1. Field
Some embodiments of the disclosure relate generally to methods and devices for determining a concentration of an analyte in a sample, such as an analyte in a sample of bodily fluid, as well as methods and devices which can be used to support the making of such determinations. Some embodiments of the disclosure relate to a method for determining a concentration of an analyte in a sample.
2. Description of Related Art
It is advantageous to measure the levels of certain analytes, such as glucose, in a bodily fluid, such as blood). This can be done, for example, in a hospital or clinical setting when there is a risk that the levels of certain analytes may move outside a desired range, which in turn can jeopardize the health of a patient. Currently known systems for analyte monitoring in a hospital or clinical setting may suffer from various drawbacks.
Example embodiments described herein have several features, no single one of which is indispensible or solely responsible for their desirable attributes. Without limiting the scope of the claims, some of the advantageous features will now be summarized.
Various embodiments of a method are disclosed that includes accessing one or more of calibration sets, each calibration set usable to estimate an analyte concentration for a sample. A measurement of the sample is then accessed, and the method determines, for each of the one or more calibration sets, whether the calibration set is eligible to estimate analyte concentration for the sample. An estimated analyte concentration is then determined based at least partly on the measurement of the sample and the calibration sets determined to be eligible. In some embodiments, a plurality of calibration sets are used. In some embodiments, if none of the one or more calibration sets is eligible, the method does not determine the estimated analyte concentration, and the method may return a no read indication to a display system. In some embodiments, the method further includes processing the measurement of the sample to reduce or minimize effects on the estimate of concentration of the analyte caused be one or more interferents.
In one embodiment, a method for estimating a concentration of an analyte in a sample is disclosed. The method comprises receiving one or more calibration sets, each usable to estimate an analyte concentration for a sample, accessing a measurement of the sample, and determining, for each of the one or more calibration sets, whether the calibration set is eligible to estimate analyte concentration from the measurement of the sample. The method then determines an estimated analyte concentration for the sample based at least in part on the measurement of the sample and the calibration set(s) determined to be eligible, via execution of instructions by a processor. The sample may include at least one component of blood, and the analyte may comprise glucose. In some examples of the method, accessing a measurement of the sample may comprise accessing a raw measurement of the sample and calculating a measurement of the sample based at least in part on the raw measurement of the sample. The method may use a measurement of a sample that is a measured spectrum (e.g., a mid-infrared spectrum). The one or more calibration sets may comprise data usable to estimate an analyte concentration and a reconstructed spectrum from at least the measured spectrum. Determining if a calibration set is eligible to estimate an analyte concentration may comprise calculating, for each of the one or more calibration sets, a reconstructed spectrum based at least in part on the measured spectrum and the calibration set and comparing the reconstructed spectrum to the measured spectrum. Each of the one or more calibration sets may comprise a prediction eligibility threshold, and comparing the reconstructed spectrum to the measured spectrum may comprise calculating a distance metric based at least on the reconstructed spectrum and the measured spectrum and comparing the distance metric to the prediction eligibility threshold to determine whether the calibration set is eligible to estimate analyte concentration. In some embodiments, a plurality of calibration sets is used. In some embodiments, if none of the one or more calibration sets is eligible, the method does not determine the estimated analyte concentration, and the method may return a no read indication to a display system. In some embodiments, the method further includes processing the measurement of the sample to reduce or minimize effects on the estimate of concentration of the analyte caused be one or more interferents.
In one embodiment, a method for estimating a concentration of an analyte in a sample is disclosed. The method comprises accessing a measured spectrum of the sample, the measured spectrum comprising measurements at a plurality of wavelengths, accessing one or more calibration sets, each calibration set usable to estimate an analyte concentration and a reconstructed spectrum from at least the measured spectrum and further comprising a prediction eligibility threshold, calculating, for each of the one or more calibration sets, a reconstructed spectrum, calculating, for each of the one or more calibration sets, a distance metric by comparing the reconstructed spectrum to the measured spectrum, comparing, for each of the one or more calibration sets, the calculated distance metric to the prediction eligibility threshold to determine whether the calibration set is eligible to estimate analyte concentration, calculating, for each of the eligible calibration sets, an estimated analyte concentration based at least in part on the measured spectrum and the calibration set, calculating, for each of the eligible calibration sets, a weighting coefficient based at least in part on the calculated distance metric and the prediction eligibility threshold of the calibration set, and determining an analyte concentration for the sample by using the weighting coefficients to combine the estimated analyte concentrations for each of the eligible calibration sets. In some embodiments, a plurality of calibration sets is used. In some embodiments, if none of the one or more calibration sets is eligible, the method does not determine the estimated analyte concentration, and the method may return a no read indication to a display system. In some embodiments, the method further includes processing the measurement of the sample to reduce or minimize effects on the estimate of concentration of the analyte caused be one or more interferents.
In one embodiment, an analyte detection system is disclosed. The system comprises a sensor system configured to provide information relating to a measurement of an analyte in a sample. The system further comprises a processor system configured to execute stored program instructions such that the analyte detection system accesses one or more calibration sets, each usable to estimate an analyte concentration for a sample, accesses a measurement of the sample, determines, for each of the one or more calibration sets, whether the calibration set is eligible to estimate analyte concentration from the measurement of the sample, and determines an estimated analyte concentration for the sample based at least in part on the measurement of the sample and the calibration set(s) determined to be eligible. In some embodiments, the analyte detection system accesses a plurality of calibration sets. In some embodiments, if none of the one or more calibration sets is eligible, the system does not determine the estimated analyte concentration, and the system may return a no read indication to a display system. In some embodiments, the analyte detection system is further configured to process the measurement of the sample to reduce or minimize effects on the estimate of concentration of the analyte caused be one or more interferents.
The following drawings and the associated descriptions are provided to illustrate embodiments of the present disclosure and do not limit the scope of the claims.
These and other features will now be described with reference to the drawings summarized above. The drawings and the associated descriptions are provided to illustrate embodiments and not to limit the scope of any claim. Throughout the drawings, reference numbers may be reused to indicate correspondence between referenced elements. In addition, where applicable, the first one or two digits of a reference numeral for an element can frequently indicate the figure number in which the element first appears.
Although certain preferred embodiments and examples are disclosed below, inventive subject matter extends beyond the specifically disclosed embodiments to other alternative embodiments and/or uses and to modifications and equivalents thereof. Thus, the scope of the claims appended hereto is not limited by any of the particular embodiments described below. For example, in any method or process disclosed herein, the acts or operations of the method or process may be performed in any suitable sequence and are not necessarily limited to any particular disclosed sequence. Various operations may be described as multiple discrete operations in turn, in a manner that may be helpful in understanding certain embodiments; however, the order of description should not be construed to imply that these operations are order dependent. Additionally, the structures, systems, and/or devices described herein may be embodied as integrated components or as separate components. For purposes of comparing various embodiments, certain aspects and advantages of these embodiments are described. Not necessarily all such aspects or advantages are achieved by any particular embodiment. Thus, for example, various embodiments may be carried out in a manner that achieves or optimizes one advantage or group of advantages as taught herein without necessarily achieving other aspects or advantages as may also be taught or suggested herein.
The systems and methods discussed herein can be used anywhere, including, for example, in laboratories, hospitals, healthcare facilities, intensive care units (ICUs), or residences. Moreover, the systems and methods discussed herein can be used for invasive techniques, as well as non-invasive techniques or techniques that do not involve a body or a patient such as, for example, in vitro techniques.
In the illustrated embodiment, the monitoring device 102 includes a display 104 such as, for example, a touch-sensitive liquid crystal display. The display 104 can provide an interface that includes alerts, indicators, charts, and/or soft buttons. The device 102 also can include one or more inputs and/or outputs 106 that provide connectivity and/or permit user interactivity.
In the embodiment shown in
In some embodiments, the apparatus 100 includes the cart 130, which has an upper shelf 131 on which the monitoring device 102 may be placed (or attached) and a bottom shelf 132 on which a battery 134 may be placed (or attached). The battery 134 may be used as a main or backup power supply for the monitoring device 102 (which may additionally or alternatively accept electrical power from a wall socket). Two or more batteries are used in certain embodiments. The apparatus 100 may be configured so that the upper and lower shelves 131, 132 are close to ground level, and the battery provides counterweight. Other types of counterweights may be used. For example, in some embodiments, portions of the cart 130 near the floor (e.g., a lower shelf) are weighted, formed from a substantial quantity of material (e.g., thick sheets of metal), and/or formed from a relatively high-density metal (e.g., lead). In some embodiments the bottom shelf 132 is approximately 6 inches to 1 foot above ground level, and the upper shelf 131 is approximately 2 feet to 4 feet above ground level. In some embodiments the upper shelf 131 may be configured to support approximately 40 pounds (lbs), and the bottom shelf 132 may be configured to support approximately 20 lbs. One possible advantage of embodiments having such a configuration is that IV pumps, bags containing saline, blood and/or drugs, and other medical equipment weighing approximately 60 lbs, collectively, can be hung on the hooks 112 above the shelves without making the apparatus 100 unstable. The apparatus 100 may be moved by applying a horizontal force on the apparatus 100, for example, by pushing and/or pulling the poles 110. In many cases, a user may exert force on an upper portion of the apparatus 100, for example, close to shoulder-height. By counterbalancing the weight as described above, the apparatus 100 may be moved in a reasonably stable manner.
In the illustrated embodiment, the cart 130 includes the bottom shelf 132 and an intermediate shelf 133, which are enclosed on three sides by walls and on a fourth side by a door 135. The door 135 can be opened (as shown in
In some embodiments (see, e.g.,
The example monitoring device 102 schematically illustrated in
In some embodiments, the apparatus 100 is not connected to a patient and may receive fluid samples from a container such as a decanter, flask, beaker, tube, cartridge, test strip, etc., or any other extracorporeal fluid source. The container may include a biological fluid sample such as, e.g., a body fluid sample. For example,
In the embodiment shown in
In the embodiment illustrated in
In other embodiments of the apparatus 100 shown in
In some embodiments, one or more components of the apparatus 100 can be located at another facility, room, or other suitable remote location. One or more components of the monitoring device 102 can communicate with one or more other components of the monitoring device 102 (or with other devices) by communication interface(s) such as, but not limited to, optical interfaces, electrical interfaces, and/or wireless interfaces. These interfaces can be part of a local network, internet, wireless network, or other suitable networks.
In some embodiments, the sampling and analysis system 400 is a mobile point-of-care apparatus that monitors physiological parameters such as, for example, blood glucose concentration. Components within the system 400 that may contact fluid and/or a patient, such as tubes and connectors, can be coated with an antibacterial coating to reduce the risk of infection. Connectors between at least some components of the system 400 can include a self-sealing valve, such as a spring valve, in order to reduce the risk of contact between port openings and fluids, and to guard against fluid escaping from the system. Other components can also be included in a system for sampling and analyzing fluid in accordance with the described embodiments.
The sampling and analysis system 400 can include a fluid source 402 (or more than one fluid source) that contain(s) fluid to be sampled. The fluid-handling system 404 of the sampling and analysis system 400 is connected to, and can draw fluid from, the fluid source 402. The fluid source 402 can be, for example, a blood vessel such as a vein or an artery, a container such as a decanter, flask, beaker, tube, cartridge, test strip, etc., or any other corporeal or extracorporeal fluid source. For example, in some embodiments, the fluid source 402 may be a vein or artery in the patient 302 (see, e.g.,
In some embodiments, the fluid-handling system 404 withdraws a sample of fluid from the fluid source 402 for analysis, centrifuges at least a portion of the sample, and prepares at least a portion of the sample for analysis by an optical sensor such as a spectrophotometer (which can be part of an optical system 412, for example). These functions can be controlled by a fluid system controller 405, which can also be integrated into the fluid-handling system 404. The fluid system controller 405 can also control the additional functions described below. In some embodiments, the sample can be withdrawn continuously or substantially continuously at certain time intervals with a given period. The time intervals at which the sample is withdrawn can be periodic or aperiodic and range from approximately 1 minute to approximately 15 minutes (e.g., the sample can be withdrawn at time intervals of 1 minute, 5 minutes, 10 minutes or 15 minutes). In some embodiments, the sample can be withdrawn at discrete time intervals (e.g., once every 30 minutes, once every 45 minutes or once every hour).
The duration of time over which the sample of fluid is withdrawn, referred to as “draw period”, may be set to avoid clinical drawbacks, and/or it can be varied according to a health-care provider's wishes. For example, in some embodiments, fluid may be continuously withdrawn into the sampling and analysis system 400 over a draw period lasting approximately 10 seconds to approximately 5 minutes.
In some embodiments, the amount of sample withdrawn from the fluid source 402 can be small. For example, in some embodiments, the volume of sample withdrawn from the fluid source can be between approximately 1.0 ml and approximately 10.0 ml in a draw period (e.g. 2.0 ml-6.0 ml of sample can be withdrawn in a draw period of approximately 1 minute). In some embodiments, the amount of sample withdrawn can be in the range of approximately 20 ml/day to approximately 500 ml/day. In some embodiments, the amount of sample withdrawn can be outside this range.
In some embodiments, at least a portion of the sample is returned to the fluid source 402. At least some of the sample, such as portions of the sample that are mixed with other materials or portions that are otherwise altered during the sampling and analysis process, or portions that, for any reason, are not to be returned to the fluid source 402, can also be placed in a waste bladder (not shown in
Components of the fluid-handling system 404 can be modularized into one or more non-disposable, disposable, and/or replaceable subsystems. In the embodiment shown in
The non-disposable subsystem 406 can include components that, while they may be replaceable or adjustable, do not generally require regular replacement during the useful lifetime of the system 400. In some embodiments, the non-disposable subsystem 406 of the fluid-handling system 404 includes one or more reusable valves and sensors. For example, the non-disposable subsystem 406 can include one or more valves (or non-disposable portions thereof), (e.g., pinch-valves, rotary valves, etc.), sensors (e.g., ultrasonic bubble sensors, non-contact pressure sensors, optical blood dilution sensors, etc). The non-disposable subsystem 406 can also include one or more pumps (or non-disposable portions thereof). For example, some embodiments can include pumps available from Hospira. In some embodiments, the components of the non-disposable subsystem 406 are not directly exposed to fluids and/or are not readily susceptible to contamination.
The first and second disposable subsystems 408, 410 can include components that are regularly replaced under certain circumstances in order to facilitate the operation of the system 400. For example, the first disposable subsystem 408 can be replaced after a certain period of use, such as a few days, has elapsed. Replacement may be necessary, for example, when a bladder within the first disposable subsystem 408 is filled to capacity. Such replacement may mitigate fluid system performance degradation associated with and/or contamination wear on system components.
In some embodiments, the first disposable subsystem 408 includes components that may contact fluids such as patient blood, saline, flushing solutions, anticoagulants, and/or detergent solutions. For example, the first disposable subsystem 408 can include one or more tubes, fittings, cleaner pouches and/or waste bladders. The components of the first disposable subsystem 408 can be sterilized in order to decrease the risk of infection and can be configured to be easily replaceable.
In some embodiments, the second disposable subsystem 410 can be designed to be replaced under certain circumstances. For example, the second disposable subsystem 410 can be replaced when the patient being monitored by the system 400 is changed. The components of the second disposable subsystem 410 may not need replacement at the same intervals as the components of the first disposable subsystem 408. For example, the second disposable subsystem 410 can include a sample holder and/or at least some components of a centrifuge, components that may not become filled or quickly worn during operation of the system 400. Replacement of the second disposable subsystem 410 can decrease or eliminate the risk of transferring fluids from one patient to another during operation of the system 400, enhance the measurement performance of system 400, and/or reduce the risk of contamination or infection.
In some embodiments, the sample holder of the second disposable subsystem 410 receives the sample obtained from the fluid source 402 via fluid passageways of the first disposable subsystem 408. The sample holder is a container that can hold fluid for the centrifuge and can include a window to the sample for analysis by a spectrometer. In some embodiments, the sample holder includes windows that are made of a material that is substantially transparent to electromagnetic radiation in the mid-infrared range of the spectrum. For example, the sample holder windows can be made of calcium fluoride.
An injector can provide a fluid connection between the first disposable subsystem 408 and the sample holder of the second disposable subsystem 410. In some embodiments, the injector can be removed from the sample holder to allow for free spinning of the sample holder during centrifugation.
In some embodiments, the components of the sample are separated by centrifuging for a period of time before measurements are performed by the optical system 412. For example, a fluid sample (e.g., a blood sample) can be centrifuged at a relatively high speed. The sample can be spun at a certain number of revolutions per minute (RPM) for a given length of time to separate blood plasma for spectral analysis. In some embodiments, the fluid sample is spun at about 7200 RPM. In some embodiments, the sample is spun at about 5000 RPM. In some embodiments, the fluid sample is spun at about 4500 RPM. In some embodiments, the fluid sample is spun at more than one rate for successive time periods. The length of time can be approximately 5 minutes. In some embodiments, the length of time is approximately 2 minutes. Separation of a sample into the components can permit measurement of solute (e.g., glucose) concentration in plasma, for example, without interference from other blood components. This kind of post-separation measurement, (sometimes referred to as a “direct measurement”) has advantages over a solute measurement taken from whole blood because the proportions of plasma to other components need not be known or estimated in order to infer plasma glucose concentration. In some embodiments, the separated plasma can be analyzed electrically using one or more electrodes instead of, or in addition to, being analyzed optically. This analysis may occur within the same device, or within a different device. For example, in certain embodiments, an optical analysis device can separate blood into components, analyze the components, and then allow the components to be transported to another analysis device that can further analyze the components (e.g., using electrical and/or electrochemical measurements).
An anticoagulant, such as, for example, heparin can be added to the sample before centrifugation to prevent clotting. The fluid-handling system 404 can be used with a variety of anticoagulants, including anticoagulants supplied by a hospital or other user of the monitoring system 400. A detergent solution formed by mixing detergent powder from a pouch connected to the fluid-handling system 404 with saline can be used to periodically clean residual protein and other sample remnants from one or more components of the fluid-handling system 404, such as the sample holder. Sample fluid to which anticoagulant has been added and used detergent solution can be transferred into the waste bladder.
The system 400 shown in
In some embodiments, the optical system 412 includes a filter wheel that contains one or more filters. In some embodiments, more than ten filters can be included, for example twelve or fifteen filters. In some embodiments, more than 20 filters (e.g., twenty-five filters) are mounted on the filter wheel. The optical system 412 includes a light source that passes light through a filter and the sample holder to a detector. In some embodiments, a stepper motor moves the filter wheel in order to position a selected filter in the path of the light. An optical encoder can also be used to finely position one or more filters. In some embodiments, one or more tunable filters may be used to filter light into multiple wavelengths. The one or more tunable filters may provide the multiple wavelengths of light at the same time or at different times (e.g., sequentially). The light source included in the optical system 412 may emit radiation in the ultraviolet, visible, near-infrared, mid-infrared, and/or far-infrared regions of the electromagnetic spectrum. In some embodiments, the light source can be a broadband source that emits radiation in a broad spectral region (e.g., from about 1500 nm to about 6000 nm). In other embodiments, the light source may emit radiation at certain specific wavelengths. The light source may comprise one or more light emitting diodes (LEDs) emitting radiation at one or more wavelengths in the radiation regions described herein. In other embodiments, the light source may comprise one or more laser modules emitting radiation at one or more wavelengths. The laser modules may comprise a solid state laser (e.g., a Nd:YAG laser), a semiconductor based laser (e.g., a GaAs and/or InGaAsP laser), and/or a gas laser (e.g., an Ar-ion laser). In some embodiments, the laser modules may comprise a fiber laser. The laser modules may emit radiation at certain fixed wavelengths. In some embodiments, the emission wavelength of the laser module(s) may be tunable over a wide spectral range (e.g., about 30 nm to about 100 nm). In some embodiments, the light source included in the optical system 412 may be a thermal infrared emitter. The light source can comprise a resistive heating element, which, in some embodiments, may be integrated on a thin dielectric membrane on a micromachined silicon structure. In one embodiment the light source is generally similar to the electrical modulated thermal infrared radiation source, IRSource™, available from the Axetris Microsystems division of Leister Technologies, LLC (Itasca, Ill.).
The optical system 412 can be controlled by an optical system controller 413. The optical system controller can, in some embodiments, be integrated into the optical system 412. In some embodiments, the fluid system controller 405 and the optical system controller 413 can communicate with each other as indicated by the line 411. In some embodiments, the function of these two controllers can be integrated and a single controller can control both the fluid-handling system 404 and the optical system 412. Such an integrated control can be advantageous because the two systems are preferably integrated, and the optical system 412 is preferably configured to analyze the very same fluid handled by the fluid-handling system 404. Indeed, portions of the fluid-handling system 404 (e.g., the sample holder described above with respect to the second disposable subsystem 410 and/or at least some components of a centrifuge) can also be components of the optical system 412. Accordingly, the fluid-handling system 404 can be controlled to obtain a fluid sample for analysis by optical system 412, when the fluid sample arrives, the optical system 412 can be controlled to analyze the sample, and when the analysis is complete (or before), the fluid-handling system 404 can be controlled to return some of the sample to the fluid source 402 and/or discard some of the sample, as appropriate.
The system 400 shown in
The system 400 shown in
Some embodiments of the systems described herein (e.g., the system 400), as well as some embodiments of each method described herein, can include a physical computer program accessible to and/or executable by a processing system, e.g., a one or more processors, controllers, and memories that are part of an embedded system. Indeed, the controllers may comprise one or more physical computing devices and/or may use software. Thus, as will be appreciated by those skilled in the art, various embodiments may be embodied as a method, an apparatus such as a special purpose apparatus, an apparatus such as a data processing system, or a carrier medium, e.g., a non-transitory computer program product. The carrier medium carries one or more computer readable code segments for controlling a processing system to implement a method. The carrier medium can include non-transitory storage media or carrier signals. Accordingly, various embodiments may take the form of a method, an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, any one or more of the disclosed methods (including but not limited to the disclosed methods of measurement analysis, interferent determination, and/or calibration constant generation) may be stored as one or more computer readable code segments or data compilations on a carrier medium. Any suitable computer readable carrier medium may be used including a magnetic storage device such as a diskette or a hard disk; a memory cartridge, module, card or chip (either alone or installed within a larger device); or an optical storage device such as a CD or DVD.
The generalized fluid-handling system 404 can have various configurations. In this context,
In addition to the reference numerals used below, the various portions of the illustrated fluid system 510 are labeled for convenience with letters to suggest their roles as follows: T# indicates a section of tubing. C# indicates a connector that joins multiple tubing sections. V# indicates a valve. BS# indicates a bubble sensor or ultrasonic air detector. N# indicates a needle (e.g., a needle that injects sample into a sample holder). PS# indicates a pressure sensor (e.g., a reusable pressure sensor). Pump# indicates a fluid pump (e.g., a syringe pump with a disposable body and reusable drive). “Hb 12” indicates a sensor for hemoglobin (e.g., a dilution sensor that can detect hemoglobin optically).
The term “valve” as used herein is a broad term and is used, in accordance with its ordinary meaning, to refer to any flow regulating device. For example, the term “valve” can include, without limitation, any device or system that can controllably allow, prevent, or inhibit the flow of fluid through a fluid passageway. The term “valve” can include some or all of the following, alone or in combination: pinch valves, rotary valves, stop cocks, pressure valves, shuttle valves, mechanical valves, electrical valves, electro-mechanical flow regulators, etc. In some embodiments, a valve can regulate flow using gravitational methods or by applying electrical voltages or by both.
The term “pump” as used herein is a broad term and is used, in accordance with its ordinary meaning, to refer to any device that can urge fluid flow. For example, the term “pump” can include any combination of the following: syringe pumps, peristaltic pumps, vacuum pumps, electrical pumps, mechanical pumps, hydraulic pumps, etc. Pumps and/or pump components that are suitable for use with some embodiments can be obtained, for example, from or through Hospira.
The function of the valves, pumps, actuators, drivers, motors (e.g., the centrifuge motor), etc. described below is controlled by one or more controllers (e.g., the fluid system controller 405, the optical system controller 413, etc.) The controllers can include software, computer memory, electrical and mechanical connections to the controlled components, etc.
At the start of a measurement cycle, most lines, including a patient tube 512 (T1), an Arrival sensor tube 528 (T4), an anticoagulant valve tube 534 (T3), and a sample cell 548 can be filled with saline that can be introduced into the system through the infusion tube 514 and the saline tube 516, and which can come from an infusion pump 518 and/or a saline bag 520. The infusion pump 518 and the saline bag 520 can be provided separately from the system 510. For example, a hospital can use existing saline bags and infusion pumps to interface with the described system. The infusion valve 521 can be open to allow saline to flow into the tube 512 (T1).
Before drawing a sample, the saline in part of the system 510 can be replaced with air. Thus, for example, the following valves can be closed: air valve 503 (PV0), the detergent tank valve 559 (V7b), 566 (V3b), 523 (V0), 529 (V7a), and 563 (V2b). At the same time, the following valves can be open: valves 531 (Via), 533 (V3a) and 577 (V4a). Simultaneously, a second pump 532 (pump #0) pumps air through the system 510 (including tube 534 (T3), sample cell 548, and tube 556 (T6)), pushing saline through tube 534 (T3) and sample cell 548 into a waste bladder 554.
Next, a sample can be drawn. With the valves 542 (PV1), 559 (V7b), and 561 (V4b) closed, a first pump 522 (pump #1) is actuated to draw sample fluid to be analyzed (e.g. blood) from a fluid source (e.g., a laboratory sample container, a living patient, etc.) up into the patient tube 512 (T1), through the tube past the two flanking portions of the open pinch-valve 523 (V0), through the first connector 524 (C1), into the looped tube 530, past the arrival sensor 526 (Hb12), and into the arrival sensor tube 528 (T4). The arrival sensor 526 may be used to detect the presence of blood in the tube 528 (T4). For example in some embodiments, the arrival sensor 526 may comprise a hemoglobin sensor. In some other embodiments, the arrival sensor 526 may comprise a color sensor that detects the color of fluid flowing through the tube 528 (T4). During this process, the valve 529 (V7a) and 523 (V0) are open to fluid flow, and the valves 531 (Via), 533 (V3a), 542 (PV1), 559 (V7b), and 561 (V4b) can be closed and therefore block (or substantially block) fluid flow by pinching the tube.
Before drawing the sample, the tubes 512 (T1) and 528 (T4) are filled with saline and the hemoglobin (Hb) level is zero. The tubes that are filled with saline are in fluid communication with the sample source (e.g., the fluid source 402). The sample source can be the vessels of a living human or a pool of liquid in a laboratory sample container, for example. When the saline is drawn toward the first pump 522, fluid to be analyzed is also drawn into the system because of the suction forces in the closed fluid system. Thus, the first pump 522 draws a relatively continuous column of fluid that first comprises generally nondiluted saline, then a mixture of saline and sample fluid (e.g., blood), and then eventually nondiluted sample fluid. In the example illustrated here, the sample fluid is blood.
The arrival sensor 526 (Hb12) can detect and/or verify the presence of blood in the tubes. For example, in some embodiments, the arrival sensor 526 can determine the color of the fluid in the tubes. In some embodiments, the arrival sensor 526 (Hb12) can detect the level of Hemoglobin in the sample fluid. As blood starts to arrive at the arrival sensor 526 (Hb12), the sensed hemoglobin level rises. A hemoglobin level can be selected, and the system can be pre-set to determine when that level is reached. A controller such as the fluid system controller 405 of
In some embodiments, when nondiluted blood is present at the first connector 524 (C1), a sample is mixed with an anticoagulant and is directed toward the sample cell 548. An amount of anticoagulant (e.g., heparin) can be introduced into the tube 534 (T3), and then the undiluted blood is mixed with the anticoagulant. A heparin vial 538 (e.g., an insertable vial provided independently by the user of the system 510) can be connected to a tube 540. An anticoagulant valve 541 (which can be a shuttle valve, for example) can be configured to connect to both the tube 540 and the anticoagulant valve tube 534 (T3). The valve can open the tube 540 to a suction force (e.g., created by the pump 532), allowing heparin to be drawn from the vial 538 into the valve 541. Then, the anticoagulant valve 541 can slide the heparin over into fluid communication with the anticoagulant valve tube 534 (T3). The anticoagulant valve 541 can then return to its previous position. Thus, heparin can be shuttled from the tube 540 into the anticoagulant valve tube 534 (T3) to provide a controlled amount of heparin into the tube 534 (T3).
With the valves 542 (PV1), 559 (V7b), 561 (V4b), 523 (V0), 531 (Via), 566 (V3b), and 563 (V2b) closed, and the valves 529 (V7a) and 533 (V3a) open, first pump 522 (pump #1) pushes the sample from tube 528 (T4) into tube 534 (T3), where the sample mixes with the heparin injected by the anticoagulant valve 541 as it flows through the system 510. As the sample proceeds through the tube 534 (T3), the air that was previously introduced into the tube 534 (T3) is displaced. The sample continues to flow until a bubble sensor 535 (BS9) indicates a change from air to a liquid, and thus the arrival of a sample at the bubble sensor. In some embodiments, the volume of tube 534 (T3) from connector 524 (C1) to bubble sensor 535 (BS9) is a known and/or engineered amount, and may be approximately 500 μL, 200 μL or 100 μL, for example. In some embodiments, the volume of tube 534 (T3) from connector 524 (C1) to bubble sensor 535 (BS9) may be approximately less than 10 ml.
When bubble sensor 535 (BS9) indicates the presence of a sample, the remainder of the sampled blood can be returned to its source (e.g., the patient veins or arteries). The first pump 522 (pump #1) pushes the blood out of the Arrival sensor tube 528 (T4) and back to the patient by opening the valve 523 (V0), closing the valves 531 (Via) and 533 (V3a), and keeping the valve 529 (V7a) open. The Arrival sensor tube 528 (T4) is preferably flushed with approximately 2 mL of saline. This can be accomplished by closing the valve 529 (V7a), opening the valve 542 (PV1), drawing saline from the saline source 520 into the tube 544, closing the valve 542 (PV1), opening the valve 529 (V7a), and forcing the saline down the Arrival sensor tube 528 (T4) with the pump 522. In some embodiments, less than two minutes elapse between the time that blood is drawn from the patient and the time that the blood is returned to the patient.
Following return of the unused patient blood sample, the sample is pushed up the anticoagulant valve tube 534 (T3), through the second connector 546 (C2), and into the sample cell 548, which can be located on the centrifuge rotor 550. This fluid movement is facilitated by the coordinated action (either pushing or drawing fluid) of the pump 522 (pump #1), the pump 532 (pump #0), and the various illustrated valves. In particular, valve 531 (V1a) can be opened, and valves 503 (PV0) and 559 (V7b) can be closed. Pump movement and valve position corresponding to each stage of fluid movement can be coordinated by one ore multiple controllers, such as the fluid system controller 405 of
After the unused sample is returned to the patient, the sample can be divided into separate slugs before being delivered into the sample cell 548. Thus, for example, valve 533 (V3a) is opened, valves 566 (V3b), 523 (V0) and 529 (V7a) are closed, and the pump 532 (pump #0) uses air to push the sample toward sample cell 548. In some embodiments, the sample (for example, 200 μL or 100 μL) is divided into multiple (e.g., more than two, five, or four) “slugs” of sample, each separated by a small amount of air. As used herein, the term “slug” refers to a continuous column of fluid that can be relatively short. Slugs can be separated from one another by small amounts of air (or bubbles) that can be present at intervals in the tube. In some embodiments, the slugs are formed by injecting or drawing air into fluid in the first connector 546 (C2).
In some embodiments, when the leading edge of the sample reaches blood sensor 552 (BS14), a small amount of air (the first “bubble”) is injected at a connector C6. This bubble helps define the first “slug” of liquid, which extends from the bubble sensor to the first bubble. In some embodiments, the valves 533 (V3a) and 566 (V3b) are alternately opened and closed to form a bubble at connector C6, and the sample is pushed toward the sample cell 548. Thus, for example, with pump 532 actuated, valve 566 V (3b) is briefly opened and valve 533 (V3a) is briefly closed to inject a first air bubble into the sample.
In some embodiments, the volume of the tube 534 (T3) from the connector 546 (C2) to the bubble sensor 552 (BS14) is less than the volume of tube 534 (T3) from the connector 524 (C1) to the bubble sensor 535 (BS9). Thus, for example and without limitation, the volume of the tube 534 (T3) from the connector 524 (C1) to the bubble sensor 535 (BS9) can be in the range of approximately 80 μL to approximately 120 μL, (e.g., 100 μL) and the volume of the tube 534 (T3) from the connector 546 (C2) to the bubble sensor 552 (BS14) can be in the range of approximately 5 μL to approximately 25 μL (e.g., 15 μL). In some embodiments, multiple blood slugs are created. For example, more than two blood slugs can be created, each having a different volume. In some embodiments, five blood slugs are created, each having approximately the same volume of approximately 20 μL each. In some embodiments, three blood slugs are created, the first two having a volume of 10 μL and the last having a volume of 20 μL. In some embodiments, four blood slugs are created; the first three blood slugs can have a volume of approximately 15 μL and the fourth can have a volume of approximately 35 μL.
A second slug can be prepared by opening the valve 533 (V3a), closing the valve 566 (V3b), with pump 532 (pump #0) operating to push the first slug through a first sample cell holder interface tube 582 (N1), through the sample cell 548, through a second sample cell holder interface tube 584 (N2), and toward the waste bladder 554. When the first bubble reaches the bubble sensor 552 (BS 14), the open/closed configurations of valves 533 (V3a) and 566 (V3b) are reversed, and a second bubble is injected into the sample, as before. A third slug can be prepared in the same manner as the second (pushing the second bubble to bubble sensor 552 (BS 14) and injecting a third bubble). After the injection of the third air bubble, the sample can be pushed through system 510 until the end of the sample is detected by bubble sensor 552 (BS 14). The system can be designed such that when the end of the sample reaches this point, the last portion of the sample (a fourth slug) is within the sample cell 548, and the pump 532 can stop forcing the fluid column through the anticoagulant valve tube 534 (T3) so that the fourth slug remains within the sample cell 548. Thus, the first three blood slugs can serve to flush any residual saline out the sample cell 548. The three leading slugs can be deposited in the waste bladder 554 by passing through the tube 556 (T6) and past the tube-flanking portions of the open pinch valve 557 (V4a).
In some embodiments, the fourth blood slug is centrifuged for a given length of time (e.g., more than 1 minute, five minutes, or 2 minutes, to take three advantageous examples) at a relatively fast speed (e.g., 7200 RPM, 5000 RPM, or 4500 RPM, to take three examples). Thus, for example, the sample cell holder interface tubes 582 (N1) and 584 (N2) disconnect the sample cell 548 from the tubes 534 (T3) and 562 (T7), permitting the centrifuge rotor 550 and the sample cell 548 to spin together. Spinning separates a sample (e.g., blood) into its components, isolates the plasma, and positions the plasma in the sample cell 548 for measurement. The centrifuge 550 can be stopped with the sample cell 548 in a beam of radiation (not shown) for analysis. The radiation, a detector, and logic can be used to analyze a portion of the sample (e.g., the plasma) spectroscopically (e.g., for glucose, lactate, or other analyte concentration). In some embodiments, some or all of the separated components (e.g., the isolated plasma) may be transported to a different analysis chamber. For example, another analysis chamber can have one or more electrodes in electrical communication with the chamber's contents, and the separated components may be analyzed electrically. At any suitable point, one or more of the separated components can be transported to the waste bladder 554 when no longer needed. In some chemical analysis systems and apparatus, the separated components are analyzed electrically. Analysis devices may be connected serially, for example, so that the analyzed substance from an optical analysis system (e.g., an “OptiScanner®” fluid analyzer) can be transferred to an independent analysis device (e.g., a chemical analysis device) for subsequent analysis. In certain embodiments, the analysis devices are integrated into a single system. Many variations are possible.
In some embodiments, portions of the system 510 that contain blood after the sample cell 548 has been provided with a sample are cleaned to prevent blood from clotting. Accordingly, the centrifuge rotor 550 can include two passageways for fluid that may be connected to the sample cell holder interface tubes 582 (N1) and 584 (N2). One passageway is sample cell 548, and a second passageway is a shunt 586. An embodiment of the shunt 586 is illustrated in more detail in
The shunt 586 can allow cleaner (e.g., a detergent such as tergazyme A) to flow through and clean the sample cell holder interface tubes without flowing through the sample cell 548. After the sample cell 548 is provided with a sample, the interface tubes 582 (N1) and 584 (N2) are disconnected from the sample cell 548, the centrifuge rotor 550 is rotated to align the shunt 586 with the interface tubes 582 (N1) and 584 (N2), and the interface tubes are connected with the shunt. With the shunt in place, the detergent tank 559 is pressurized by the second pump 532 (pump #0) with valves 561 (V4b) and 563 (V2b) open and valves 557 (V4a) and 533 (V3a) closed to flush the cleaning solution back through the interface tubes 582 (N1) and 584 (N2) and into the waste bladder 554. Subsequently, saline can be drawn from the saline bag 520 for a saline flush. This flush pushes saline through the Arrival sensor tube 528 (T4), the anticoagulant valve tube 534 (T3), the sample cell 548, and the waste tube 556 (T6). Thus, in some embodiments, the following valves are open for this flush: 529 (V7a), 533 (V3a), 557 (V4a), and the following valves are closed: 542 (PV1), 523 (V0), 531 (V1a), 566 (V3b), 563 (V2b), and 561 (V4b).
Following analysis, the second pump 532 (pump #0) flushes the sample cell 548 and sends the flushed contents to the waste bladder 554. This flush can be done with a cleaning solution from the detergent tank 558. In some embodiments, the detergent tank valve 559 (V7b) is open, providing fluid communication between the second pump 532 and the detergent tank 558. The second pump 532 forces cleaning solution from the detergent tank 558 between the tube-flanking portions of the open pinch valve 561 and through the tube 562 (T7). The cleaning flush can pass through the sample cell 548, through the second connector 546, through the tube 564 (T5) and the open valve 563 (V2b), and into the waste bladder 554.
Subsequently, the first pump 522 (pump #1) can flush the cleaning solution out of the sample cell 548 using saline in drawn from the saline bag 520. This flush pushes saline through the Arrival sensor tube 528 (T4), the anticoagulant valve tube 534 (T3), the sample cell 548, and the waste tube 556 (T6). Thus, in some embodiments, the following valves are open for this flush: 529 (V7a), 533 (V3a), 557 (V4a), and the following valves are closed: 542 (PV1), 523 (V0), 531 (V1a), 566 (V3b), 563 (V2b), and 561 (V4b).
When the fluid source is a living entity such as a patient, a low flow of saline (e.g., 1-5 mL/hr) is preferably moved through the patient tube 512 (T1) and into the patient to keep the patient's vessel open (e.g., to establish a keep vessel open, or “KVO” flow). This KVO flow can be temporarily interrupted when fluid is drawn into the fluid system 510. The source of this KVO flow can be the infusion pump 518, the third pump 568 (pump #3), or the first pump 522 (pump #1). In some embodiments, the infusion pump 518 can run continuously throughout the measurement cycle described above. This continuous flow can advantageously avoid any alarms that may be triggered if the infusion pump 518 senses that the flow has stopped or changed in some other way. In some embodiments, when the infusion valve 521 closes to allow pump 522 (pump #1) to withdraw fluid from a fluid source (e.g., a patient), the third pump 568 (pump #3) can withdraw fluid through the connector 570, thus allowing the infusion pump 518 to continue pumping normally as if the fluid path was not blocked by the infusion valve 521. If the measurement cycle is about two minutes long, this withdrawal by the third pump 568 can continue for approximately two minutes. Once the infusion valve 521 is open again, the third pump 568 (pump #3) can reverse and insert the saline back into the system at a low flow rate. Preferably, the time between measurement cycles is longer than the measurement cycle itself (for example, the time interval can be longer than ten minutes, shorter than ten minutes, shorter than five minutes, longer than two minutes, longer than one minute, etc.). Accordingly, the third pump 568 can insert fluid back into the system at a lower rate than it withdrew that fluid. This can help prevent an alarm by the infusion pump.
In some embodiments, the system 400 (see
In certain embodiments, analyte levels in a patient can be adjusted directly (e.g., by infusing or extracting glucose) or indirectly (e.g., by infusing or extracting insulin).
In some embodiments, a non-disposable fluid-handling subsystem 708 is disposed within the body portion 702 of the monitoring device 700. The first removable portion 710 can include one or more openings that allow portions of the non-disposable fluid-handling subsystem 708 to interface with the removable portion 710. For example, the non-disposable fluid-handling subsystem 708 can include one or more pinch valves that are designed to extend through such openings to engage one or more sections of tubing. When the first removable portion 710 is present in a corresponding first receptacle 704, actuation of the pinch valves can selectively close sections of tubing within the removable portion. The non-disposable fluid-handling subsystem 708 can also include one or more sensors that interface with connectors, tubing sections, or pumps located within the first removable portion 710. The non-disposable fluid-handling subsystem 708 can also include one or more actuators (e.g., motors) that can actuate moveable portions (e.g., the plunger of a syringe) that may be located in the removable portion F10. A portion of the non-disposable fluid-handling subsystem 708 can be located on or in the moveable portion F06 (which can be a door having a slide or a hinge, a detachable face portion, etc.).
In the embodiment shown in
In some embodiments, the removable portion 710 and the sample holder are adapted to be in fluid communication with each other. For example, the removable portion 710 can include a retractable injector that injects fluids into a sample holder. In some embodiments, the sample holder can comprise or be disposed in a second removable portion (not shown). In some embodiments, the injector can be retracted to allow the centrifuge to rotate the sample holder freely.
The body portion 702 of the monitoring device 700 can also include one or more connectors for an external battery (not shown). The external battery can serve as a backup emergency power source in the event that a primary emergency power source such as, for example, an internal battery (not shown) is exhausted.
Depicted in
In some embodiments, a sample cell holder 820 can engage a centrifuge motor 818 mounted within the casing 802 of the device 800. A filter wheel motor 822 disposed within the housing 802 rotates a filter wheel 824, and in some embodiments, aligns one or more filters with an optical path. An optical path can originate at a source 826 within the housing 802 that can be configured to emit a beam of radiation (e.g., infrared radiation, visible radiation, ultraviolet radiation, etc.) through the filter and the sample cell holder 820 and to a detector 828. A detector 828 can measure the optical density of the light when it reaches the detector.
The cartridge 1000 can include one or more ports for connecting to material sources or receptacles. Such ports can be provided to connect to, for example, a saline source, an infusion pump, a sample source, and/or a source of gas (e.g., air, nitrogen, etc.). The ports can be connected to sections of tubing within the cartridge 1000. In some embodiments, the sections of tubing are opaque or covered so that fluids within the tubing cannot be seen, and in some embodiments, sections of tubing are transparent to allow interior contents (e.g., fluid) to be seen from outside.
The cartridge 1000 shown in
The housing of the cartridge 1000 can include a tubing portion 1008 containing within it a card having one or more sections of tubing. In some embodiments, the body of the cartridge 1000 includes one or more apertures 1009 through which various components, such as, for example, pinch valves and sensors, can interface with the fluid-handling portion contained in the cartridge 1000. The sections of tubing found in the tubing portion 1008 can be aligned with the apertures 1009 in order to implement at least some of the functionality shown in the fluid system 510 of
The cartridge 1000 can include a pouch space (not shown) that can comprise one or more components of the fluid system 510. For example, one or more pouches and/or bladders can be disposed in the pouch space (not shown). In some embodiments, a cleaner pouch and/or a waste bladder can be housed in a pouch space. The waste bladder can be placed under the cleaner pouch such that, as detergent is removed from the cleaner pouch, the waste bladder has more room to fill. The components placed in the pouch space (not shown) can also be placed side-by-side or in any other suitable configuration.
The cartridge 1000 can include one or more pumps 1016 that facilitate movement of fluid within the fluid system 510. Each of the pump housings 1016 can contain, for example, a syringe pump having a plunger. The plunger can be configured to interface with an actuator outside the cartridge 1000. For example, a portion of the pump that interfaces with an actuator can be exposed to the exterior of the cartridge 1000 housing by one or more apertures 1018 in the housing.
The cartridge 1000 can have an optical interface portion 1030 that is configured to interface with (or comprise a portion of) an optical system. In the illustrated embodiment, the optical interface portion 1030 can pivot around a pivot structure 1032. The optical interface portion 1030 can house a sample holder (not shown) in a chamber that can allow the sample holder to rotate. The sample holder can be held by a centrifuge interface 1036 that can be configured to engage a centrifuge motor (not shown). When the cartridge 1000 is being inserted into a system, the orientation of the optical interface portion 1030 can be different than when it is functioning within the system.
In some embodiments, the cartridge 1000 is designed for single patient use. The cartridge 1000 may also be disposable and/or designed for replacement after a period of operation. For example, in some embodiments, if the cartridge 1000 is installed in a continuously operating monitoring device that performs four measurements per hour, the waste bladder may become filled or the detergent in the cleaner pouch depleted after about three days. The cartridge 1000 can be replaced before the detergent and waste bladder are exhausted. In some embodiments, a portion of the cartridge 1000 can be disposable while another portion of the cartridge 1000 is disposable, but lasts longer before being discarded. In some embodiments, a portion of the cartridge 1000 may not be disposable at all. For example, a portion thereof may be configured to be cleaned thoroughly and reused for different patients. Various combinations of disposable and less- or non-disposable portions are possible.
The cartridge 1000 can be configured for easy replacement. For example, in some embodiments, the cartridge 1000 is designed to have an installation time of only minutes. For example, the cartridge can be designed to be installed in less than about five minutes, or less than two minutes. During installation, various fluid lines contained in the cartridge 1000 can be primed by automatically filling the fluid lines with saline. The saline can be mixed with detergent powder from the cleaner pouch in order to create a cleaning solution.
The cartridge 1000 can also be designed to have a relatively brief shut down time. For example, the shut down process can be configured to take less than about fifteen minutes, or less than about ten minutes, or less than about five minutes. The shut down process can include flushing the patient line; sealing off the insulin pump connection, the saline source connection, and the sample source connection; and taking other steps to decrease the risk that fluids within the used cartridge 1000 will leak after disconnection from the monitoring device.
Some embodiments of the cartridge 1000 can comprise a flat package to facilitate packaging, shipping, sterilizing, etc. Advantageously, however, some embodiments can further comprise a hinge or other pivot structure. Thus, as illustrated, an optical interface portion 1030 can be pivoted around a pivot structure 1032 to generally align with the other portions of the cartridge 1000. The cartridge can be provided to a medical provider sealed in a removable wrapper, for example.
In some embodiments, the cartridge 1000 is designed to fit within standard waste containers found in a hospital, such as a standard biohazard container. For example, the cartridge 1000 can be less than one foot long, less than one foot wide, and less than two inches thick. In some embodiments, the cartridge 1000 is designed to withstand a substantial impact, such as that caused by hitting the ground after a four foot drop, without damage to the housing or internal components. In some embodiments, the cartridge 1000 is designed to withstand significant clamping force applied to its casing. For example, the cartridge 1000 can be built to withstand five pounds per square inch of force without damage. In some embodiments, the cartridge 1000 can be designed to be less sturdy and more biodegradable. In some embodiments, the cartridge 1000 can be formed and configured to withstand more or less than five pounds of force per square inch without damage. In some embodiments, the cartridge 1000 is non pyrogenic and/or latex free.
The depicted fluid-routing card 1038 can have additional openings that allow operative portions of actuators and/or valves to protrude through the fluid-routing card 1038 and interface with the tubes.
An opposite opening 1530 can provide an alternative optical pathway between a radiation source and a radiation detector (e.g., the radiation source 826 of
The receiving nubs 1512B communicate with a shunt passage 1586 inside the sample cell holder 820 (which can correspond, for example to the shunt 586 of
Other features of the sample cell holder 820 can provide balancing properties for even rotation of the sample cell holder 820. For example, the wide trough 1562 and the narrower trough 1564 can be sized or otherwise configured so that the weight and/or mass of the sample cell holder 820 is evenly distributed from left to right in the view of
As described above with reference to
The energy source 2012 may comprise a broad-band and/or a narrow-band source of electromagnetic energy. In some embodiments, the energy source 2012 comprises optical elements such as, e.g., filters, collimators, lenses, mirrors, etc., that are adapted to produce a desired energy beam E. For example, in some embodiments, the energy beam E is an infrared beam in a wavelength range between about 2 μm and 20 μm. In some embodiments, the energy beam E comprises an infrared beam in a wavelength range between about 4 μm and 10 μm. In the infrared wavelength range, water generally is the main contributor to the total absorption together with features from absorption of other blood components, particularly in the 6 μm-10 μm range. The 4 μm to 10 μm wavelength band has been found to be advantageous for determining glucose concentration, because glucose has a strong absorption peak structure from about 8.5 μm to 10 μm, whereas most other blood components have a relatively low and flat absorption spectrum in the 8.5 μm to 10 μm range. Two exceptions are water and hemoglobin, which are interferents in this range.
The energy beam E may be temporally modulated to provide increased signal-to-noise ratio (S/N) of the measurements provided by the analyzer 2010 as further described below. For example, in some embodiments, the beam E is modulated at a frequency of about 10 Hz or in a range from about 1 Hz to about 30 Hz. A suitable energy source 2012 may be an electrically modulated thin-film thermoresistive element such as the HawkEye IR-50 available from Hawkeye Technologies of Milford, Conn.
As depicted in
In the embodiment shown in
The filter wheel motor 2016 rotates the filter wheel 2018 to position the filters 2015 in the energy beam E to sequentially vary the wavelengths or the wavelength bands used to analyze the fluid sample. In some embodiments, each individual filter 2015 is disposed in the energy beam E for a dwell time during which optical properties in the passband of the filter are measured for the sample. The filter wheel motor 2016 then rotates the filter wheel 2018 to position another filter 2015 in the beam E. In some embodiments, 25 narrow-band filters are used in the filter wheel 2018, and the dwell time is about 2 seconds for each filter 2015. A set of optical measurements for all the filters can be taken in about 2 minutes, including sampling time and filter wheel movement. In some embodiments, the dwell time may be different for different filters 2015, for example, to provide a substantially similar S/N ratio for each filter measurement. Accordingly, the filter wheel assembly 2021 functions as a varying-passband filter that allows optical properties of the sample to be analyzed at a number of wavelengths or wavelength bands in a sequential manner.
In some embodiments of the analyzer 2010, the filter wheel 2018 includes 25 finite-bandwidth infrared filters having a Gaussian transmission profile and full-width half-maximum (FWHM) bandwidth of 28 cm−1 corresponding to a bandwidth that varies from 0.14 μm at 7.08 μm to 0.28 μm at 10 The central wavelength of the filters are, in microns: 7.082, 7.158, 7.241, 7.331, 7.424, 7.513, 7.605, 7.704, 7.800, 7.905, 8.019, 8.150, 8.271, 8.598, 8.718, 8.834, 8.969, 9.099, 9.217, 9.346, 9.461, 9.579, 9.718, 9.862, and 9.990.
With further reference to
As illustrated in
In the embodiment of the analyzer 2010 shown in
The centrifuge wheel 2050 can be spun by a centrifuge motor 2026. In some embodiments of the analyzer 2010, the fluid sample (e.g., a whole blood sample) is spun at a certain number of revolutions per minute (RPM) for a given length of time to separate blood plasma for spectral analysis. In some embodiments, the fluid sample is spun at about 7200 RPM. In some embodiments, the fluid sample is spun at about 5000 RPM or 4500 RPM. In some embodiments, the fluid sample is spun at more than one rate for successive time periods. The length of time can be approximately 5 minutes. In some embodiments, the length of time is approximately 2 minutes. In some embodiments, an anti-clotting agent such as heparin may be added to the fluid sample before centrifuging to reduce clotting. With reference to
The embodiment of the analyzer 2010 illustrated in
In some embodiments, the reference detector 2036 and the sample detector 2030 comprise broadband pyroelectric detectors. As known in the art, some pyroelectric detectors are sensitive to vibrations. Thus, for example, the output of a pyroelectric infrared detector is the sum of the exposure to infrared radiation and to vibrations of the detector. The sensitivity to vibrations, also known as “microphonics,” can introduce a noise component to the measurement of the reference and sample energy beams Er, Es using some pyroelectric infrared detectors. Because it may be desirable for the analyzer 2010 to provide high signal-to-noise ratio measurements, such as, e.g., S/N in excess of 100 dB, some embodiments of the analyzer 2010 utilize one or more vibrational noise reduction apparatus or methods. For example, the analyzer 2010 may be mechanically isolated so that high S/N spectroscopic measurements can be obtained for vibrations below an acceleration of about 1.5 G.
In some embodiments of the analyzer 2010, vibrational noise can be reduced by using a temporally modulated energy source 2012 combined with an output filter. In some embodiments, the energy source 2012 is modulated at a known source frequency, and measurements made by the detectors 2036 and 2030 are filtered using a narrowband filter centered at the source frequency. For example, in some embodiments, the energy output of the source 2012 is sinusoidally modulated at 10 Hz, and outputs of the detectors 2036 and 2030 are filtered using a narrow bandpass filter of less than about 1 Hz centered at 10 Hz. Accordingly, microphonic signals that are not at 10 Hz are significantly attenuated. In some embodiments, the modulation depth of the energy beam E may be greater than 50% such as, for example, 80%. The duty cycle of the beam may be between about 30% and 70%. The temporal modulation may be sinusoidal or any other waveform. In embodiments utilizing temporally modulated energy sources, detector output may be filtered using a synchronous demodulator and digital filter. The demodulator and filter are software components that may be digitally implemented in a processor such as the algorithm processor 416. Synchronous demodulators, coupled with low pass filters, are often referred to as “lock in amplifiers.”
The analyzer 2010 may also include a vibration sensor 2032 (e.g., one or more accelerometers) disposed near one (or both) of the detectors 2036 and 2030. The output of the vibration sensor 2032 is monitored, and suitable actions are taken if the measured vibration exceeds a vibration threshold. For example, in some embodiments, if the vibration sensor 2032 detects above-threshold vibrations, the system discards any ongoing measurement and “holds off” on performing further measurements until the vibrations drop below the threshold. Discarded measurements may be repeated after the vibrations drop below the vibration threshold. In some embodiments, if the duration of the “hold off” is sufficiently long, the fluid in the sample cell 2030 is flushed, and a new fluid sample is delivered to the cell 2030 for measurement. The vibration threshold may be selected so that the error in analyte measurement is at an acceptable level for vibrations below the threshold. In some embodiments, the threshold corresponds to an error in glucose concentration of 5 mg/dL. The vibration threshold may be determined individually for each filter 2015.
Certain embodiments of the analyzer 2010 include a temperature system (not shown in
The analyzer 2010 illustrated in
In some embodiments, the analyzer 2010 performs spectroscopic measurements on the fluid sample (known as a “wet” reading) and on one or more reference samples. For example, an “air” reading occurs when the sample detector 2036 measures the sample signal without the sample cell 2048 in place along the optical axis X. (This can occur, for example, when the opposite opening 1530 is aligned with the optical axis X). A “water” or “saline” reading occurs when the sample cell 2048 is filled with water or saline, respectively. The algorithm processor 416 may be programmed to calculate analyte concentration using a combination of these spectral measurements.
In some embodiments, a pathlength corrected spectrum is calculated using wet, air, and reference readings. For example, the transmittance at wavelength denoted by Ti, may be calculated according to Ti=(Si(wet)/Ri(wet))/(Si(air)/Ri(air)), where Si denotes the sample signal from the sample detector 2036 and Ri denotes the corresponding reference signal from the reference detector 2030. In some embodiments, the algorithm processor 416 calculates the optical density, ODi, as a logarithm of the transmittance, e.g., according to ODi=−Log(Ti). In one implementation, the analyzer 2010 takes a set of wet readings in each of the N filter passbands and then takes a set of air readings in each of the N filter passbands. In other embodiments, the analyzer 2010 may take an air reading before (or after) the corresponding wet reading.
The optical density ODi is the product of the absorption coefficient at wavelength λi, αi, times the pathlength L over which the sample energy beam Es interacts with the substance in the sample cell 2048, e.g., ODi=αi L. The absorption coefficient αi of a substance may be written as the product of an absorptivity per mole times a molar concentration of the substance.
The pathlength L may be used to determine the absorption coefficients of the fluid sample at each wavelength. Molar concentration of an analyte of interest can be determined from the absorption coefficient and the known molar absorptivity of the analyte. In some embodiments, a sample measurement cycle comprises a saline reading (at one or more wavelengths), a set of N wet readings (taken, for example, through a sample cell 2048 containing saline solution), followed by a set of N air readings (taken, for example, through the opposite opening 1530). As discussed above, the sample measurement cycle can be performed in a given length of time that may depend, at least in part, on filter dwell times. For example, the measurement cycle may take five minutes when the filter dwell times are about five seconds. In some embodiments, the measurement cycle may take about two minutes when the filter dwell times are about two seconds. After the sample measurement cycle is completed, a detergent cleaner may be flushed through the sample cell 2048 to reduce buildup of organic matter (e.g., proteins) on the windows of the sample cell 2048. The detergent is then flushed to a waste bladder.
In some embodiments, the system stores information related to the spectral measurements so that the information is readily available for recall by a user. The stored information can include wavelength-dependent spectral measurements (including fluid sample, air, and/or saline readings), computed analyte values, system temperatures and electrical properties (e.g., voltages and currents), and any other data related to use of the system (e.g., system alerts, vibration readings, S/N ratios, etc.). The stored information may be retained in the system for a time period such as, for example, 30 days. After this time period, the stored information may be communicated to an archival data storage system and then deleted from the system. In some embodiments, the stored information is communicated to the archival data storage system via wired or wireless methods, e.g., over a hospital information system (HIS).
The algorithm processor 416 (
Interferents can comprise components of a material sample being analyzed for an analyte, where the presence of the interferent affects the quantification of the analyte. Thus, for example, in the spectroscopic analysis of a sample to determine an analyte concentration, an interferent could be a compound having spectroscopic features that overlap with those of the analyte, in at least a portion of the wavelength range of the measurements. The presence of such an interferent can introduce errors in the quantification of the analyte. More specifically, the presence of one or more interferents can affect the sensitivity of a measurement technique to the concentration of analytes of interest in a material sample, especially when the system is calibrated in the absence of, or with an unknown amount of, the interferent.
Independently of or in combination with the attributes of interferents described above, interferents can be classified as being endogenous (i.e., originating within the body) or exogenous (i.e., introduced from or produced outside the body). As an example of these classes of interferents, consider the analysis of a blood sample (or a blood component sample or a blood plasma sample) for the analyte glucose. Endogenous interferents include those blood components having origins within the body that affect the quantification of glucose, and can include water, hemoglobin, blood cells, and any other component that naturally occurs in blood. Exogenous interferents include those blood components having origins outside of the body that affect the quantification of glucose, and can include items administered to a person, such as medicaments, drugs, foods or herbs, whether administered orally, intravenously, topically, etc.
Independently of or in combination with the attributes of interferents described above, interferents can comprise components which are possibly, but not necessarily, present in the sample type under analysis. In the example of analyzing samples of blood or blood plasma drawn from patients who are receiving medical treatment, a medicament such as acetaminophen is possibly, but not necessarily, present in this sample type. In contrast, water is necessarily present in such blood or plasma samples.
Certain disclosed analysis methods are particularly effective if each analyte and interferent has a characteristic signature in the measurement (e.g., a characteristic spectroscopic feature), and if the measurement is approximately affine (e.g., includes a linear term and an offset) with respect to the concentration of each analyte and interferent. In such methods, a calibration process is used to determine a set of one or more calibration coefficients and a set of one or more optional offset values that permit the quantitative estimation of an analyte. For example, the calibration coefficients and the offsets may be used to calculate an analyte concentration from spectroscopic measurements of a material sample (e.g., the concentration of glucose in blood plasma). In some of these methods, the concentration of the analyte is estimated by multiplying the calibration coefficient by a measurement value (e.g., an optical density) to estimate the concentration of the analyte. Both the calibration coefficient and measurement can comprise arrays of numbers. For example, in some embodiments, the measurement comprises spectra Cs(λi) measured at the wavelengths λi, and the calibration coefficient and optional offset comprise an array of values corresponding to each wavelength λi. In some embodiments, as further described below, a hybrid linear analysis (HLA) technique is used to estimate analyte concentration in the presence of a set of interferents, while retaining a high degree of sensitivity to the desired analyte. The data used to accommodate the set of possible interferents can include (a) signatures of each of the members of the family of potential additional substances and (b) a typical quantitative level at which each additional substance, if present, is likely to appear. In some embodiments, the calibration coefficient (and optional offset) are adjusted to minimize or reduce the sensitivity of the calibration to the presence of interferents that are identified as possibly being present in the fluid sample.
In some embodiments, the analyte analysis method uses a set of training spectra each having known analyte concentration and produces a calibration that minimizes the variation in estimated analyte concentration with interferent concentration. The resulting calibration coefficient indicates sensitivity of the measurement to analyte concentration. The training spectra need not include a spectrum from the individual whose analyte concentration is to be determined. That is, the term “training” when used in reference to the disclosed methods does not require training using measurements from the individual whose analyte concentration will be estimated (e.g., by analyzing a bodily fluid sample drawn from the individual).
Several terms are used herein to describe the analyte analysis process. The term “Sample Population” is a broad term and includes, without limitation, a large number of samples having measurements that are used in the computation of calibration values (e.g., calibration coefficients and optional offsets). In some embodiments, the term Sample Population comprises measurements (such as, e.g., spectra) from individuals and may comprise one or more analyte measurements determined from those same individuals. Additional demographic information may be available for the individuals whose sample measurements are included in the Sample Population. For an embodiment involving the spectroscopic determination of glucose concentration, the Sample Population measurements may include a spectrum (measurement) and a glucose concentration (analyte measurement).
Various embodiments of Sample Populations may be used in various embodiments of the systems and methods described herein. Several examples of Sample Populations will now be described. These examples are intended to illustrate certain aspects of possible Sample Population embodiments but are not intended to limit the types of Sample Populations that may be generated. In certain embodiments, a Sample Population may include samples from one or more of the example Sample Populations described below.
In some embodiments of the systems and methods described herein, one or more Sample Populations are included in a “Population Database.” The Population Database may be implemented and/or stored on a computer-readable medium. In certain embodiments, the systems and methods may access the Population Database using wired and/or wireless techniques. Certain embodiments may utilize several different Population Databases that are accessible locally and/or remotely. In some embodiments, the Population Database includes one or more of the example Sample Populations described below. In some embodiments, two or more databases can be combined into a single database, and in other embodiments, any one database can be divided into multiple databases.
An example Sample Population may comprise samples from individuals belonging to one or more demographic groups including, for example, ethnicity, nationality, gender, age, etc. Demographic groups may be established for any suitable set of one or more distinctive factors for the group including, for example, medical, cultural, behavioral, biological, geographical, religious, and genealogical traits. For example, in certain embodiments, a Sample Population includes samples from individuals from a specific ethnic group (e.g., Caucasians, Hispanics, Asians, African Americans, etc.). In another embodiment, a Sample Population includes samples from individuals of a specific gender or a specific race. In some embodiments, a Sample Population includes samples from individuals belonging to more than one demographic group (e.g., samples from Caucasian women).
Another example Sample Population can comprise samples from individuals having one or more medical conditions. For example, a Sample Population may include samples from individuals who are healthy and unmedicated (sometimes referred to as a Normal Population). In some embodiments, the Sample Population includes samples from individuals having one or more health conditions (e.g., diabetes). In some embodiments, the Sample Population includes samples from individuals taking one or more medications. In certain embodiments, Sample Population includes samples from individuals diagnosed to have a certain medical condition or from individuals being treated for certain medical conditions or some combination thereof. The Sample Population may include samples from individuals such as, for example, ICU patients, maternity patients, and so forth.
An example Sample Population may comprise samples that have the same interferent or the same type of interferents. In some embodiments, a Sample Population can comprise multiple samples, all lacking an interferent or a type of interferent. For example, a Sample Population may comprise samples that have no exogenous interferents, that have one or more exogenous interferents of either known or unknown concentration, and so forth. The number of interferents in a sample depends on the measurement and analyte(s) of interest, and may number, in general, from zero to a very large number (e.g., greater than 300). All of the interferents typically are not expected to be present in a particular material sample, and in many cases, a smaller number of interferents (e.g., 0, 1, 2, 5, 10, 15, 20, or 25) may be used in an analysis. In certain embodiments, the number of interferents used in the analysis is less than or equal to the number of wavelength-dependent measurements N in the spectrum Cs(λi).
Certain embodiments of the systems and methods described herein are capable of analyzing a material sample using one or more Sample Populations (e.g., accessed from the Population Database). Certain such embodiments may use information regarding some or all of the interferents which may or may not be present in the material sample. In some embodiments, a list of one or more possible interferents, referred to herein as forming a “Library of Interferents,” can be compiled. Each interferent in the Library can be referred to as a “Library Interferent.” The Library Interferents may include exogenous interferents and endogenous interferents that may be present in a material sample. For example, an interferent may be present due to a medical condition causing abnormally high concentrations of the exogenous and endogenous interferents. In some embodiments, the Library of Interferents may not include one or more interferents that are known to be present in all samples. Thus, for example, water, which is a glucose interferent for many spectroscopic measurements, may not be included in the Library of Interferents. In certain embodiments, the systems and methods use samples in the Sample Population to train calibration methods.
The material sample being measured, for example a fluid sample in the sample cell 2048, may also include one or more Library Interferents which may include, but is not limited to, an exogenous interferent or an endogenous interferent. Examples of exogenous interferent can include medications, and examples of endogenous interferents can include urea in persons suffering from renal failure. In addition to components naturally found in the blood, the ingestion or injection of some medicines or illicit drugs can result in very high and rapidly changing concentrations of exogenous interferents.
In some embodiments, measurements of a material sample (e.g., a bodily fluid sample), samples in a Sample Population, and the Library Interferents comprise spectra (e.g., infrared spectra). The spectra obtained from a sample and/or an interferent may be temperature dependent. In some embodiments, it may be beneficial to calibrate for temperatures of the individual samples in the Sample Population or the interferents in the Library of Interferents. In some embodiments, a temperature calibration procedure is used to generate a temperature calibration factor that substantially accounts for the sample temperature. For example, the sample temperature can be measured, and the temperature calibration factor can be applied to the Sample Population and/or the Library Interferent spectral data. In some embodiments, a water or saline spectrum is subtracted from the sample spectrum to account for temperature effects of water in the sample.
In other embodiments, temperature calibration may not be used. For example, if Library Interferent spectra, Sample Population spectra, and sample spectra are obtained at approximately the same temperature, an error in a predicted analyte concentration may be within an acceptable tolerance. If the temperature at which a material sample spectrum is measured is within, or near, a temperature range (e.g., several degrees Celsius) at which the plurality of Sample Population spectra are obtained, then some analysis methods may be relatively insensitive to temperature variations. Temperature calibration may optionally be used in such analysis methods.
An example embodiment of the method 2100 of
In block 2120, a statistical comparison of the absorption spectrum of the sample S with a spectrum of the Sample Population and combinations of individual Library Interferent spectra is performed. The statistical comparison provides a list of Library Interferents that are possibly contained in sample S and can include either no Library Interferents or one or more Library Interferents. In this example, in block 2130, one or more sets of spectra are generated from spectra of the Sample Population and their respective known analyte concentrations and known spectra of the Library Interferents identified in block 2120. In block 2130, the generated spectra are used to calculate a model for predicting the analyte concentration from the obtained measurement. In some embodiments, the model comprises one or more calibration coefficients κ(λi) that can be used with the sample measurements Cs(λi) to provide an estimate of the analyte concentration, gest. In block 2140, the estimated analyte concentration is determined form the model generated in block 2130. For example, in some embodiments of HLA, the estimated analyte concentration is calculated according to a linear formula: gest=κ(λi)·Cs(λi). Because the absorption measurements and calibration coefficients may represent arrays of numbers, the multiplication operation indicated in the preceding formula may comprise a sum of the products of the measurements and coefficients (e.g., an inner product or a matrix product). In some embodiments, the calibration coefficient is determined so as to have reduced or minimal sensitivity to the presence of the identified Library Interferents.
An example embodiment of block 2120 of the method 2100 will now be described with reference to
In some embodiments of block 2210, the statistical sample model comprises a mean spectrum and a covariance matrix calculated for the Sample Population. For example, if each spectrum measured at N wavelengths λi is represented by an N×1 array, C, then the mean spectrum, μ, is an N×1 array having values at each wavelength averaged over the range of spectra in the Sample Population. The covariance matrix, V, is calculated as the expected value of the deviation between C and μ and can be written as V=E((C−μ)(C−μ)T) where E(·) represents the expected value and the superscript T denotes transpose. In other embodiments, additional statistical parameters may be included in the statistical model of the Sample Population spectra.
Additionally, a Library of Interferents may be assembled in block 2220. A number of possible interferents can be identified, for example, as a list of possible medications or foods that might be ingested by the population of patients at issue. Spectra of these interferents can be obtained, and a range of expected interferent concentrations in the blood, or other expected sample material, can be estimated. In certain embodiments, the Library of Interferents includes, for each of “M” interferents, the absorption spectrum normalized to unit interferent concentration of each interferent, IF={IF1, IF2, . . . , IFM}, and a range of concentrations for each interferent from Tmax={Tmax1, Tmax2, . . . TmaxM) to Tmin={Tmin1, Tmin2, . . . , TminM). Information in the Library may be assembled once and accessed as needed. For example, the Library and the statistical model of the Sample Population may be stored in a storage device associated with the algorithm processor 416 (see,
Continuing in block 2225, the algorithm processor 416 assembles one or more subsets comprising a number K of spectra taken from the Library of Interferents. The number K may be an integer such as, for example, 1, 2, 3, 4, 5, 6, 10, 16, or more. In some embodiments, the subsets comprise all combinations of the M Library spectra taken K at a time. In these embodiments, the number of subsets having K spectra is M!/(K! (M−K)!), where ! represents the factorial function.
Continuing in block 2230, the obtained measurement data (e.g., the sample spectrum) and the statistical Sample Population model (e.g., the mean spectrum and the covariance matrix) are compared with data for each subset of interferents determined in block 2225 in order to determine the presence of possible interferents in the sample (block 2240). In some embodiments, the statistical test for the presence of an interferent subset in block 2240 comprises determining the concentrations of each subset of interferences that minimize a statistical measure of “distance” between a modified spectrum of the material sample and the statistical model of the Sample Population (e.g., the mean μ and the covariance V). The term “concentration” used in this context refers to a computed value, and, in some embodiments, that computed value may not correspond to an actual concentration. The concentrations may be calculated numerically. In some embodiments, the concentrations are calculated by algebraically solving a set of linear equations. The statistical measure of distance may comprise the well-known Mahalanobis distance (or square of the Mahalanobis distance) and/or some other suitable statistical distance metric (e.g., Hotelling's T-square statistic). In certain implementations, the modified spectrum is given by C′s(T)=Cs−IF·T where T=(T1, T2, . . . TK)T is a K-dimensional column vector of interferent concentrations and IF={IF1, IF2, . . . IFK} represents the K interferent absorption spectra of the subset. In some embodiments, concentration of the ith interferent is assumed to be in a range from a minimum value, Tmini, to a maximum value, Tmaxi. The value of Tmini may be zero, or may be a value between zero and Tmaxi, such as a fraction of Tmaxi, or may be a negative value. Negative values represent interferent concentrations that are smaller than baseline interferent values in the Sample Population.
In block 2250, a list of a number NS of possible interferent subsets ξ may be identified as the particular subsets that pass one or more statistical tests (in block 2240) for being present in the material sample. One or more statistical tests may be used, alone or in combination, to identify the possible interferents. For example, if a statistical test indicates that an ith interferent is present in a concentration outside the range Tmini to Tmaxi, then this result may be used to exclude the ith interferent from the list of possible interferents. In some embodiments, only the single most probable interferent subset is included on the list, for example, the subset having the smallest statistical distance (e.g., Mahalanobis distance). In an embodiment, the list includes the subsets having statistical distances smaller than a threshold value. In certain embodiments, the list includes a number NS of subsets having the smallest statistical distances, e.g., the list comprises the “best” candidate subsets. The number NS may be any suitable integer such as 10, 20, 50, 100, 200, or more. An advantage of selecting the “best” NS subsets is reduced computational burden on the algorithm processor 416. In some embodiments, the list includes all the Library Interferents. In certain such embodiments, the list is selected to comprise combinations of the NS subsets taken L at a time. For example, in some embodiments, pairs of subsets are taken (e.g., L=2). An advantage of selecting pairs of subsets is that pairing captures the most likely combinations of interferents and the “best” candidates are included multiple times in the list of possible interferents. In embodiments in which combinations of L subsets are selected, the number of combinations of subsets in the list of possible interferent subsets is Ns!/(L! (NS−L)!).
In other embodiments, the list of possible interferent subsets ξ is determined using a combination of some or all of the above criteria. In another embodiment, the list of possible interferent subsets ξ includes each of the subsets assembled in block 2225. Many selection criteria are possible for the list of possible interferent subsets ξ.
Returning to
In one example embodiment for block 2310, synthesized Sample Population spectra are generated by adding random concentrations of each interferent in one of the possible interferent subsets ξ. These spectra are referred to herein as an Interferent-Enhanced Spectral Database or IESD. In one example method, the IESD is formed as follows. A plurality of Randomly-Scaled Single Interferent Spectra (RSIS) are formed for each interferent in the interferent subset ξ. Each RSIS is formed by combinations of the interferent having spectrum IF multiplied by the maximum concentration Tmax, which is scaled by a random factor between zero and one. In certain embodiments, the scaling places the maximum concentration at the 95th percentile of a log-normal distribution in order to generate a wide range of concentrations. In some embodiments, the log-normal distribution has a standard deviation equal to half of its mean value.
In this example method, individual RSIS are then combined independently and in random combinations to form a large family of Combination Interferent Spectra (CIS), with each spectrum in the CIS comprising a random combination of RSIS, selected from the full set of identified Library Interferents. An advantage of this method of selecting the CIS is that it produces adequate variability with respect to each interferent, independently across separate interferents.
The CIS and replicates of the Sample Population spectra are combined to form the IESD. Since the interferent spectra and the Sample Population spectra may have been obtained from measurements having different optical pathlengths, the CIS may be scaled to the same pathlength as the Sample Population spectra. The Sample Population Database is then replicated R times, where R depends on factors including the size of the Database and the number of interferents. The IESD includes R copies of each of the Sample Population spectra, where one copy is the original Sample Population Data, and the remaining R−1 copies each have one randomly chosen CIS spectra added. Accordingly, each of the IESD spectra has an associated analyte concentration from the Sample Population spectra used to form the particular IESD spectrum. In some embodiments, a 10-fold replication of the Sample Population Database is used for 130 Sample Population spectra obtained from 58 different individuals and 18 Library Interferents. A smaller replication factor may be used if there is greater spectral variety among the Library Interferent spectra, and a larger replication factor may be used if there is a greater number of Library Interferents.
After the IESD is generated in block 2310, in block 2360, the IESD spectra and the known, random concentrations of the subset interferents are used to generate a calibration coefficient for estimating the analyte concentration from a sample measurement. The calibration coefficient is calculated in some embodiments using a hybrid linear analysis (HLA) technique. In certain embodiments, the HLA technique uses a reference analyte spectrum to construct a set of spectra that are free of the desired analyte, projecting the analyte's spectrum orthogonally away from the space spanned by the analyte-free calibration spectra, and normalizing the result to produce a unit response. Further description of embodiments of HLA techniques may be found in, for example, “Measurement of Analytes in Human Serum and Whole Blood Samples by Near-Infrared Raman Spectroscopy,” Chapter 4, Andrew J. Berger, Ph. D. thesis, Massachusetts Institute of Technology, 1998, and “An Enhanced Algorithm for Linear Multivariate Calibration,” by Andrew J. Berger, et al., Analytical Chemistry, Vol. 70, No. 3, Feb. 1, 1998, pp. 623-627, the entirety of each of which is hereby incorporated by reference herein. In other embodiments, the calibration coefficients may be calculated using other techniques including, for example, regression techniques such as, for example, ordinary least squares (OLS), partial least squares (PLS), and/or principal component analysis.
In block 2365, the processor 416 determines whether additional interferent subsets ξ remain in the list of possible interferent subsets. If another subset is present in the list, the acts in blocks 2310-2360 are repeated for the next subset of interferents using different random concentrations. In some embodiments, blocks 2310-2360 are performed for only the most probable subset on the list.
The calibration coefficient determined in block 2360 corresponds to a single interferent subset from the list of possible interferent subsets and is denoted herein as a single-interferent-subset calibration coefficient κavg(ξ). In this example method, after all subsets have been processed, the method continues in block 2370, in which the single-interferent-subset calibration coefficient is applied to the measured spectra Cs to determine an estimated, single-interferent-subset analyte concentration, g(ξ)=κavg(ξ)·Cs, for the interferent subset ξ. The set of the estimated, single-interferent-subset analyte concentrations g(ξ) for all subsets in the list may be assembled into an array of single-interferent-subset concentrations. As noted above, in some embodiments the blocks 2310-2370 are performed once for the most probable single-interferent-subset on the list (e.g., the array of single-interferent analyte concentrations has a single member).
Returning to block 2140 of
In certain embodiments, the estimated analyte concentration, gest, is determined (in block 2140) by combining the single-interferent-subset estimates, g(ξ), and the weighting functions, p(ξ), to generate an average analyte concentration. The average concentration may be computed according to gest=Σg(ξ) p(ξ), where the sum is over the interferent subsets processed from the list of possible interferent subsets. In some embodiments, the weighting function p(ξ) is a constant value for each subset (e.g., a standard arithmetic average is used for determining average analyte concentration). By testing the above described example method on simulated data, it has been found that the average analyte concentration advantageously has errors that may be reduced in comparison to other methods (e.g., methods using only a single most probable interferent).
Although the flowchart in
For example, in some embodiments of the method 2100, the calibration coefficient is computed without synthesizing spectra and/or partitioning the data into calibration sets and test sets. Such embodiments are referred to herein as “Parameter-Free Interferent Rejection” (PFIR) methods. In one example embodiment using PFIR, for each of the possible interferent subsets ξ, the following calculations may be performed to compute an estimate of a calibration coefficient for each subset ξ. An average concentration may be estimated according to gest=Σg(ξ) p(ξ), where the sum is over the interferent subsets processed from the list of possible interferent subsets.
An example of an alternative embodiment of block 2130 includes the following steps and calculations.
Step 1: For a subset's NIF interferents, form a scaled interferent spectra matrix. In certain embodiments, the scaled interferent spectra matrix is the product of an interferent spectral matrix, IF, multiplied by an interferent concentration matrix, Tmax, and can be written as: IF Tmax. In certain such embodiments, the interferent concentration matrix Tmax is a diagonal matrix having entries given by the maximum plasma concentrations for the various interferents.
Step 2: Calculate a covariance for the interferent component. If X denotes the IESD, the covariance of X, cov(X), is defined as the expectation E((X−mean(X))(X−mean(X))T) and is
cov(X)≈XXT/(N−1)−mean(X)mean(X)T. (1)
As described above, the IESD (e.g., X) is obtained as a combination of Sample Population Spectra, C, with Combination Interferent Spectra (CIS): Xj=Cj+IFj ξj, therefore the covariance is:
cov(X)≈CCT/(N−1)+IFΞΞTIFT/(N−1)−mean(X)mean(X)T, (2)
which can be written as,
cov(X)≈cov(C)+IFcov(Ξ) IFT. (3)
If the weights in the weighting matrix Ξ are independent and identically distributed, the covariance of Ξ, cov(Ξ), is a diagonal matrix having along the diagonal the variance, v, of the samples in Ξ. The last equation may be written as
cov(X)≈V0+vΦ, (4)
where V0 is the covariance of the original sample population and Φ is the covariance of the IF spectral set.
Step 3: The group's covariance may be at least partially corrected for the presence of a single replicate of the Sample Population spectra with the IESD as formed from NIF replicates of the Sample Population Spectra with Combined Interferent Spectra. This partial correction may be achieved by multiplying the second term in the covariance formula given above by a correction factor ρ:
V=V
0
+ρvΦ, (5)
where ρ is a scalar weighting function that depends on the number of interferents in the group. In some embodiments, the scalar weighting function is ρ=NIF/(NIF+1). In certain embodiments, the variance v of the weights is assumed to be the variance of a log-normal random variable having a 95th percentile at a value of 1.0, and a standard deviation equal to half of the mean value.
Step 4: The eigenvectors and the corresponding eigenvalues of the covariance matrix V are determined using any suitable linear algebraic methods. The number of eigenvectors (and eigenvalues) is equal to the number of wavelengths L in the spectral measurements. The eigenvectors may be sorted based on decreasing order of their corresponding eigenvalues.
Step 5: The matrix of eigenvectors is decomposed so as to provide an orthogonal matrix Q. For example, in some embodiments, a QR-decomposition is performed, thereby yielding the matrix Q having orthonormal columns and rows.
Step 6: The following matrix operations are performed on the orthogonal matrix Q. For n=2 to L−1, the product P∥n=Q(:,1:n) Q(:,1:n)T is calculated, where Q(:,1:n) denotes the submatrix comprising the first n columns of the full matrix Q. The orthogonal projection, P⊥n, away from the space spanned by Q(:,1:n) is determined by subtracting P∥n from the L×L identity matrix I. The nth calibration vector is then determined from κn=P⊥nαXαXTP⊥nαX, and the nth error variance En is determined as the projection of the full covariance V onto the subspace spanned by κn as follows: En=κnTV κn.
The steps 4-6 of this example are an embodiment of the HLA technique.
In some embodiments, the calibration coefficient κ is selected as the calibration vector corresponding to the minimum error variance En. Thus, for example, the average group calibration coefficient κ may be found by searching among all the error variances for the error variance κ that has the minimum value. The calibration coefficient is then selected as the nth calibration vector κn corresponding to the minimum error variance En. In other embodiments, the calibration coefficient is determined by averaging some or all of the calibration vectors κn.
Embodiments of the above-described methods have been used to estimate blood plasma glucose concentrations in humans. Four example experiments will now be described. The population of individuals from whom samples were obtained for analysis (estimation of glucose concentration) will be referred to as the “target population.” Infrared spectra obtained from the target population will be referred to as the “target spectra.” In the four example experiments, the target population included 41 intensive care unit (ICU) patients. Fifty-five samples were obtained from the target population.
In this example experiment, a partial least squares (PLS) regression method was applied to the infrared target spectra of the target patients' blood plasma to obtain the glucose estimates. In example experiment 1, estimated glucose concentration was not corrected for effects of interferents. The Sample Population used for the analysis included infrared spectra and independently measured glucose concentrations for 92 individuals selected from the general population. This Sample Population will be referred to as a “Normal Population.”
In example experiment 2, an embodiment of the Parameter-Free Interferent Rejection (PFIR) method was used to estimate glucose concentration for the same target population of patients in example experiment 1. The Sample Population was the Normal Population. In this example, calibration for Library Interferents was applied to the measured target spectra. The Library of Interferents included spectra of the 59 substances listed below:
In some embodiments, the calibration data set is determined according to two criteria: the calibration method itself (e.g., HLA, PLS, OLS, PFIR) and the intended application of the method. The calibration data set may comprise spectra and corresponding analyte levels derived from a set of plasma samples from the Sample Population. In some embodiments, e.g., those where an HLA calibration method is used, the calibration data set may also include spectra of the analyte of interest.
In the example experiments 1 and 2, the Sample Population was the Normal Population. Thus, samples were drawn from a population of normal individuals who did not have identifiable medical conditions that might affect the spectra of their plasma samples. For example, the sample plasma spectra typically did not show effects of high levels of medications or other substances (e.g., ethanol), or effects of chemicals that are indicative of kidney or liver malfunction.
In some embodiments, an analysis method may calibrate for deviations from the distribution defined by the calibration plasma spectra by identifying a “base” set of interferent spectra likely to be responsible for the deviation. The analysis method may then recalibrate with respect to an enhanced spectral data set. In some embodiments, the enhancement can be achieved by including the identified interferent spectra into the calibration plasma spectra. When it is anticipated that the target population may have been administered significant amounts of substances not present in the samples of the calibration set, or when the target population have many distinct interferents, estimation of the interferents present in the target spectrum may be subject to a large degree of uncertainty. In some cases, this may cause analyte estimation to be subject to errors.
Accordingly, in certain embodiments, the calibration data set may be enhanced beyond the base of “normal” samples to include a population of samples intended to be more representative of the target population. The enhancement of the calibration set may be generated, in some embodiments, by including samples from a sufficiently diverse range of individuals in order to represent the range of likely interferents (both in type and in concentration) and/or the normal variability in underlying plasma characteristics. The enhancement may, additionally or alternatively, be generated by synthesizing interferent spectra having a range of concentrations as described above (see, e.g., discussion of block 2310 in
Example experiments 3 and 4 use the analysis methods of example experiments 1 and 2, respectively (PLS without interferent correction and PFIR with interferent correction). However, example experiments 3 and 4 use a Sample Population having blood plasma spectral characteristics different from the Normal Population used in example experiments 1 and 2. In example experiments 3 and 4, the Sample Population was modified to include spectra of both the Normal Population and spectra of an additional population of 55 ICU patients. These spectra will be referred to as the “Normal+Target Spectra.” In experiments 3 and 4, the ICU patients included Surgical ICU patients, Medical ICU patients as well as victims of severe trauma, including a large proportion of patients who had suffered major blood loss. Major blood loss may necessitate replacement of the patient's total blood volume multiple times during a single day and subsequent treatment of the patient via electrolyte and/or fluid replacement therapies. Major blood loss may also require administration of plasma-expanding medications. Major blood loss may lead to significant deviations from the blood plasma spectra representative of a Normal Population. The population of 55 ICU patients (who provided the Target Spectra) has some similarities to the individuals for whom the analyses in experiments 1-4 were performed (e.g., all were ICU patients), but in these experiments, target spectra from individuals in the target population were not included in the Target Spectra.
Results of example experiments 1-4 are shown in the following table. The glucose concentrations estimated from the analysis method were compared to independently determined glucose measurements to provide an average prediction error and a standard deviation of the average prediction error. The table demonstrates that independent of the Sample Population used (e.g., either the Normal Population or the Normal+Target Population), calibrating for interferents reduces both the average prediction error and the standard deviation (e.g., compare the results for experiment 2 to the results for experiment 1 and compare the results for experiment 4 to the results for experiment 3). The table further demonstrates that independent of the analysis method used (e.g., either PLS or PFIR), using a Sample Population with more similarity to the target population (e.g., the Normal+Target Population) reduces both the average prediction error and the standard deviation (e.g., compare the results for experiment 3 to the results for experiment 1 and compare the results for experiment 4 to the results for experiment 2).
Accordingly, embodiments of analysis methods that use a Sample Population that includes both normal spectra and spectra from individuals similar to those of the target population and that calibrate for possible interferents provide a good match between the estimated glucose concentration and the measured glucose concentration. As discussed above, a suitable Sample Population may be assembled from the Population Database in order to include normal spectra plus suitable target spectra from individuals that match a desired target population including, for example, ICU patients, trauma patients, a particular demographic group, a group having a common medical condition (e.g., diabetes), and so forth.
The system 400 can include a display system 414, for example, as depicted in
The display system 414 can include a user interface 2400 by which users can conveniently and efficiently interact with the system 400. The user interface 2400 may be displayed on the output device of the system 400 (e.g., the touchscreen display). In some embodiments, the user interface 2400 is implemented and/or stored as one or more code modules, which may be embodied in hardware, firmware, and/or software.
The user interface 2400 can include an analyte concentration status graphic 2414 that indicates status of the patient's current analyte concentration compared with a reference standard. For example, the analyte may be glucose, and the reference standard may be a hospital ICU's tight glycemic control (TGC). In
The user interface 2400 can include one or more trend indicators 2416 that provide a graphic indicating the time history of the concentration of an analyte of interest. In
The user interface 2400 can include one or more buttons 2420-2426 that can be actuated by a user to provide additional functionality or to bring up suitable context-sensitive menus and/or screens. For example, in the embodiments shown in
The button 2422 (“Pause”) may be actuated by the user if patient monitoring is to be interrupted but is not intended to end. For example, the “Pause” button 2422 may be actuated if the patient is to be temporarily disconnected from the system 400 (e.g., by disconnecting the tubes 306). After the patient is reconnected, the button 2422 may be pressed again to resume monitoring. In some embodiments, after the “Pause” button 2422 has been pressed, the button 2422 displays “Resume.”
The button 2424 (“Delay 5 Minutes”) causes the system 400 to delay the next measurement by a delay time period (e.g., 5 minutes in the depicted embodiments). Actuating the delay button 2424 may be advantageous if taking a reading would be temporarily inconvenient, for example, because a health care professional is attending to other needs of the patient. The delay button 2424 may be pressed repeatedly to provide longer delays. In some embodiments, pressing the delay button 2424 is ineffective if the accumulated delay exceeds a maximum threshold. The next-reading graphic 2410 automatically increases the displayed time until the next reading for every actuation of the delay button 2424 (up to the maximum delay).
The button 2426 (“Dose History”) may be actuated to bring up a dosing history window that displays patient dosing history for an analyte or medicament of interest. For example, in some embodiments, the dosing history window displays insulin dosing history of the patient and/or appropriate hospital dosing protocols. A nurse attending the patient can actuate the dosing history button 2426 to determine the time when the patient last received an insulin dose, the last dosage amount, and/or the time and amount of the next dosage. The system 400 may receive the patient dosing history via wired or wireless communications from a hospital information system.
In other embodiments, the user interface 2400 can include additional and/or different buttons, menus, screens, graphics, etc. that are used to implement additional and/or different functionalities.
Related Components
The patient monitoring system 2630 shown in
The patient monitoring system 2630 can include an optional interface cable 2642. In some embodiments, the interface cable 2642 connects the monitoring apparatus 2632 to a patient monitor (not shown). The interface cable 2642 can be used to transfer data from the monitoring apparatus 2632 to the patient monitor for display. In some embodiments, the patient monitor is a bedside cardiac monitor having a display that is located in the patient room (see, e.g., the user interface 2400 shown in
In the embodiment shown in
The patient monitoring system 2630 includes a fluid system kit 2634 connected to the monitoring apparatus 2632. In some embodiments, the fluid system kit 2634 includes fluidic tubes that connect a fluid source to an analytic subsystem. For example, the fluidic tubes can facilitate fluid communication between a blood source or a saline source and an assembly including a sample holder and/or a centrifuge. In some embodiments, the fluid system kit 2634 includes many of the components that enable operation of the monitoring apparatus 2632. In some embodiments, the fluid system kit 2634 can be used with anti-clotting agents (such as heparin), saline, a saline infusion set, a patient catheter, a port sharing IV infusion pump, and/or an infusion set for an IV infusion pump, any or all of which may be made by a variety of manufacturers. In some embodiments, the fluid system kit 2634 includes a monolithic housing that is sterile and disposable. In some embodiments, at least a portion of the fluid system kit 2634 is designed for single patient use. For example, the fluid system kit 2634 can be constructed such that it can be economically discarded and replaced with a new fluid system kit 2634 for every new patient to which the patient monitoring system 2630 is connected. In addition, at least a portion of the fluid system kit 2634 can be designed to be discarded after a certain period of use, such as a day, several days, several hours, three days, a combination of hours and days such as, for example, three days and two hours, or some other period of time. Limiting the period of use of the fluid system kit 2634 may decrease the risk of malfunction, infection, or other conditions that can result from use of a medical apparatus for an extended period of time.
In some embodiments, the fluid system kit 2634 includes a connector with a luer fitting for connection to a saline source. The connector may be, for example, a three-inch pigtail connector. In some embodiments, the fluid system kit 2634 can be used with a variety of spikes and/or IV sets used to connect to a saline bag. In some embodiments, the fluid system kit 2634 also includes a three-inch pigtail connector with a luer fitting for connection to one or more IV pumps. In some embodiments, the fluid system kit 2634 can be used with one or more IV sets made by a variety of manufacturers, including IV sets obtained by a user of the fluid system kit 2634 for use with an infusion pump. In some embodiments, the fluid system kit 2634 includes a tube with a low dead volume luer connector for attachment to a patient vascular access point. For example, the tube can be approximately seven feet in length and can be configured to connect to a proximal port of a cardiovascular catheter. In some embodiments, the fluid system kit 2634 can be used with a variety of cardiovascular catheters, which can be supplied, for example, by a user of the fluid system kit 2634.
As shown in
In the embodiment shown in
In some embodiments, the computer system 2646 runs optional data analysis software 2648 that organizes and presents information obtained from one or more monitoring devices. In some embodiments, the data analysis software 2648 collects and analyzes data from the monitoring devices in an ICU. The data analysis software 2648 can also present charts, graphs, and statistics to a user of the computer system 2646.
In some embodiments, the computer system 2646 runs optional hospital information system (HIS) interface software 2650 that provides an interface point between one or more monitoring devices and an HIS. The HIS interface software 2650 may also be capable of communicating data between one or more monitoring devices and a laboratory information system (LIS).
In some embodiments, the computer system 2646 runs optional insulin dosing software 2652 that provides a platform for implementation of an insulin dosing regimen. In some embodiments, the hospital tight glycemic control protocol is included in the software. The protocol allows computation of proper insulin doses for a patient connected to a monitoring device 2646. The insulin dosing software 2652 can communicate with the monitoring device 2646 to ensure that proper insulin doses are calculated.
In some embodiments, it may be advantageous to control a level of an analyte (e.g., glucose) in a patient using an embodiment of an analyte detection system described herein. Although certain examples of glucose control are described below, embodiments of the systems and methods disclosed herein may be used to monitor and/or control other analytes (e.g., lactate).
For example, diabetic individuals control their glucose levels by administration of insulin. If a diabetic patient is admitted to a hospital or ICU, the patient may be in a condition in which he or she cannot self-administer insulin. Advantageously, embodiments of the analyte detection systems disclosed herein may be used to control the level of glucose in the patient. Additionally, it has been found that a majority of patients admitted to the ICU exhibit hyperglycemia without having diabetes. In such patients it may be beneficial to monitor and control their blood glucose level to be within a particular range of values. Further, it has been shown that tightly controlling blood glucose levels to be within a stringent range may be beneficial to patients undergoing surgical procedures.
A patient admitted to the ICU or undergoing surgery may be administered a variety of drugs and fluids such as Hetastarch, intravenous antibiotics, intravenous glucose, intravenous insulin, intravenous fluids such as saline, etc., which may act as interferents and make it difficult to determine the blood glucose level. Moreover, the presence of additional drugs and fluids in the blood stream may require different methods for measuring and controlling blood glucose level. Also, the patient may exhibit significant changes in hematocrit levels due to blood loss or internal hemorrhage, and there can be unexpected changes in the blood gas level or a rise in the level of bilirubin and ammonia levels in the event of an organ failure. Embodiments of the systems and methods disclosed herein advantageously may be used to monitor and control blood glucose (and/or other analytes) in the presence of possible interferents to estimation of glucose and for patients experiencing health problems.
In some environments, Tight Glycemic Control (TGC) can include: (1) substantially continuous monitoring (which can include periodic monitoring, at relatively frequent intervals of every 1, 5, 15, 30, 45, and/or 60 minutes, for example) of glucose levels; (2) determination of substances that tend to increase glucose levels (e.g., sugars such as dextrose) and/or decrease glucose levels (e.g., insulin); and/or (3) responsive delivery of one or more of such substances, if appropriate under the controlling TGC protocol. For example, one possible TGC protocol can be achieved by controlling glucose within a relatively narrow range (for example between 70 mg/dL to 110 mg/dL). As will be further described, in some embodiments, TGC may be achieved by using an analyte monitoring system to make continuous and/or periodic but frequent measurements of glucose levels.
In some embodiments, the analyte detection system schematically illustrated in
In an example of glycemic control, a system can be used to determine and monitor the concentration of glucose in the sample. If the concentration of glucose falls below a lower threshold, glucose from an external source can be supplied. If the concentration of glucose increases above an upper threshold, insulin from an external source can be supplied. In some embodiments, glucose or insulin may be infused in a patient continuously over a certain time interval or may be injected in a large quantity at once (referred to as “bolus injection”).
In some embodiments, a glycemic control system may be capable of delivering glucose, dextrose, glycogen, and/or glucagon from an external source relatively quickly in the event of hypoglycemia. As discussed, embodiments of the glycemic control system may be capable of delivering insulin from an external source relatively quickly in the event of hyperglycemia.
Returning to
As schematically illustrated in
In an example implementation for glycemic control, if the analyte detection system determines that the level of glucose has fallen below a lower threshold value (e.g., the patient is hypoglycemic), a control system (e.g., the fluid system controller 405 in some embodiments) controlling an infusion delivery system may close the pinch valves 521 and/or 542 to prevent infusion of insulin and/or saline into the patient. The control system may open the pinch valve 2786 and dextrose solution from the source 2782 can be infused (or alternatively injected as a bolus) into the patient. After a suitable amount of dextrose solution has been infused to the patient, the pinch valve 2786 can be closed, and the pinch valves 521 and/or 542 can be opened to allow flow of insulin and/or saline. In some systems, the amount of dextrose solution for infusion (or bolus injection) may be calculated based on one or more detected concentration levels of glucose. The source 2782 advantageously may be located at a short enough fluidic distance from the patient such that dextrose can be delivered to the patient within a time period of about one to about ten minutes. In other embodiments, the source 2782 can be located at the site where the patient tube 512 interfaces with the patient so that dextrose can be delivered within about one minute.
If the analyte detection system determines that the level of glucose has increased above an upper threshold value (e.g., the patient is hyperglycemic), the control system may close the pinch valves 542 and/or 2786 to prevent infusion of saline and/or dextrose into the patient. The control system may open the pinch valve 521, and insulin can be infused (or alternatively injected as a bolus) into the patient. After a suitable amount of insulin has been infused (or bolus injected) to the patient, the control system can close the pinch valve 521 and open the pinch valves 542 and/or 2786 to allow flow of saline and/or glucose. The suitable amount of insulin may be calculated based on one or more detected concentration levels of glucose in the patient. The insulin source 518 advantageously may be located at a short enough fluidic distance from the patient such that insulin can be delivered to the patient within about one to about ten minutes. In other embodiments, the insulin source 518 may be located at the site where the patient tube 512 interfaces with the patient so that insulin can be delivered to the patient within about one minute.
In some embodiments, sampling bodily fluid from a patient and providing medication to the patient may be achieved through the same lines of the fluid handling system. For example, in some embodiments, a port to a patient can be shared by alternately drawing samples and medicating through the same line. In some embodiments, a bolus can be provided to the patient at regular intervals (in the same or different lines). For example, a bolus of insulin can be provided to a patient after meals. In another embodiment comprising a shared line, a bolus of medication can be delivered when returning part of a body fluid sample back to the patient. In some implementations, the bolus of medication is delivered midway between samples (e.g., every 7.5 minutes if samples are drawn every 15 minutes). In other embodiment, a dual lumen tube can be used, wherein one lumen is used for the sample and the other lumen to medicate. In yet another embodiment, an analyte detection system (e.g., an “OptiScanner®” monitor) may provide suitable commands to a separate insulin pump (on a shared port or different line).
Example Method for Glycemic Control
In block 2725, if the estimated glucose concentration is outside the acceptable range of concentrations, then the method 2700 proceeds to block 2740 in which the estimated glucose concentration is compared with a desired glucose concentration. The desired glucose concentration may be based on, for example, the acceptable range of glucose concentrations, the parameters of the particular glycemic protocol, the patient's estimated glucose concentration, and so forth. If the estimated glucose concentration is below the desired concentration (e.g., the patient is hypoglycemic), a dose of dextrose to be delivered to the patient is calculated in block 2745. This calculation may take into account various factors including, for example, one or more estimated glucose concentrations, presence of additional drugs in the patient's system, time taken for dextrose to be assimilated by the patient, and the delivery method (e.g., continuous infusion or bolus injection). In block 2750, a fluid delivery system (e.g., a system such as the optional subsystem 2780 shown in
In block 2740, if the estimated glucose concentration is greater than the desired concentration (e.g., the patient is hyperglycemic), a dose of insulin to be delivered is calculated in block 2755. The dose of insulin may depend on various factors including, for example, one or more estimated glucose concentrations in the patient, presence of other drugs, type of insulin used, time taken for insulin to be assimilated by the patient, method of delivery (e.g., continuous infusion or bolus injection), etc. In block 2750, a fluid delivery system (e.g., the optional subsystem 2780 shown in
In block 2765, the method 2700 returns to block 2705 to await the start of the next measurement cycle, which may be within about one to about thirty minutes (e.g., every fifteen minutes). In some embodiments, the next measurement cycle begins at a different time than normally scheduled in cases in which the estimated glucose concentration lies outside the acceptable range of concentrations under the glycemic protocol. Such embodiments advantageously allow the system to monitor response of the patient to the delivered dose of dextrose (or insulin). In some such embodiments, the time between measurement cycles is reduced so the system can more accurately monitor analyte levels in the patient.
Examples of Some Possible Additional or Alternative Analytes
Although examples of control and/or monitoring has been described in the illustrative context of glycemic control, embodiments of the systems and methods can be configured for control and/or monitoring of one or more of many possible analytes, in addition to or instead of glucose. Monitor and/or control of analytes may be particularly helpful in ICUs, which receive patients experiencing trauma. For example, another parameter that can be monitored is level of Hemoglobin (Hb). If the Hb level of a patient goes down without an apparent external reason, the patient could be suffering from internal bleeding. Indeed, many ICU patients (some estimate as many as 10%) suffer from what appears to be spontaneous internal bleeding that may not be otherwise detectable until the consequences are too drastic to easily overcome. In some embodiments, level of Hb can be measured indirectly, because its relationship to oxygen in the veins and arteries (at different points in the vasculature with respect to the heart and lungs) is understood. In some embodiments, the apparatus, systems and methods described herein can be useful for measuring a level of Hb.
Another parameter that can be monitored is lactate level, which can be related to sepsis or toxic shock. Indeed, high levels and/or rapid rise in lactate levels can be correlated to organ failure and oxygenation problems in the blood and organs. However, other direct measures of the biological effects related to lactate level problems can be difficult to measure, for example, only becoming measurable with a delay (e.g., 2-6 hours later). Thus, measurement of lactate level can help provide a valuable early warning of other medical problems. Indeed, if a problem with lactate levels is detected, a nurse or doctor may be able to prevent the correlated problems by providing more fluids.
Another parameter that can be monitored is central venous oxygen saturation (ScvO2). It can be advantageous to try to maintain an ScvO2 of 65-70% or greater in ICU patients (to help avoid sepsis, for example). In some embodiments, the apparatus, systems, and methods described herein can be useful for measuring a level of ScvO2.
Levels of lactate and ScvO2 in a patient can be used together to provide information and/or warnings to a health care provider, which can be especially useful in an ICU setting. For example, if lactate and ScvO2 are both high, a warning can be provided (e.g., automatically using an alarm). If lactate is high, but ScvO2 is low, a patient may benefit from additional fluids. If ScvO2 is high, but lactate is low, a cardiac problem may be indicated. Thus, a system that provides information about both lactate and ScvO2 can be very beneficial to a patient, especially, for example, in the ICU environment. Although lactate and ScvO2 have been used as an illustrative example, in other embodiments different combinations of analytes may be monitored and used to provide information and/or warnings to a health care provider.
Method for Estimating Analyte Concentration in a Sample
One example method for estimating an analyte concentration of a sample is schematically illustrated in the flowchart 2800 of
One embodiment of the example method outlined in the flowchart 2800 for the determination of an analyte concentration in a sample will now be discussed. In this embodiment, the analyte of interest is glucose, and the sample is a blood sample S. The blood sample S may be whole blood, plasma, or blood that has been processed, refined, diluted, or treated in any number of ways. The algorithm processor 416 may access one or more calibration sets, as schematically illustrated at Block 2810. In some implementations, a plurality of calibration sets are accessed such as, e.g., two, three, four, five, six, or more calibration sets. Accordingly, in certain illustrative examples described herein, the methods and systems may be described as accessing, receiving, or using a plurality of calibration sets. However, this is not a limitation, and in other implementations a single calibration set can be used. Each of these calibration sets may include data usable to estimate the analyte concentration for a sample, such as a glucose concentration in the blood sample S. Each calibration set from the plurality of calibration sets may also include data that is sufficient to reconstruct the measurement of the sample. In the case where the measurement of the sample is an absorbance spectrum Cs(λi), each calibration set may include data sufficient to calculate a reconstructed spectrum from a measured spectrum derived from the blood sample S. In a further embodiment, each calibration set from the plurality of calibration sets may also contain a prediction eligibility threshold. The prediction eligibility threshold may be used to determine a maximum permissible difference between the reconstructed spectrum and the measured spectrum derived from the blood sample S. A prediction eligibility threshold may be calculated in a number of different ways. Prediction eligibility thresholds are discussed in more detail below.
In one embodiment, an algorithm processor 416 may access a measurement of the sample, as schematically illustrated at Block 2820. The measurement of the sample may be an absorbance spectrum Cs(λi) of a blood sample S. The absorbance spectrum, or measured spectrum, of the blood sample S may be a raw measurement from the optical system 412, or the raw measurement from the optical system 412 may be processed to create the measured spectrum accessed at Block 2820. The raw measurement from the optical system 412 may be processed in any number of different ways in order to create the measured spectrum accessed at Block 2820. For example, the raw measurement from the optical system 412 may be processed by removing the influence of saline from the measurement. This may be useful if a sample contains a predictable amount of saline, in order to help isolate the effect of the analyte of interest. The raw measurement from the optical system 412 may also be processed by normalizing the absorbance spectrum. This may be useful to minimize path length effects. The raw measurement from the optical system 412 may also be processed by centering it on itself by subtracting its average. These methods of processing the spectrum may each be used individually or together. Other methods of processing the spectrum may also be used.
In one embodiment, the algorithm processor 416 may determine, for each of a plurality of calibration sets, whether a calibration set is eligible to estimate the analyte concentration of the sample as schematically illustrated at Block 2830. One way to make this determination is to calculate a distance metric Mm for each of the plurality of calibration sets, based at least on the calibration set and on a measured spectrum, and compare this distance metric Mm with a prediction eligibility threshold.
A distance metric Mm may be calculated in numerous ways. For example, each calibration set may contain data usable to create a reconstructed spectrum based at least on the measured spectrum. If such a reconstructed spectrum is constructed, this reconstructed spectrum may be compared to the measured spectrum in order to calculate a distance metric in various ways. For example, a distance metric Mm may be a sum of the differences at each point between the reconstructed spectrum and the measured spectrum. The distance metric Mm may also be a weighted average of the differences at each point between the reconstructed spectrum and the measured spectrum. Weightings for this weighted average may be determined based upon sample measured spectra that were taken and compared to known analyte concentrations, and used to calculate which weightings may produce the most accurate analyte measurements. A distance metric Mm may also be calculated by calculating a ratio between the reconstructed spectrum and the measured spectrum or on other combinations of the reconstructed spectrum and the measured spectrum.
For example, one possible distance metric Mm may be based on a root-mean squared difference between each point on the reconstructed spectrum and the corresponding point on the measured spectrum. Other methods may also be used to determine a distance metric Mm. For example, a distance metric Mm may be calculated using the following formula:
M
m
=f
m(Rm(λm),Cs(λi)) (6)
where Mm is the distance metric for the mth calibration set, Rm(λi) is the value of the reconstructed spectrum for the mth calibration set at the ith wavelength, Cs(λi) is the value of the measured spectrum at the ith wavelength, and where fm is a function to calculate the distance metric Mm using one of the listed above methods, or any other method. For example, the distance metric Mm may be calculated by calculating a number of spectral residuals between the reconstructed spectrum Rm(λi) and the measured spectrum Cs(λi), squaring each of the spectral residuals, and summing those squares to form a sum of the squares of the residuals (SSR) value, according to the formula:
where Mm is a distance metric for the mth calibration set, Rm(λi) is the value of the reconstructed spectrum for the mth calibration set at the ith wavelength, Cs(λi) is the value of the measured spectrum at the ith wavelength, and where there are n wavelengths in both the reconstructed spectrum and the measured spectrum. This is merely one method which may be used to calculate a distance metric Mm. For example, in other embodiments, the square in formula (2) may be replaced by an absolute value or other function that returns a positive value for its argument(s).
Each calibration set may also include a prediction eligibility threshold value. This prediction eligibility threshold may be a scalar. The prediction eligibility threshold may be a vector value. In one embodiment, the prediction eligibility threshold for each of a plurality of calibration sets may be compared to the distance metric Mm for that calibration set to determine whether the calibration set is eligible to estimate the analyte concentration of a sample. Prediction eligibility thresholds are discussed in more depth herein.
The algorithm processor 416 may determine the estimated analyte concentration of the sample based at least in part on the eligible calibration sets, as schematically illustrated at Block 2840. If there is a single eligible calibration set, the estimated analyte concentration can be the analyte concentration estimated by the single eligible calibration set.
If there are multiple eligible calibration sets, an estimated analyte concentration from each of the multiple eligible calibration sets may be combined into a final estimated analyte concentration. This may be done in a number of different ways in different implementations. For example, each of the estimated analyte concentrations from the multiple eligible calibration sets may be arithmetically averaged to determine the final estimated analyte concentration. Alternatively, each of the estimated analyte concentrations from the multiple eligible calibration sets may be assigned a weight and a weighted average used. In some cases, the weights can be based upon the difference or the quotient between a distance metric Mm and a prediction eligibility threshold associated with that calibration set. In this embodiment, a calibration set with a distance metric Mm lower than its prediction eligibility threshold may be more heavily weighted than a calibration set with a distance metric Mm that is closer to its prediction eligibility threshold. In some embodiments, if the distance metric Mm for a calibration set is greater than the prediction eligibility threshold for that calibration set, the weight assigned to that calibration set may be zero, and that calibration set will not influence the final estimated analyte concentration. This may allow calibration sets that provide a more accurate estimate to be given more weight in estimating analyte concentration.
Alternatively, each of the estimated analyte concentrations from the multiple eligible calibration sets may be assigned a weight based upon whether that estimated analyte concentration is within a pre-determined range. This may be useful if there are undesirable analyte concentration values, and so it may be important to avoid giving analyte readings outside this undesirable area, if the true analyte concentration is within this dangerous area.
For example, if the analyte of interest is blood-glucose, it may be important that an individual avoid hypoglycemia. Hypoglycemia is a condition that occurs when an individual's blood glucose is too low. Blood glucose below about 70 mg/dL may be considered low, and blood glucose below this level may be harmful to the patient. Therefore, it may be useful in some implementations to avoid providing a blood glucose estimate that is above 70 mg/dL if the actual blood glucose level is below 70 mg/dL, because if a patient is hypoglycemic, it may be useful to provide them with carbohydrates such as dextrose, either orally or intravenously, or to lower their dose of insulin if the patient is diabetic Thus, if there are multiple eligible calibration sets, it may be advantageous to more heavily weight estimated glucose concentrations that are at or below 70 mg/dL, in order to minimize or reduce the chance of a “false negative” reading that suggests an individual is not hypoglycemic when, in fact, that individual is hypoglycemic. This may be useful as it may be important for health care professionals or others to know when an individual is hypoglycemic, so that the proper steps may be taken.
In the case where none of the plurality of calibration sets are eligible to estimate the analyte concentration of the sample (for example, where the distance metric Mm is greater than the prediction eligibility threshold for each calibration set m), the method may take an appropriate action. For example, the method may display a warning message, may notify a health care professional or other person responsible for monitoring the individual (perhaps through an electronic or computerized means such as email), may have a light come on, or may generate a null or “No Read” value. In one embodiment, the algorithm processor 416 may be connected to a display system 414. In the case of a no read, the display system 414 may display a null value (e.g. “---”) or indicate that the system is unable to provide an estimate of analyte concentration at this time. In another embodiment, the display system 414 may turn on a red light, or alert an operator that system is unable to provide an estimate of analyte concentration at this time. If there are no eligible calibration sets, the system may also attempt to analyze another sample more quickly than it otherwise would, in order to be able to provide an estimated analyte concentration. It may be preferable in these situations to display no analyte concentration rather than display a concentration that may be inaccurate.
Calibration Sets
At Block 2810, one or more (or a plurality) of calibration sets can be accessed, where each calibration set is usable to estimate an analyte concentration for a sample. The plurality of calibration sets may consist of two, three, four, five or more calibration sets. In one embodiment, the analyte of interest is glucose, and the sample is an absorbance spectrum based at least in part upon measurements of a body fluid sample provided by the optical system 412. The absorbance spectrum in this embodiment may comprise a number of wavelengths. For example, the absorbance system may contain 1, 2, 5, 12, 16, 24, 48, or 50 wavelengths. The wavelengths in the absorbance spectrum may be from any part of the electromagnetic spectrum. The number of wavelengths that are chosen, and the portion of the spectrum that the wavelengths are centered in may depend upon the properties of the analyte of interest. For example, if the analyte of interest is glucose, it may be useful to use wavelengths that are in the mid-infrared spectrum (e.g., 4-10 μm), because glucose has several absorbance peaks in the 7-10 μm region.
The one or more (or plurality) of calibration sets may be constructed outside of the system, and may be loaded onto the system as a part of a manufacturing process or as part of an upgrade to an existing system. These calibration sets may be stored locally on each system, or may be contained on a central network that the system connects to. Calibration sets may be updated from time-to-time, in order to increase their accuracy. The system may also be able to receive new calibration sets to add to those it already contains. When using calibration sets loaded from outside the system, it may be useful to cross-calibrate the calibration sets for each individual system, due to a need for precision and due to very slight irregularities in the manufacturing process. For example, the sensors in each system may provide slightly different readings of the same sample, and it may be useful to compensate for this effect when loading calibration sets that were made using a different system.
A calibration set may be constructed in many different ways. For example, calibration sets may be constructed using a plurality of paired measured spectrum and known glucose concentration values. These paired measured spectrum and known glucose concentration values may be measured highly accurate, lab-grade equipment. For example, the known glucose concentration value may be measured on a Yellow Springs Instrument (YSI) model 2300 Plus blood glucose instrument, while the measured spectrum may be measured on a high-resolution Fourier transform infrared spectrometer (FTIR).
In one embodiment, the plurality of paired measured spectrum and glucose concentration values may be processed using a partial least squares (PLS) algorithm. PLS is a statistical method that finds a linear regression model by projecting predicted variables, here analyte concentration, and observable variables, here measured spectrum, to a new space. In one embodiment, the PLS algorithm can be used to produce an analyte vector and one or more spectral loadings. The analyte vector may contain a number of coefficients and, optionally, a bias. The number of coefficients in the analyte vector may be equal to the number of wavelengths in the measured spectrum. For example, if a measured spectrum is configured to contain 24 wavelengths, the analyte vector may contain 24 coefficients and a bias. The PLS algorithm can also be used to create a number of spectral loadings (which may also be known as basis factors or basis vectors).
The number of spectral loadings may be equal to the number of wavelengths in the measured spectrum, or it may be fewer. This is because when embodiments of the PLS algorithm are executed, the PLS algorithm can generate a number of spectral loadings equal to the number of wavelengths in the measured spectrum. The first spectral loading generally spans the largest amount of variance in the spectral data. The second spectral loading generally spans the largest portion of the variance in the spectral data not accounted for by the first spectral loading, and so forth for the remaining spectral loadings. Thus, each of these spectral loadings may be of diminishing significance on the reconstructed spectrum. For example, the first spectral loading will represent a value that may be akin to an average of the measured spectra used to create the calibration set. This spectral loading may be significant in creating reconstructed spectra. Higher order spectral loadings may be affected more and more by noise that may be present in the measurements. Thus, some of the higher order spectral loading may tend to reflect noise in the measurements. Some or all of the higher order spectral loadings may not be statistically significant, and, in some embodiments, some or all of the higher order loadings may be discarded and not used in reconstructing a spectrum from measurements. For example, if the measured spectrum contains 24 wavelengths, a calibration set may have 24 spectral loadings or some number of these spectral loadings may be discarded.
Thus, the calibration set may contain 24 spectral loadings, or may contain only a subset of these, perhaps containing, e.g., 1, 2, 5, 8, 12, 16, or 20 spectral loadings. These spectral loadings may be used to calculate a reconstructed spectrum from the measured spectrum. The number of spectral loadings that are used may depend upon the algorithm used to create calibration sets, the number of paired measured spectra and analyte concentration values used to construct the calibration set, the noise in the measured spectra and analyte concentration values, or any number of other factors. Further details regarding PLS (and other analysis techniques) that are usable with various embodiments of the systems and methods described herein are provided in “Chemometric Techniques for Quantitative Analysis,” by Richard Kramer, published by Marcel Dekker, Inc., New York, 1998, which is hereby incorporated by reference herein in its entirety so as to form part of this specification.
Each calibration set may also contain a prediction eligibility threshold. A prediction eligibility threshold may be used to determine whether a calibration set is eligible to estimate the analyte concentration of a sample. In one embodiment, where the calibration set was created using a PLS algorithm and contains a number of spectral loadings, the prediction eligibility threshold may comprise a scalar value. A prediction eligibility threshold may also be a vector. In one embodiment, a prediction eligibility threshold may be generated by selecting one or more matched measured spectrum and analyte value pairs for which the calibration set is able to provide an accurate estimate of analyte concentration. For each of these measured spectra, the calibration set may be used to create a reconstructed spectrum. A distance metric may then be calculated by comparing the reconstructed spectrum to the measured spectrum. For example, the distance metric may be calculated by calculating a number of spectral residuals between the reconstructed spectrum and the measured spectrum, squaring each of the spectral residuals, and summing those squares to form a sum of the squares of the residuals (SSR) value, for example, as shown in formula (7). The distance metrics for each of the one or more matched measured spectra and analyte value pairs may then be averaged, and an average distance metric may then be used to determine a prediction eligibility threshold. For example, it may be desirable to set the prediction eligibility threshold to some multiple of the average distance metric, such as, e.g., three times the average distance metric or five times the average distance metric. This may allow the prediction eligibility threshold to predict situations where it is likely that a calibration set will be able to provide an accurate estimate of analyte concentration.
Each calibration set may be created using any number of matched spectrum and analyte value pairs. For example, a calibration set may be constructed from one matched spectrum and analyte value pair, or may be constructed from 10, 100, 1000 or more matched spectrum and analyte value pairs. Generally, it may be preferable to construct calibration sets from a large number of matched spectrum and analyte value pairs. This may allow a calibration set to be more accurate, and able to accurately predict analyte values for a greater number and diversity of individuals. Calibration sets may include matched spectrum and analyte value pairs from a multitude of different types of individuals, or may only contain matched spectrum and analyte value pairs from individuals who fall into certain categories. For example, a calibration set may include matched spectrum and analyte value pairs from healthy individuals exclusively, or may include matched spectrum and analyte value pairs from both healthy individuals and individuals in intensive-care units. Calibration sets may also be constructed from matched spectrum and analyte value pairs of individuals according to their demographics (e.g., location, gender, age, race, nationality, etc.), medical conditions (e.g., diabetes, sepsis, etc.), analyte concentration level, prescription or non-prescription drug intake, or other attributes (e.g., presence in the ICU or other type of care facility). For example, calibration sets may be created where the matched spectrum and analyte value pairs all fall within a certain range of analyte concentration level. These calibration sets may be useful for predicting an analyte concentration when the analyte concentration also falls within that certain range. For example, in some implementations configured to measure the analyte glucose, the range of glucose concentrations may generally be between about 0 mg/dL and 600 mg/dL. In some such implementations, multiple calibration sets (e.g., two or more) can be used to cover this range. For example, a first calibration set may cover glucose concentrations in a range from about 0 mg/dL to about 75 mg/dL, a second calibration set may cover glucose concentrations in a range from about 75 mg/dL to about 150 mg/dL, and a third calibration set may cover glucose concentrations in a range from about 150 mg/dL to about 600 mg/dL. In other implementations, two, four, five, six, or more calibration sets may be used. In some implementations, some or all of the analyte concentration ranges for the calibration sets may at least partially overlap with one or more neighboring calibration sets. Some such implementations may advantageously provide redundancy to for the estimation of the analyte concentration.
In one embodiment, additional calibration sets may be constructed by using a first calibration set to estimate analyte concentration values from a plurality of matched spectra and analyte value pairs, and selecting the matched spectra and analyte value pairs for which the first calibration set is least successful at estimating. In this way, the additional calibration set may be specialized to work with measured spectra for which the first calibration set cannot provide an accurate estimate of analyte concentration.
In some implementations, the calibration sets included within the one or more (or plurality) of calibration sets may each be constructed from unique and non-overlapping sets of matched spectrum and analyte value pairs, or the calibration sets may include overlapping sets of matched spectrum and analyte value pairs. For example, a plurality of calibration sets may include a first calibration set constructed from matched spectrum and analyte value pairs of healthy individuals, and a second calibration set constructed from the same matched spectrum and analyte value pairs of healthy individuals as well as matched spectrum and analyte value pairs from individuals with major diseases and illnesses. In this way, calibration sets included within the plurality of calibration sets may be overlapping, as multiple calibration sets may be constructed, at least in part, from the same set of matched spectrum and analyte value pairs.
Determining Whether a Calibration Set is Eligible to Estimate an Analyte Concentration
In the example method illustrated by flowchart 2800, at Block 2830, for each of the plurality of calibration sets, it can be determined whether a calibration set is eligible to estimate an analyte concentration of a sample.
At Block 2910, a calibration set and a measured spectrum are accessed, for example, from a data repository or memory storage. At Block 2920, the calibration set and the measured spectrum are used to calculate a reconstructed spectrum. At Block 2930, the measured spectrum and the reconstructed spectrum are compared in order to calculate a distance metric. At Block 2940, the distance metric is compared to a prediction eligibility threshold for the calibration set. At Block 2950, if the distance metric is lower than the prediction eligibility threshold, the calibration set is eligible to estimate analyte concentration. At Block 2960, if the distance metric is greater than the prediction eligibility threshold, the calibration set is ineligible to estimate analyte concentration. Each step may be done for each calibration set to determine which calibration sets are eligible to estimate analyte concentration.
In this embodiment, each calibration set may include data usable to estimate analyte concentration from a measured spectrum, data usable to construct a reconstructed spectrum, and a prediction eligibility threshold. For example, if calibration sets are created using a PLS algorithm, as above, the data usable to estimate analyte concentration may consist of an analyte vector that consists of a number of coefficients and a bias. In one embodiment, glucose is the analyte of interest, and the measured spectrum contains 24 wavelengths. In this embodiment, the data usable to estimate analyte concentration may consist of a glucose vector that consists of 24 coefficients and a bias. The data sufficient to calculate a reconstructed a spectrum may consist of one of more spectral loadings (or basis factors or basis vectors) created using a PLS algorithm, each of which contains a number of coefficients. For example, the data sufficient to construct a reconstructed spectrum may consist of 12 spectral loadings, each of which contains 24 coefficients. Other numbers of spectral loadings may also be used, as discussed herein. The number of spectral loadings that are used may depend upon the algorithm used to create calibration sets, the number of paired measured spectra and analyte concentration values used to construct the calibration set, or any number of other factors. In one embodiment, the prediction eligibility threshold may consist of a scalar value.
In one embodiment, the calibration set and the measured spectrum can be used to calculate a reconstructed spectrum, as at Block 2920. In some embodiments, the reconstructed spectrum may be created as a linear combination of a number of spectral loadings. For example, if there are 12 spectral loadings containing 24 coefficients per loading, these spectral loadings may be used as a basis set to create the reconstructed spectrum. In some embodiments, the reconstructed spectrum can be calculated in a way that reduces or minimizes a distance metric that measures the difference between the reconstructed spectrum and the measured spectrum.
In this embodiment, the reconstructed spectrum and the measured spectrum may be compared to calculate a distance metric, as at Block 2930. The distance metric may be used to quantify the difference between the reconstructed spectrum and the measured spectrum. The distance metric may be calculated in many different ways. For example, the distance metric may be calculated by using a number of spectral residuals which are the difference between points of the reconstructed spectrum and the measured spectrum. Each of these spectral residuals may be squared, and these squares may be summed to form a sum of the squares of the residuals (SSR) value. Distance metrics may also be calculated in alternative ways, such as by taking absolute values of spectral residuals. It may be preferable to calculate the distance metric the same way as the prediction eligibility threshold is calculated.
In this embodiment, the distance metric may then be compared to the prediction eligibility threshold, to determine whether the distance metric is greater than or less than the prediction eligibility threshold, as at Block 2940. If the distance metric is less than the prediction eligibility threshold, the calibration set may be eligible to estimate analyte concentration, as at Block 2950. If the distance metric is greater than the prediction eligibility threshold, the calibration set may be ineligible to estimate analyte concentration, as at Block 2960.
Determining an Estimated Analyte Concentration from Eligible Calibration Sets
In one embodiment, the algorithm processor 416 may determine an estimated analyte concentration of the sample based at least on the eligible calibration sets, as schematically illustrated in Block 2840. As described above, each calibration set may contain data usable to estimate an analyte concentration of a sample. In one embodiment, a calibration may estimate the analyte concentration of the sample using an analyte vector containing a number of coefficients and a bias. In one embodiment, where the analyte of interest is glucose, the following equation may be used to estimate a glucose concentration using a glucose vector:
where gm is a glucose concentration estimate created by an mth calibration set, Cs(λi) is a measured absorbance value at the ith wavelength of n total wavelengths, κmi is an ith coefficient of the glucose vector for the mth calibration set, and βm is the bias for the glucose vector of the mth calibration set.
If only one calibration set is eligible to estimate the analyte concentration, the estimated analyte concentration may consist of an estimate of analyte concentration from the one eligible calibration set.
However, if more than one calibration set is determined to be eligible to estimate analyte concentration of a sample, an estimate of analyte concentration from one or more of the more than one eligible calibration set may be used to calculate an estimated analyte concentration. For example, estimates from each eligible calibration set may be used to calculate an estimated analyte concentration. The estimates of analyte concentration from each of the more than one eligible calibration set may be combined in a number of different ways. Some or all of the estimates of analyte concentration from each of the more than one eligible calibration set may be averaged to create the estimated analyte concentration. For example, some or all of the estimates of analyte concentration from each of the more than one eligible calibration set may be averaged using an arithmetic mean, weighted mean, or other mathematical technique.
For example, the estimated analyte concentration may be calculated using only a portion of the eligible calibration sets. If the calibration sets use distance metrics and prediction eligibility thresholds, as discussed herein, these values may be compared to determine which of the calibration sets is closest to or furthest from the measured spectrum. The algorithm may selectively discard some number of the furthest calibration sets. For example, the algorithm may discard the furthest one, two, three, five, or more calibration sets. Alternatively, the algorithm may selectively use some number of the closest calibration sets. For example, the algorithm may use the closest one, two, three, five, or more calibration sets.
Alternatively, if there are multiple calibration sets eligible to estimate analyte concentration, an estimate of analyte concentration from each of the multiple eligible calibration sets may be weighted in order to give different weight to certain of the estimates of analyte concentration. In one embodiment, where the analyte of interest is glucose, the following equation may be used to estimate a glucose concentration of a sample:
where gest is an estimated glucose concentration of a sample, wm is a weight given to an estimate of glucose concentration given by an mth eligible calibration set, and gm is the estimate of glucose concentration given by an mth eligible calibration set, and there are Ne eligible calibration sets. In this example, Ne may satisfy the equation:
1≦Ne≦Ne (10)
where there are Ne eligible calibration sets, at least one of those calibration sets is eligible, and where there are Ne total calibration sets. In some embodiments, the weights wm can be normalized, for example, such that
In one embodiment, where a calibration set contains a prediction eligibility threshold and a distance metric may be calculated, the calibration set may be given a weight based at least in part on a difference or ratio between the distance metric and the prediction eligibility threshold. For example, if there are two eligible calibration sets, and a first calibration set has a distance metric very similar to the prediction eligibility threshold and a second calibration set has a distance metric much less than that the prediction eligibility threshold, then in one embodiment, more weight may be given to the second calibration set. It may be found that according varying levels of weight to the estimates of calibration sets based upon the difference between the distance metric and the prediction eligibility threshold of that calibration set will increase the accuracy of the estimated analyte concentration.
In a further embodiment, a weight may be given to a calibration set based at least in part on an estimate of analyte concentration of the calibration set. For example, a calibration set may be accorded more weight if it provides a lower estimate of analyte concentration, or if it provides an estimate of analyte concentration above or below a certain level. For example, where the analyte of interest is glucose, it may be undesirable to provide an estimate of glucose concentration that is above a certain level if the actual level of glucose concentration is below that level, at discussed herein. In this embodiment, it may be useful to recognize hypoglycemia, which may occur when a blood plasma glucose level is below about 70 mg/dL. In one embodiment, whenever a calibration set would provide an estimate of glucose concentration below a certain level, the algorithm may give the calibration set a higher weight, in order to reduce or mitigate the risk of an over-estimate of glucose concentration when actual glucose concentration is below a certain level. For example, a calibration set may be accorded a higher weight if the calibration set provides an estimate glucose concentration below a threshold level (e.g. 80, 100, or 120 mg/dL), in order to mitigate the risk of an over-estimate below a second threshold level (e.g. 100, 120, or 140 mg/dL).
In a further embodiment, a weight may be given to each eligible calibration set based upon an estimate of the accuracy of each calibration set. For example, if there are two calibration sets, it may be calculated that a first calibration set is more likely to be accurate than a second calibration set, and therefore, the first calibration set may be given a higher weight than the second calibration set in the case where both calibration sets are eligible to estimate analyte concentration.
In one embodiment, if none of a plurality of calibration sets is eligible to estimate analyte concentration of a sample, the algorithm will not determine an estimated analyte concentration. For example, the system may take an appropriate action, as discussed above, rather than determining an estimated analyte concentration. Any of these weighting schemes can be used alone, together, or in any suitable combination. In other embodiments, other weighting schemes may also be used.
In this illustrative example, at measurement time 1, the distance metric for calibration set A is above the prediction eligibility threshold for calibration set A 3015. This means that at measurement time 1, calibration set A is not eligible to estimate analyte concentration. Also at measurement time 1, the distance metric for calibration set B is below the eligibility threshold for calibration set A 3025. This means that at measurement time 1, calibration set B is eligible to estimate analyte concentration. Therefore, at measurement time 1, the estimated analyte concentration would be based upon an estimated analyte concentration from calibration set B but not calibration set A. This is also true of measurement times 2, 3 and 4, as the distance metric for calibration set A remains above the prediction eligibility threshold for calibration set A 3015, while the distance metric for calibration set B remains below the eligibility threshold for calibration set A 3025.
Continuing with this illustrative example, at measurement times 5, 6, 7 and 8, the distance metric for calibration set A is below the prediction eligibility threshold for calibration set A 3015, while the distance metric for calibration set B is above the prediction eligibility threshold for calibration set B 3025. Therefore, at measurement times 5, 6, 7 and 8, the estimated analyte concentration may be based upon an estimated analyte concentration from calibration set A but not calibration set B.
Continuing with this illustrative example, at measurement time 9, the distance metric for calibration set A is above the prediction eligibility threshold for calibration set A 3015, while the distance metric for calibration set B is above the prediction eligibility threshold for calibration set B 3025. Therefore, at measurement time 9, neither calibration set A nor calibration set B are eligible to provide an estimate for analyte concentration. Therefore, if calibration set A and B are the only two calibration sets being used, the algorithm may not be able to estimate analyte concentration and an appropriate action may be taken. In one embodiment, the algorithm may be attached to a display system 416, which may be programmed to display an appropriate message. The display system 416 may show that a reading (e.g. an estimate of analyte concentration in a sample) was not generated by displaying a null value (e.g. “---”) instead of an analyte concentration level, may turn on a warning light, or may take other appropriate action, as discussed above. The algorithm may also be attached to a system capable of notifying the operator of the system that the algorithm has returned a “No Read,” or the system may request another measurement of the sample be taken in order to obtain an estimate of analyte concentration.
Continuing with this illustrative example, at measurement times 10 and 11, the distance metric for calibration set A is below the prediction eligibility threshold for calibration set A 3015, and the distance metric for calibration set B is also below the prediction eligibility threshold for calibration set B 3025. Therefore, at measurement times 10 and 11, the estimated analyte concentration can be based upon an estimated analyte concentration from calibration set A and an estimated analyte concentration from calibration set B. For example, the estimate of analyte concentration may be determined by the formula:
g
est
=w
A
g
A
+w
B
g
B (11)
where gest is an estimated analyte concentration of a sample, wA is a weight given to an estimate of analyte concentration given by calibration set A, wB is a weight given to an estimate of analyte concentration given by calibration set B, gA is the estimate of analyte concentration given by calibration set A, and gB is the estimate of analyte concentration given by calibration set B. The values of wA and wB may be determined in a number of different ways, as discussed herein. In some embodiments, the weights may be normalized such that their sum is equal to one (e.g., wA+wB=1, in this illustrative example).
The graph 3100 in
As described herein, an estimated glucose concentration may be calculated, at least in part, by using the glucose concentration values predicted by some or all of the eligible calibration sets. One method that may be used to determine whether a calibration set is eligible is to compare that distance metric for that calibration set with a prediction eligibility threshold for that calibration set. For example, in the graph 3100 at 9:30 AM, the distance metric for Calibration Set A 3125 may be compared to the prediction eligibility threshold for Calibration Set A 3120. It may be observed that the prediction eligibility threshold for Calibration Set A 3125 is approximately 3×10−6. However, at 9:30 AM, the distance metric for Calibration Set A 3125 is approximately 2×10−7. Because 2×10−7 is less than 3×10−6, this means that Calibration Set A is eligible to estimate the glucose concentration of the sample at 9:30 AM. The prediction eligibility threshold for Calibration Set B 3115 as illustrated in the graph 3100 is approximately 2×10−6. At 9:30 AM, the distance metric for Calibration Set B 3130 is approximately 2.5×10−6. Because 2.5×10−6 is greater than 2×10−6, Calibration Set B is not eligible to provide an estimate of glucose concentration at 9:30 AM. Therefore, the estimated glucose concentration may be based, at least in part, on the glucose concentration estimated by Calibration Set A, but not based upon the estimate of glucose concentration provided by Calibration Set B. In other implementations, if the distance metric for a calibration set exceeds the prediction eligibility threshold, the calibration set may nonetheless be included in the analyte estimation calculation but, for example, with a much reduced weight.
Continuing with the graph 3100 in
Continuing with the graph 3100 in
At Block 3210, a calibration set from the plurality of calibration sets can be used to create a reconstructed spectrum from the measured spectrum. At Block 3215, a distance metric can be calculated by comparing the reconstructed spectrum to the measured spectrum. At Block 3220, a prediction eligibility threshold for the calibration set can be accessed. At Block 3225, the distance metric may be compared to the prediction eligibility threshold to determine whether the distance metric is less than the prediction eligibility threshold. If the distance metric is less than the prediction eligibility threshold, at Block 3230, an estimate of analyte concentration may be calculated based upon the measured spectrum and the calibration set. In one embodiment, this may be done using an analyte vector, as described above. At Block 3235 a weight for the calibration set can be determined, based upon the distance metric and the prediction eligibility threshold for the calibration set. If the distance metric is less than the prediction eligibility threshold at Block 3225, the calibration set may be ineligible to provide an estimate of analyte concentration at Block 3240. After either determining a calibration set is ineligible or calculating a weight for an eligible calibration set, at Block 3245 it can be determined whether there are more calibration sets left to analyze the measured spectrum. If there are more calibration sets, Block 3210 and the following blocks may be repeated for each calibration set. If there are no more calibration sets, at Block 3250, it may be determined whether all calibration sets were ineligible. If there are no eligible calibration sets, at Block 3255, a “No Read” can be returned. This may result in an appropriate action being taken, as discussed above. If one or more calibration sets are eligible, the weights for the eligible calibration sets, calculated at Block 3235, may be normalized at Block 3260. at Block 3265, an estimated analyte concentration may be calculated based upon the normalized weights calculated at Block 3260 and the estimates of analyte concentration calculated at Block 3230. At Block 3270, the estimated analyte concentration may be returned.
Multi-Step Method of Estimating Analyte Concentration
One example multi-step method for estimating an analyte concentration of a blood sample is described herein. First, a blood sample S may be obtained. This blood sample may be physically treated in a number of ways. A measurement of blood sample S may be made, such as a spectroscopic measurement of blood sample S. The measurement of blood sample S may then be pre-processed to reduce or minimize potentially negative effects of certain potential interferents on the accuracy of an estimate of analyte concentration. This pre-processed measurement of blood sample S may be used to provide an estimate of analyte concentration in blood sample S. This accuracy of this estimate may be compared to a threshold value, in order to determine whether the method should return the estimate of analyte concentration to a user, or whether the method should return a “No Read” signal.
A blood sample S may be obtained from a patient in order to determine the concentration of one or more analytes within blood sample S. Blood sample S may be whole blood, plasma, or blood that has been processed, refined, diluted, or treated in any number of ways. For example, blood sample S may be whole blood that has been drawn from a patient and centrifuged in order to fractionate the blood, separating red blood cells from plasma. Blood sample S may also be filtered in a number of ways, in order to remove certain portions of the blood or remove impurities from the blood. Other forms of processing, refining, diluting, or treating the blood may also be used, either alone or in conjunction with filtering and/or centrifuging the blood.
Next, a measurement of blood sample S may be obtained and pre-processed to help reduce or minimize the effects of potential interferents. For example, in some embodiments of the method 2800 described with reference to
The first stage of pre-processing may comprise pre-processing to reduce or minimize the negative effect of certain common potential interferents. These common potential interferents may be present in many or most blood samples, and may negatively affect the accuracy of an estimate of analyte concentration in blood sample S if they are not pre-processed. Common interferents include, but are not limited to, water and saline. Other common interferents can include substances included in the Library of Interferents described herein. For example, the common interferents can include substances in the table included above and described with reference to Example Experiment 2. Water or saline may be present in many or most blood samples, and may negatively affect the accuracy of an estimate of analyte concentration in blood sample S. Thus, it may be beneficial to pre-process the measurement of blood sample S in order to help reduce of mitigate the effects of water or saline on the accuracy of an estimate of analyte concentration. There may also be a number of other common potential interferents which may be pre-processed out of the measurement of blood sample S. For example, it may be beneficial to pre-process certain proteins or protein generally, certain common drugs, or other common sources of error in estimates of analyte concentration. Any number of common potential interferents may be pre-processed. For example, one, two, three, four, five, or ten or more common potential interferents may be pre-processed in order to reduce or mitigate their negative effect on the estimate of analyte concentration. These common potential interferents may be pre-processed individually in an iterative manner, or may be pre-processed together in a single step (e.g., in a parallel manner). If the measurement of blood sample S is a spectroscopic measurement, pre-processing may comprise subtracting or removing a value from the spectroscopic measurement of blood sample S, where the value is based at least in part upon a known spectrum of the potential interferent or interferents which may be subject to pre-processing. For example, if the potential interferent is saline, pre-processing the measurement of blood sample S may comprise removing the known spectra of saline from a spectroscopic measurement of blood sample S. In other implementations, saline may be introduced into a spectroscopic sample cell and a saline measurement taken by the analyte detection system. The sample measurement may be processed to remove the effects of saline by subtracting the saline measurement from the sample measurement. Such processing may also remove effects caused by the sample cell. It may also be advantageous to allow a user to specify which common potential interferents should be pre-processed. For example, if a user is a doctor and has administered a drug to a patient, the doctor may realize that the patient's blood will contain that drug. It may be beneficial to provide the doctor with the ability to tell the method which drugs are likely to be in a patient's blood, so they may be pre-processed as necessary. It may also be beneficial if the system has access to a patient's medical records, in order to determine which potential interferents may be in a patient's blood. Examples of this type of spectroscopic pre-processing to account for the presence of certain interferents are described in more detail in U.S. Pat. Nos. 7,009,180 and 7,271,912, each of which is incorporated by reference in its entirety herein for all it discloses.
In a second stage of pre-processing, the measurement of blood sample S may be pre-processed to reduce or minimize the effects of other potential interferents that may negatively affect the estimate of analyte concentration. This second stage of pre-processing may be done in a similar manner to the first stage described above, or may be done in a different manner. It may be beneficial to pre-process the measurement of blood sample S for common potential interferents such as saline (or water), and to then determine whether the measurement of blood sample S requires additional pre-processing to account for other potential interferents which may be present in blood sample S. This may be advantageous if, for example, these other potential interferents are less commonly found, and if pre-processing for these potential interferents when they are not present may introduce error into a later estimate of analyte concentration, or if pre-processing for all potential interferents may be too time-consuming or use too many limited computing resources. Interferents processed in the second stage can include substances included in the Library of Interferents described herein. For example, such interferents can include substances in the table included above and described with reference to Example Experiment 2. The interferents processed in the second stage generally include substances not corrected for in the first stage of pre-processing.
Generally, it may be found that the removal of a potential interferent from a measurement of a sample increases the accuracy of an estimate of analyte concentration in that sample if the potential interferent is present in sufficient quantity, but may reduce the accuracy of the estimate if the potential interferent is not present in a sufficient quantity. It may also be found that each pre-processing step may introduce a small amount of noise or error into an estimate of analyte concentration. This small amount of noise or error may compound as more pre-processing steps are taken. Thus, it may be beneficial to limit the number of pre-processing steps, in order to minimize the introduction of noise and/or error, and to increase the accuracy of an estimate of analyte concentration of a sample. For example, it may be found that an estimate of analyte concentration is more accurate if fewer than 3, 5, 9, 13, 17, 23 or some other number of pre-processing steps are conducted on a measurement of a sample. It may also be found that an estimate of analyte concentration is more accurate if potential interferents are only pre-processed if they are present in a sufficient quantity within the sample.
Because an estimate of analyte concentration may be more accurate if fewer unnecessary pre-processing steps are taken, it may be advantageous to determine which potential interferents should be pre-processed from the measurement of blood sample S. This may also be advantageous if there is a maximum number of pre-processing steps that may be taken, but there is a larger number of potential interferents. For example, if it was determined that an estimate of analyte concentration is less likely to be accurate if more than 9 pre-processing steps are conducted, but there are 15, 30, 50, or 100 potential interferents, it may be beneficial to determine which of these potential interferents, if any, should be pre-processed out of the measurement of blood sample S.
There are many ways in which it may be determined which potential interferents to pre-process out of a measurement of a blood sample. For example, it may be beneficial to determine whether certain potential interferents are present in a blood sample S. It may also be beneficial to determine whether there certain potential interferents are present in a quantity above some threshold level. For example, if a measurement of blood sample S is a spectroscopic measurement, this measurement may be compared to the known spectra of certain potential interferents to determine whether or not these potential interferents may be present in the sample. This comparison may also be used to estimate whether or not the quantity of a certain potential interferent is above a threshold level. If the quantity of a potential interferent is above a threshold level, that particular potential interferent may be pre-processed in order to reduce of minimize its effect on the estimate of analyte concentration. These threshold levels may be particular to each potential interferent or to each type of potential interferent. There may be a list of potential interferents to process out of a pre-processed measurement. This list may include some or all interferents in the Library of Interferents. This list may be in a certain order, such that it is determined whether to process out certain interferents before other interferents. This order may be beneficial if only a limited number of processing steps may occur, as this order may allow certain higher-priority potential interferents to be processed out of a pre-processed measurement of a blood sample before lower-priority potential interferents. This order may be determined in a number of ways. For example, it may be found that certain interferents have a greater negative effect on the accuracy of an estimate of analyte concentration if they are not processed out. It may be found that certain interferents may be more effectively processed out of a measurement of a blood sample, so it may be beneficial to determine the presence of these interferents first. It may be found that certain interferents are more common, or may be more common in certain locations or among certain groups of patients. For example, it may be found that a certain treatment center is more likely to administer certain drugs which may interfere with an estimate of analyte concentration, or that people with certain conditions are more likely to have certain types of interferents in their blood samples. These factors and others may help to determine an order that potential interferents may be processed out. It may also be beneficial to determine the potential presence or absence of multiple different potential interferents, and to determine the possible effects each of these interferents may have on the estimate of analyte concentration, in order to process out the interferents that may have the largest negative effect. For example, it may be possible to determine an approximate level of a number of potential interferents, and use a database of interferent data to determine the effect that each of these potential interferents may have on the accuracy of an estimate of analyte concentration. It may also be beneficial to provide the system with access to a patient's medical records, or to allow a user to alert the method to the potential presence of certain drugs or other compounds in a patient's blood stream, and to create a list of potential interferents based, at least in part, on this data.
After the measurement of the blood sample S has been pre-processed, this measurement may be used to provide an estimate of the analyte concentration in blood sample S. This estimate may be made using a number of methods. For example, this estimate may be made using any of the methods described herein. In some embodiments, it may be beneficial to use multiple calibration sets, or it may be beneficial to use only a single calibration set. For example, multiple calibration sets may be used to estimate analyte concentration, and their estimates may be combined as described herein. Alternatively, it may be found that a single calibration set may be more accurate than using multiple calibration sets, and the use of multiple calibration sets may therefore decrease accuracy, since this may result in combining a single, more-accurate estimate with other less-accurate estimates. Calibration sets may be constructed in a number of ways, using data from a number of matched pair samples, as detailed herein. It may be beneficial to construct a calibration set using a large number of matched spectrum and analyte value pairs. These matched spectrum and analyte value pairs may come from a variety of individuals, including healthy individuals and also individuals who may be receiving medical treatment for a wide variety of ailments. Alternatively, it may be beneficial to construct a calibration set using only a small number of individuals. This small number of individuals may be comprised only of healthy individuals. It may be found that using a small number of matched spectrum and analyte value pairs, taken from healthy individuals, may provide better and more accurate estimates of analyte concentrations. A calibration set based on only a small number of individuals may be more accurate if it is found that a large calibration set tends to introduce more data into the calibration set that may appear statistically significant, but really represents noise. A calibration set based only on matched spectrum and analyte value pairs from healthy individuals may provide more accurate estimates of analyte concentrations if it is found that spectra from unhealthy individuals are simply too varied, case-specific, and unique to provide accurate estimates of analyte concentration in others. The use of a smaller calibration set may result in a larger number of instances of a “No Read” situations occurring, as described herein. However, this larger number of “No Read” situations may be advantageous, as this may increase the accuracy of the estimates of analyte concentration in the instances where the calibration set is eligible to provide an estimate of analyte concentration and may result in fewer inaccurate estimates of analyte concentration.
Reference throughout this specification to “some embodiments” or “an embodiment” means that a particular feature, structure or characteristic described in connection with the embodiment is included in at least some embodiments. Thus, appearances of the phrases “in some embodiments” or “in an embodiment” in various places throughout this specification are not necessarily all referring to the same embodiment and may refer to one or more of the same or different embodiments. Furthermore, the particular features, structures or characteristics may be combined in any suitable manner, as would be apparent to one of ordinary skill in the art from this disclosure, in one or more embodiments.
As used in this application, the terms “comprising,” “including,” “having,” and the like are synonymous and are used inclusively, in an open-ended fashion, and do not exclude additional elements, features, acts, operations, and so forth. Also, the term “or” is used in its inclusive sense (and not in its exclusive sense) so that when used, for example, to connect a list of elements, the term “or” means one, some, or all of the elements in the list.
Similarly, it should be appreciated that in the above description of embodiments, various features are sometimes grouped together in a single embodiment, figure, or description thereof for the purpose of streamlining the disclosure and aiding in the understanding of one or more of the various inventive aspects. This method of disclosure, however, is not to be interpreted as reflecting an intention that any claim require more features than are expressly recited in that claim. Rather, inventive aspects lie in a combination of fewer than all features of any single foregoing disclosed embodiment. Accordingly, no feature or group of features is necessary or indispensable to each embodiment.
Embodiments of the disclosed systems and methods may be used and/or implemented with local and/or remote devices, components, and/or modules. The term “remote” may include devices, components, and/or modules not stored locally, for example, not accessible via a local bus. Thus, a remote device may include a device which is physically located in the same room and connected via a device such as a switch or a local area network. In other situations, a remote device may also be located in a separate geographic area, such as, for example, in a different location, building, city, country, and so forth.
Methods and processes described herein may be embodied in, and partially or fully automated via, software code modules executed by one or more general and/or special purpose physical computing systems. The word “module” refers to logic embodied in hardware and/or firmware, or to a collection of software instructions, possibly having entry and exit points, written in a programming language, such as, for example, C or C++. A software module may be compiled and linked into an executable program, installed in a dynamically linked library, or may be written in an interpreted programming language such as, for example, BASIC, Perl, or Python. It will be appreciated that software modules may be callable from other modules or from themselves, and/or may be invoked in response to detected events or interrupts. Software instructions may be embedded in firmware, such as an erasable programmable read-only memory (EPROM). It will be further appreciated that hardware modules may be comprised of connected logic units, such as gates and flip-flops, and/or may be comprised of programmable units, such as programmable gate arrays, application specific integrated circuits, and/or processors. The modules described herein are preferably implemented as software modules, but may be represented in hardware and/or firmware. Moreover, although in some embodiments a module may be separately compiled, in other embodiments a module may represent a subset of instructions of a separately compiled program, and may not have an interface available to other logical program units.
In certain embodiments, code modules may be implemented and/or stored in any type of computer-readable medium or other computer storage device (e.g., hard disks, RAM, ROM, flash memory, etc.). Computer-readable media include non-transitory computer-readable media such as magnetic storage, optical storage (e.g., CD-ROMs or DVDs), semiconductor storage, etc. In some systems, data (and/or metadata) input to the system, data generated by the system, and/or data used by the system can be stored in any type of computer data repository, such as a relational database and/or flat file system. Any of the systems, methods, and processes described herein may include an interface configured to permit interaction with patients, health care practitioners, administrators, other systems, components, programs, and so forth.
A number of applications, publications, and external documents may be incorporated by reference herein. Any conflict or contradiction between a statement in the body text of this specification and a statement in any of the incorporated documents is to be resolved in favor of the statement in the body text.
Although described in the illustrative context of certain preferred embodiments and examples, it will be understood by those skilled in the art that the disclosure extends beyond the specifically described embodiments to other alternative embodiments and/or uses and obvious modifications and equivalents. Thus, it is intended that the scope of the claims which follow should not be limited by the particular embodiments described above.
This application claims benefit under 35 U.S.C. §119(e) of U.S. Provisional Patent Application No. 61/509,487, titled “Method and Apparatus for Analyte Measurements Using Multiple Calibration Sets,” filed on Jul. 19, 2011. The entire disclosure of the above-identified application is incorporated by reference herein and made part of this specification.
Number | Date | Country | |
---|---|---|---|
61509487 | Jul 2011 | US |