Sensitivity calibration of in vivo sensors used to measure analyte concentration

Abstract

Methods, devices, systems, and computer program products are provided to improve sensitivity calibration of an in vivo analyte sensor and usability of an associated analyte monitoring system. In certain embodiments, methods are provided that improve the user experience of using an analyte monitoring system. Certain embodiments of the present disclosure include features that reduce the amount of calibration or re-calibration performed by the analyte monitoring system. More specifically methods of using a suspect calibration attempt to avoid having to recalibrate by adjusting the calibration or mitigating effects of sensor signal attenuation that caused the calibration attempt to be suspect are provided. Additional features are disclosed.

Description

BACKGROUND

The detection of the concentration level of glucose or other analytes in certain individuals may be vitally important to their health. For example, the monitoring of glucose levels is particularly important to individuals with diabetes or pre-diabetes. People with diabetes may need to monitor their glucose levels to determine when medication (e.g., insulin) is needed to reduce their glucose levels or when additional glucose is needed.

Devices have been developed for automated in vivo monitoring of analyte concentrations, such as glucose levels, in bodily fluids such as in the blood stream or in interstitial fluid. Some of these analyte level measuring devices are configured so that at least a portion of the devices are positioned below a skin surface of a user, e.g., in a blood vessel or in the subcutaneous tissue of a user. As used herein, the term analyte monitoring system is used to refer to any type of in vivo monitoring system that uses a sensor disposed with at least a portion subcutaneously to measure and store sensor data representative of analyte concentration levels automatically over time. Analyte monitoring systems include both (1) systems such as continuous glucose monitors (CGMs) which transmit sensor data continuously or at regular time intervals (e.g., once per minute) to a processor/display unit and (2) systems that transfer stored sensor data in one or more batches in response to a request from a processor/display unit (e.g., based on an activation action and/or proximity, for example, using a near field communications protocol) or at a predetermined but irregular time interval.

Determining an analyte concentration level in blood based on the analyte concentration measured using an analyte monitoring system typically involves calibrating the in vivo sensor of the analyte monitoring system using a reference measurement. For example, a finger stick blood sample may be used as a reference to determine the blood analyte concentration at a particular time and the result is paired with a corresponding measurement from the analyte monitoring system. The sensitivity of the analyte monitoring system's sensor is adjusted based on the difference and other factors. However, several factors related to the measurement from the analyte monitoring system can affect the accuracy of the calibration. Thus, what is needed are systems, methods and apparatus to improve sensitivity calibration of the in vivo sensors used to measure analyte concentration.

SUMMARY

Methods, devices, and systems are provided to improve sensitivity calibration of an in vivo analyte sensor and usability of the associated analyte monitoring system. In some embodiments, the present disclosure includes a computer-implemented method for defining a set of system checks associated with an analyte monitoring system; receiving a request to perform calibration of an in vivo sensor of an analyte monitoring system; receiving a signal representative of sensor data from the analyte monitoring system related to an analyte level of a patient measured over time; conducting calibration using a reference measurement estimated at calibration time from a first predetermined duration of reference observation paired, with a signal representative of sensor data up to the calibration time; updating calibration using the same reference measurement at a second predetermined duration of reference observation spanning the first predetermined duration and an additional period, paired with a signal representative of sensor data up to a third predetermined period relative to the calibration time; using an adjustment map that balances the risk of over and under calibration based on a priori information; deferring calibration if attenuation of a signal from the sensor is detectable at a time in which the request is received, wherein the calibration is delayed until the signal attenuation is no longer detected; performing calibration if the set of system checks pass and signal attenuation is not being detected; storing the signal from the sensor over time for a period of time spanning from before the calibration to after the calibration; determining if previously undetected signal attenuation occurred during the calibration based on the stored signal after the calibration has completed; reconstructing the signal and revising the calibration of the sensor based on the reconstructed signal if previously undetected signal attenuation occurred and if the signal can be adjusted to compensate for the previously undetected signal attenuation; invalidating the calibration and requesting a new calibration if previously undetected signal attenuation occurred and if the signal cannot be adjusted to compensate for the previously undetected signal attenuation; and displaying an analyte concentration value if previously undetected signal attenuation did not occur during calibration based on the stored signal. Related systems and computer program products are also disclosed. Numerous other aspects and embodiments are provided. Other features and aspects of the present disclosure will become more fully apparent from the following detailed description, the appended claims, and the accompanying drawings.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 depicts a flowchart illustrating an example of a first method for calibrating a sensor of an analyte monitoring system in accordance with some embodiments of the present disclosure.

FIG. 2 depicts a flowchart illustrating an example of a second method for calibrating a sensor of an analyte monitoring system in accordance with some embodiments of the present disclosure.

FIG. 3 depicts a flowchart illustrating an example of a third method for calibrating a sensor of an analyte monitoring system in accordance with some embodiments of the present disclosure.

FIG. 4 depicts a flowchart illustrating an example of a fourth method for calibrating a sensor of an analyte monitoring system in accordance with some embodiments of the present disclosure.

FIG. 5 depicts a graph of example data on a Clarke error grid in accordance with some embodiments of the present disclosure.

FIG. 6 depicts a graph of example data on a Clarke error grid in accordance with some embodiments of the present disclosure.

FIG. 7 depicts a graph summarizing the effect of stratifying the example data in FIG. 5 in accordance with some embodiments of the present disclosure.

FIG. 8 depicts a graph of an example correlation between an accuracy metric and the ratio of calibration sensitivity (Sc) to ideal sensitivity (Sn) in accordance with some embodiments of the present disclosure.

FIG. 9 depicts a graph of an example of a calibration adjustment factor derived by inverting the relative accuracy metric of FIG. 8 in accordance with some embodiments of the present disclosure.

FIG. 10 depicts a flowchart illustrating an example of a fifth method for calibrating a sensor of an analyte monitoring system in accordance with some embodiments of the present disclosure.

FIG. 11 depicts a flowchart illustrating an example of a sixth method for calibrating a sensor of an analyte monitoring system in accordance with some embodiments of the present disclosure.

FIG. 12 depicts a flowchart illustrating an example of a seventh method for calibrating a sensor of an analyte monitoring system in accordance with some embodiments of the present disclosure.

DETAILED DESCRIPTION

The present disclosure provides systems, methods, and apparatus to improve sensitivity calibration of an in vivo analyte sensor and usability of the associated analyte monitoring system. In some embodiments, the present disclosure provides methods that improve the user experience of using an analyte monitoring system. For example, in some embodiments, the present disclosure includes features that reduce the amount of calibration or re-calibration performed by the analyte monitoring system. More specifically, in some embodiments, the present disclosure provides methods of using a suspect calibration attempt to avoid having to recalibrate by adjusting the calibration or mitigating effects of sensor signal attenuation that caused the calibration attempt to be suspect. For example, sensor signal attenuating phenomena such as Early Sensitivity Attenuation (ESA) and dropouts that occur during a calibration attempt (and render the calibration attempt suspect) can be addressed in some embodiments. When ESA or a dropout occurs, the sensor still generates a signal with a lower amplitude than the true signal, but may still take on a value that is physiologically feasible. Hence, unlike signal loss, detection of either ESA or a dropout is not trivial. Unlike typical noise artifact, ESA and dropouts consistently bias the sensor output down. Standard filtering methods are not effective to compensate for neither ESA nor dropouts without incurring additional drawbacks. ESA typically occurs within the first few hours of sensor start time, and disappears thereafter. Dropouts last for much shorter periods and might occur, for example, during bedtime independent of the sensor start time.

In some embodiments, the present disclosure provides methods of analyzing sensor data received after a reference analyte concentration value is received (e.g., initiated by the user) to perform a calibration in place of a future system requested calibration. Thus, for example, if a user happens to perform a finger stick in vitro blood glucose (BG) measurement and the outcome is provided to the analyte monitoring system, the system can use the information as a reference analyte concentration value for a calibration attempt. In conventional BG systems, a reference analyte BG reading is computed based on a time series measurement in response to an analyte sample collected in a test strip. The BG system takes a predetermined duration in order to collect the time series measurement, typically in the five second range. In addition, instead of determining a sensitivity of the in vivo sensor using a real time algorithm that only considers limited sensor data in order to provide real time responsiveness regarding the success or failure of the calibration attempt, the present disclosure can take the time to consider a wider window of sensor data since the user is not waiting for an outcome indication of the calibration attempt. Therefore, not only are the number of system requested calibrations reduced, more accurate calibrations can be performed by using more data. In some embodiments, the present disclosure provides methods that improve the sensitivity calibration by using historical data to hedge against the aggregate risks of calibration errors, such as, for example, due to ESA. In some embodiments, the present disclosure uses “extra” data obtained from a test strip that remains in a BG meter after an initial reading has been made, to improve the accuracy of the initial reading and to improve the certainty of the analyte monitoring system's calibration based on the initial reading. In some embodiments, the present disclosure uses fault identification, instead of merely detecting faults, to improve the likelihood of a successful calibration attempt.

Embodiments of the present disclosure are described primarily with respect to continuous glucose monitoring devices and systems for illustrative purposes but the present disclosure can be applied to other analytes, other analyte characteristics, and other analyte measurement systems, as well as data from measurement systems that transmit sensor data from a sensor unit to another unit such as a processing or display unit in response a request from the other unit. For example, other analytes that can be monitored include, but are not limited to, acetyl choline, amylase, bilirubin, cholesterol, chorionic gonadotropin, creatine kinase (e.g., CK-MB), creatine, DNA, fructosamine, glutamine, growth hormones, hormones, ketones, lactate, peroxide, prostate-specific antigen, prothrombin, RNA, thyroid stimulating hormone, and troponin. The concentration of drugs, such as, for example, antibiotics (e.g., gentamicin, vancomycin, and the like), digitoxin, digoxin, drugs of abuse, theophylline, and warfarin, can also be monitored. In those embodiments that monitor more than one analyte, the analytes can be monitored at the same or different times. In addition, in some embodiments, the present disclosure can be applied to non-analyte sensor data. For example, non-analyte sensor data can include temperature estimation of a target physiological compartment that is made based on measuring the temperature of a nearby compartment, where the measured temperature is related to the temperature of the target compartment. The present disclosure also provides numerous additional embodiments.

Some embodiments of the present disclosure include a programmed computer system adapted to receive and store data from an analyte monitoring system. The computer system can include one or more processors for executing instructions or programs that implement the methods described herein. The computer system can include memory and persistent storage devices to store and manipulate the instructions and sensor data received from the analyte monitoring system. The computer system can also include communications facilities (e.g., wireless and/or wired) to enable transfer of the sensor data from the analyte monitoring system to the computer. The computer system can include a display and/or output devices for identifying dropouts in the sensor data to a user. The computer system can include input devices and various other components (e.g., power supply, operating system, clock, etc.) that are typically found in a conventional computer system. In some embodiments, the computer system is integral to the analyte monitoring system. For example, the computer system can be embodied as a handheld or portable receiver unit within the analyte monitoring system. In some embodiments, the analyte monitoring system may include a computer system and in some embodiments, the analyte monitoring system may include (or be in signal communication with) an analyte meter (e.g., a BG meter) configured to measure an analyte concentration in vitro.

In some embodiments, the various methods described herein for performing one or more processes, also described herein, can be embodied as computer programs (e.g., computer executable instructions and data structures). These programs can be developed using an object oriented programming language, for example, that allows the modeling of complex systems with modular objects to create abstractions that are representative of real world, physical objects and their interrelationships. However, any practicable programming language and/or techniques can be used. The software for performing the inventive processes, which can be stored in a memory or storage device of the computer system described herein, can be developed by a person of ordinary skill in the art based upon the present disclosure and can include one or more computer program products. The computer program products can be stored on a non-transitory computer readable medium such as a server memory, a computer network, the Internet, and/or a computer storage device.

Analyte monitoring systems such as Continuous Glucose Monitor (CGM) systems that use reference Blood Glucose (BG) values for calibration, pair the reference BG values with a corresponding sensor data point received in an uncalibrated sensor signal from the CGM system. There are a number of conditions which can invalidate a calibration attempt due to known hardware system, environmental, and physiological factors. A hardware system invalidating factor example includes the sensor signal transmission to the receiver, handheld, or monitoring unit has been compromised with noise or interrupted altogether. A potential environmental invalidating factor example occurs when a measured temperature is out of the operating range of the system. A physiological invalidating example occurs when the computed rate is too high, either physiologically so, such that the accuracy of certain numerical calculations can be suspect, or non-physiologically high such that the reliability of the latest sensor signal is questionable.

Temporal features of the sensor signal that do not correlate to glucose excursion can significantly affect calibration. Examples of such temporal features include Early Sensitivity Attenuation (ESA) and dropouts. Methods to detect and evaluate these features during operation of the analyte monitoring system as they occur (i.e., in real time) allows the system to determine the proper course of action. For example, if a dropout is determined to be taking place while the system is scheduled to perform a calibration, the system can delay the calibration request until the undesired artifact is deemed to have ended. On occasion, re-evaluating data from before and up to after a completed calibration can reveal that the completed calibration actually took place while an undesired temporal feature, such as a dropout, occurred. This retrospective conclusion can result in two possible scenarios. The first is that the errors in calibration due to the dropout can be corrected, so that the system revises the calibration in light of this new information. The second is that the errors in calibration due to the dropout cannot be corrected. In this case, the system may decide to invalidate the recently performed calibration either immediately or within a certain elapsed time after this determination. When the recently performed calibration is invalidated, the system will determine when to request a new calibration based on various factors including the likelihood of the persistence of undesired temporal features.

In some embodiments, the present disclosure attempts to adjust the calibration and/or mitigate the effect of temporal signal attenuations on calibration. By combining existing calibration checks and calculations with retrospective calibration and both real-time and retrospective artifact detection, such as for example, using a dropout detector, the present disclosure is able to salvage suspect calibration attempts that might otherwise require additional calibration.

Turning now to FIG. 1, an example of a method 100 according to the present disclosure is illustrated in a flowchart. As mentioned above, an analyte monitoring system can include a set of system tests or checks to determine if a calculated sensitivity value is acceptable and can be used to configure a sensor (102). For example, the system checks may include an outlier check, a comparison against sensor-code nominal sensitivity in the calibration validations check, a stability check, a calibration sensitivity check, an ESA check, and the like. When the system determines that calibration of the sensor should be performed, the system can request that the user provide a reference analyte concentration value (104). Sensor data continues to be received from the in vivo sensor while the request is pending (106). Known methods and apparatus for detecting undesired signal attenuation are used to determine if calibration should be deferred dues to signal attenuation (108). For example, U.S. Pat. No. 7,630,748 entitled “Method and System For Providing Analyte Monitoring” and incorporated herein by reference, describes a real-time dropout detector that may be used.

If undesired signal attenuation is detected, the system will delay the calibration request until the sensor has recovered and a true signal is generated. Otherwise, the system will allow calibration provided that other system checks are passed (110). The system stores the sensor data from before and after the calibration (112). Within a predetermined time after a calibration has been performed, the system retrospectively evaluates the sensor's signal quality around the moment of calibration to determine whether a previously undetected undesired signal attenuation was missed by the real-time detector(s) (114). If an undesired signal attenuation is determined to have taken place during the calibration, and if the true analyte signal can be reconstructed with good confidence, then the system revises the result of the calibration without the need of further user interaction (116). An example of such retrospective calibration is described in U.S. Pat. No. 7,618,369 entitled “Method and System For Dynamically updating Calibration Parameters For an Analyte Sensor” and incorporated herein by reference. However, if an undesired signal attenuation is determined to have taken place and the true analyte signal cannot be reconstructed with good confidence, then the system may request a new calibration (118).

In some embodiments, the system can take alternative courses of action depending on the characterization of the attenuation. In order to characterize attenuation, a detector estimates the degree of attenuation during a window of time by comparing the measured aggregate level of the signal in that window relative to a baseline level, which is computed from the same window of time using a different filter, or computed from a different window of time using a similarly structured filter. The comparison between the measured aggregate level and the baseline level is used to identify whether or not an attenuation is occurring. If adjacent time windows are deemed to be attenuated, then the time windows may be considered as a single attenuation event. For each event, a predicted time of onset and recovery may be computed. Some of the characterizations of attenuation include the direct ratio between the measured aggregate level and its baseline, a scaled ratio between the measured aggregate level and its baseline, a function such as a quadratic absolute ratio between the measured aggregate level and its baseline, the peak ratio, the duration between onset and recovery, or the area under the curve generated by the ratio and duration. One or more of these characterizations can be used to determine the various courses of action needed. If the attenuation completely invalidates the previously completed calibration, then the system can immediately expire that calibration, and the system may no longer display an analyte concentration value until a successful calibration has been performed. If the calibration is sufficiently accurate enough to provide useful analyte concentration value information for at least a short time after the calibration, the system momentarily allows the analyte concentration value display to continue, and then expires the calibration soon after. When appropriate, the system requests a new calibration in order to resume display of the analyte concentration value (120).

In some embodiments, in addition to compensating for signal attenuation, the methods of the present disclosure can assess the feasibility of a calibration attempt based on past and stored data to identify faults. Thus, instead of merely invalidating a calibration attempt because a check has failed, the present disclosure can determine useful information to avoid repeating failed calibrations in the future. The present disclosure assesses the sensor signal and reference analyte value history around the present and past calibration attempts. In some embodiments, the present disclosure determines whether a current failed calibration attempt is caused by the sensor currently producing unreliable sensor data, or whether the reference analyte concentration value is not reliable.

To illustrate the additional information provided by checking the reference value and the sensor data instead of just the sensitivity value, consider a case in which the latest calibration attempt is close enough to a prior attempt, whether or not the prior attempt was used to update sensitivity for glucose calculation. In addition to the system checks, these relatively closely spaced reference values can be compared to assess the feasibility of the reference value of the latest calibration attempt. For example, if the difference between the latest and previous reference values is too large considering the time interval between them and the assumed reference value measurement error variability, then the latest calibration attempt should be failed, even though none of the sensitivity based checks failed. In some cases, an outlier reference value could occur such that the sensitivity happens to be relatively unchanged because the error in reference value is almost cancelled by another erroneous factor in the calculation of sensitivity.

If there are several recent past reference value data available, other methods can be used. One example is to monitor the standard deviation of the predicted reference value using a regularization procedure or other data smoothing procedure. A dramatic increase in standard deviation around the latest attempt implies that the latest reference value has a relatively large uncertainty.

In addition to using reference value, one can make a better assessment of the latest calibration attempt by considering the sensor data. For example, given the best sensitivity used up to the latest attempt, the computed analyte level can be compared from the sensor data around the instances where recent past calibration attempts were made. The computed analyte level can then be tested against similar methods as described in the previous section. In addition, methods that require relatively constant sample time, such as an Auto Regressive (AR) model and Auto Regressive Moving Average (ARMA), can be used to see whether the latest sensor measurements correlated reasonably well to past data in the same manner as observed around the recent past attempts. Because of the relatively rich data content from the sensor, assessments can be made involving combinations of point wise analyte values as well as higher order signals. For example, one can track the analyte rate of change during recent past and latest calibration attempts, to determine whether including the latest calibration attempt significantly changes the rate distribution statistics of the moving average population.

The likelihood of undesired sensor transients can be checked to see if they differ dramatically between the latest attempts. For example, assuming a dropout detector is used to assess the likelihood of the sensor signal exhibiting dropouts, if the detector generates a much higher likelihood for dropouts than the previous attempts, then it is possible that at least the absolute raw sensor signal and/or the raw sensor rate of change may be far enough from the true value that a sensitivity check may not indicate an actual change in sensitivity. The combined knowledge of the relative confidence of the sensor, reference BG, and the resulting sensitivity calculation amongst relatively recent attempts can be useful in a number of ways. For example, the reference value and sensor signal comparisons are useful in qualifying the relative uncertainty of calibration attempts. By taking the combined information, one can weigh both the reference BG and sensor signal of recent past attempts, so that future calibration attempts can be compared against these values without having to resort to using equal weighting for each of the attempts. This results in a more reliable relative assessment. In some embodiments, the relative weighting can be used when generating the latest weighted sensitivity to be used in converting raw sensor signal into final glucose values as well as glucose rate of change. In other words, instead of using a fixed weighting factor that was predetermined offline, each calibration update can use all near past and latest attempts with weights that are derived from the combination of the above comparisons.

Additionally, a consistent pattern of increase or change in uncertainty over time of the sensor signal, reference value, or computed sensitivity can be used to infer that state of the system. Several possible scenarios and corresponding system actions are described below. For example, if the latest few attempts indicate that the sensor signal or sensor rate of change uncertainty increases dramatically compared to the older attempts, the system could conclude that the sensor is momentarily not ready to produce reliable measurements. The system could notify the user to delay the follow-up calibration attempt in order not to waste their effort. The delay time may depend on other checks, such as the various components of data quality check as well as the ESA detector. If the last attempt indicates that the latest reference value uncertainty is dramatically larger compared to the other past attempts, the system could conclude that the attempt must be failed because the latest reference value is likely to be an outlier. The system could notify the user to try again, possibly reminding the user to consult the user guide for the proper finger stick method.

If more than one recent reference value attempts show a relatively larger uncertainty than the older records, then the system can conclude that something systematic could be wrong with the reference value measurement process. Examples include using a new test strip vial but not correctly updating the sensor code, using alternate site testing (AST) without properly preparing the test area, and possibly using control solution instead of patient's blood as a sample source. The system could remind the user to consult the user guide for the proper finger stick method. If the persistent increase in uncertainty is too big, the system may decide to terminate operation to prevent producing results with poor accuracy.

If more than one attempt indicate that while both the reference value and the sensor signal are relatively reasonable, but the latest computed sensitivities result in a much larger uncertainty than before, then it is possible that a dramatic change has occurred. One possible cause would be a change in the effective dynamics between, for example, blood and interstitial glucose, such that the reference value and interstitial analyte value independently show no dramatic change in behavior, but their dynamic and possibly static relationship has changed. Another possible cause would be the loss of counter electrode function, in which a bias is introduced between measured interstitial analyte and blood analyte as measured by the reference value. In some embodiments, the system could elect to adjust the calibration factor with an offset factor as necessary. If several follow-up confirmatory finger sticks do not indicate that the offset-adjusted model has a good fit, then the system may decide to terminate operation.

FIG. 2 is a flow chart depicting an example method 200 of assessing the sensor signal and reference analyte value history around the present and past calibration attempts to determine whether a current failed calibration attempt is caused by the sensor producing unreliable sensor data, or whether the reference analyte concentration value is not reliable. Initially, a set of system checks associated with an analyte monitoring system is defined (202). A signal representative of sensor data is received from an analyte monitoring system related to an analyte level of a patient measured over time (204). A reference analyte concentration value is received (206). A sensitivity value is calculated based on the signal and the reference analyte concentration value (208). The system checks are performed on the sensitivity value (210). If the system checks indicate the sensitivity value is invalid, the system determines if the signal is a cause of the sensitivity value being invalid (212). If the system checks indicate the sensitivity value is invalid, the system determines if the reference analyte concentration value is a cause of the sensitivity value being invalid (214). The sensitivity value is recalculated using new sensor data after a delay if the signal was a cause of the sensitivity value being invalid (216). The sensitivity value is recalculated using a new reference analyte concentration value if the reference analyte concentration value was a cause of the sensitivity value being invalid (218). The analyte level of the patient is displayed based on the recalculated sensitivity value if the system checks indicate the sensitivity value is invalid (220).

Turning now to FIG. 3, another aspect of the present disclosure is described. In some embodiments, where a user-initiated provision of a reference analyte concentration value that happens to occur within a predetermined time period of a scheduled calibration request, is to be used for calibration, a retrospective review of the calibration may be used to improve the calibration. In other words, instead of determining a sensitivity of the in vivo sensor using a real-time algorithm that only considers limited sensor data in order to provide real-time responsiveness regarding the success or failure of the calibration attempt, the present disclosure can consider a wider window of sensor data since the user is not expecting feedback regarding a calibration attempt or, for that matter, the user is not aware of the calibration attempt at all.

FIG. 3 is a flowchart depicting an example method 300 of using retrospective analysis on calibration attempts that use a user-initiated provision of a reference analyte concentration value. For example, if a user happens to perform a finger stick in vitro blood glucose (BG) measurement and the outcome is provided to the analyte monitoring system within a predetermined time period of a scheduled calibration request, the system can use the information as a reference analyte concentration value for a calibration attempt and the calibration attempt can be retrospectively analyzed to refine the calibration.

In some embodiments, an analyte monitoring system is provided configured to request a reference analyte concentration value for use in calibrating an in vivo sensor of the analyte monitoring system (302). The request may be satisfied by a user-initiated supply of the reference analyte concentration value before the request or by a user response to the request. A signal representative of sensor data is received from the analyte monitoring system related to an analyte level of a patient measured over time (304). A first method of computing a sensitivity value for calibrating the sensor is provided wherein the sensitivity value is computed based on an amount of sensor data that is less than a predetermined amount (306). A second method of computing a sensitivity value for calibrating the sensor is provided wherein the sensitivity value is computed based on an amount of sensor data that is more than a predetermined amount (308). The sensor is calibrated using the first method when the request is satisfied by a user response to the request (310). The sensor is calibrated using the second method when the request is satisfied by a user initiated supply of the reference analyte concentration before the request (312). The analyte level of the patient is displayed based on the calibration of the sensor (314).

In some embodiments, the system attempts a full retrospective calibration on a first BG measurement and a partial and/or real-time calibration on a second BG measurement as soon as a scheduled calibration is required. The result can be a weighted average based on the relative uncertainty of the two results. In some embodiments, the weighting scheme can be dependent on the percentage of data points available to perform retrospective calibration. Once a weighted average of the two calibration values are obtained, then the system can use that value as the calibration value and the user will be relieved of having to take a measurement for the next scheduled calibration. If one of the values fails a calibration-related acceptance criteria, then the system ignores that value. If none of the values pass the system checks, then the normally scheduled calibration will occur.

In some embodiments, the system can wait for the second user-initiated BG measurement to have sufficient associated sensor data so that retrospective calibration can be performed on both user-initiated BG measurements without attempting a real-time calibration. If both calibrations pass the calibration-related acceptance criteria, then the weighted average will be used as the calibration value for the next scheduled calibration. If one calibration fails, then only the successful calibration is used. If both fail, then the normally scheduled calibration will occur.

In some embodiments, any user-initiated BG measurements prior to a next scheduled calibration can be used to update calibration parameters that are used to screen out unsuitable conditions. For example, pre-calibration screening routines may check the glucose rate of change based on an assumed sensitivity value and the analyte monitoring system's sensor signal close in time to a next scheduled calibration. Updating the sensitivity value using the user-initiated BG measurements can improve the reliability of such a pre-calibration screening routine, thereby reducing the frequency of unnecessary calibration requests that have a relatively elevated chance of failure. In some embodiments, any user-initiated BG measurement taken just prior to a BG measurement initiated by the system (e.g., a system requested measurement) for a next scheduled calibration, can be combined to obtain calibration results that are relatively insensitive to BG measurement errors, an analyte monitoring system sensor signal noise, and potential changes in system equilibrium since the last scheduled calibration. In some embodiments, the decision to use any one or combination of the above embodiments can be determined by a processor-based online logic system and/or by using rules derived offline from analysis of stored data. For example, the system can determine that a stepwise change in one or more system related parameters may be likely around the third to fourth day of the sensor wear.

In another aspect of the present disclosure, the system may take advantage of additional data available when a test strip is left in a reference BG meter beyond the time required by the meter's primary algorithm to determine a reference analyte concentration value. This extended time during which an analyte measurement system with an integrated reference BG meter would not normally monitor the output of the reference BG meter can be used to refine or improve the initial determination of the reference analyte concentration value by using the additional data. A second algorithm allows for an updated reference analyte concentration value and the uncertainty limits can be used by the analyte measurement system to further refine its calibration value.

In some embodiments, for each calibration instance, the primary algorithm produces a reference BG value to be used by the analyte measurement system's calibration algorithm. The reference BG value, together with the most recent sensor signal measurements, can be used to both evaluate the suitability of the current calibration attempt and to obtain the sensitivity (i.e., a scaling calibration factor), in order to produce an analyte measurement from the analyte measurement system. Immediately after, the second algorithm can use data from the existing primary algorithm plus any further signal from the BG meter's strip port until the strip is removed. The second algorithm then provides an updated reference BG value to the analyte measurement system's calibration algorithm, which uses the updated reference BG value to refine the sensitivity value. The update can occur several minutes after the test strip has been removed and requires no special action by the user.

In some embodiments a stream of recursive revisions of the reference BG meter glucose estimate is used to obtain a measure of the measurement's variability. The variability in the reference BG meter glucose value can then be translated into the variability of the computed sensitivity for calibration. This results in both upper and lower bounds of the estimated sensitivity. Knowing these bounds allow the relative precision of each calibration to be compared. In addition, any threshold checks could be done with these bounds taken into account. For example, if the latest rate check results in an estimate of 1.9±0.5 mg/(dL min), then a +2 mg/(dL min) maximum rate threshold may be too close for the latest calibration attempt to be allowed to succeed.

In some embodiments, the reference BG meter variability is first obtained by comparing the characteristics of a primary reference BG meter algorithm against the secondary reference BG meter algorithm operating at different extended time durations. These measurements are also compared against the known glucose concentration in the medium that the reference BG meter strip takes its measurement from. Any possible bias and offset correction as a function of the difference between the primary and secondary reference BG meter algorithm outputs as well as the duration of the extended time, can then be used to correct the reference BG meter reading used for calibration in the analyte measurement system. This online correction will be applied as soon as the secondary reference BG meter algorithm finishes. The calibration algorithm in the analyte measurement system can then be used to revise the sensitivity value based on the bias and/or offset adjusted reference BG meter reading.

In some embodiments, the correlation between the degree of certainty of a reference BG meter reading and the difference between primary and secondary reference BG meter algorithm outputs as a function of extended time duration is obtained. In an online implementation, as soon as the secondary reference BG meter algorithm finishes, the weighting of the latest sensitivity value can be adjusted by taking into account for the degree of certainty obtained offline.

Turning now to FIG. 4, an example embodiment of a method 400 is depicted as a flowchart. An analyte monitoring system is provided that is configured to receive a reference analyte concentration value for use in calibrating an in vivo sensor of the analyte monitoring system (402). The reference analyte concentration value is calculated based on a test strip inserted in a port of an in vitro analyte meter for a predetermined period of time. A signal representative of sensor data is received from the analyte monitoring system related to an analyte level of a patient measured over time (404). A first method of calculating the reference analyte concentration value is provided wherein the reference analyte concentration value is calculated based on data collected from the test strip within the predetermined period of time (406). A second method of calculating the reference analyte concentration value is provided wherein the reference analyte concentration value is calculated based on data collected from the test strip within the predetermined period of time and data collected from the test strip after the predetermined period of time (408). A sensitivity value is computed for calibrating the sensor based on the reference analyte concentration value computed using the first method (410). The sensitivity value for calibrating the sensor is re-computed based on the reference analyte concentration value computed using the second method if the test strip remains in the port of the in vitro analyte meter beyond the predetermined period (412). The sensor is calibrated using the sensitivity value (414). The analyte level of the patient is displayed based on the calibration of the sensor (416).

FIG. 5 depicts a Clarke Grid 500 of sensor data and reference BG points from a collection of experiments. The Clarke Grid shows a certain amount of data dispersion on the higher end. To relate to the ESA case, a subset of the data presented in FIG. 5 is shown in FIG. 6. In FIG. 6, only paired points where calibration took place during repressed sensitivity are displayed. As sensitivity recovers to its regular value, which is higher relative to when sensitivity is repressed, the sensor glucose values tend to display higher numbers. Composite sensitivity (Sc) is the effective sensitivity used to determine displayed glucose values at each data sample instance. As an illustrative example, the data points in Clarke Grid 600 in FIG. 6 are restricted to points whose calibration sensitivity value, Sc, is less than or equal to 7/10^th of the true sensitivity. True sensitivity is computed after the fact from the median of all available sensitivities, Sm. Note that this subset of the population is the group that contributes to the high-end scatter seen in FIG. 5. As the stratification is changed from 7/10^th to 8/10^th, and so on, more and more of the high-end scatter seen in FIG. 5 will get included, and less and less of the ideal scatter near the 45 degree line 502 will get excluded.

FIG. 7 summarizes the effect of stratifying the data in FIG. 5 based on the ratio between the sensitivity during calibration, Sc, and the best estimate of true sensor sensitivity, Sm. The horizontal axis is the overall MARD (Mean Absolute Relative Difference) of each group, which is a metric that roughly describes the variability of the accuracy. The vertical axis is the low end MRD (Mean Relative Difference) of each group, which is a metric that roughly describes low-end bias of the accuracy. The relative size of each circle reflects the number of paired points that fall into each group, with the “All” group having the largest number of paired points. The patterns are indicative of the Percent A region of the Clarke Grid Statistics of each group, closer to the bottom being a higher Percent A value than closer to the top.

The stratification based on Sc/Sm ratio shown in FIG. 7 is presented in 10% increments from the full population (i.e. “All” group). The groups below median are those whose Sc/Sm values are less than or equal to 0.9, 0.8, 0.7 and 0.6. Similarly, the groups above median are those whose Sc/Sm values are greater than or equal to 1.1, 1.2, and 1.3. There are not enough points for the 1.4 group.

Given the equal 10% increments, it can be observed that the performance degradation in terms of marginal increase in full MARD, marginal increase in the absolute value of low end MRD, and the marginal decrease in Percent A, moves faster for the groups below the median than for the groups above the median. Practically, this means that based on the common accuracy performance metrics, the system is much more robust to a high sensitivity calibration compared to a low sensitivity calibration of the same additive percentage change. This is still the case if the comparison were to be made in the geometric percentage change sense. For example, the group where Sc/Sm is less than or equal to 0.8 should be geometrically comparable to the group where Sc/Sm is greater than or equal to 1/0.8=1.25. One would expect that this group would lie between the 1.2 and 1.3 groups in FIG. 5. While the low end MRD is not significantly different, the MARD and percent A statistics are worse for the 0.8 group.

In the general context of calibration, it can be assumed that the effective sensitivity being somewhat higher than computed is preferred over using the “as is” value or a mildly scaled down value. This observation can be combined with real-time observations of past calibrations as well as aggregate sensor sensitivity information obtained from the manufacturing lot to improve the robustness of each calibration instance. In the particular case related to ESA, where early sensitivity calculations are most likely to be underestimated, a proper scaling of this calculated value could mean improved accuracy.

Prior information regarding the sensor based on past calibrations and/or sample lot statistics is used to obtain the best nominal estimate of the sensor sensitivity. Any statistical property of this estimate, such as mean and standard deviation, can be adjusted by the risk distribution derived by information obtained from the process summarized in FIG. 10. The result is a table of correction factors on each calibration instance based on how the current calibration value compares to prior information. This table could also be a function of elapsed time since sensor start as well as elapsed time since the last calibration. For example, instead of using median sensitivity to normalize all Sc values after the fact, nominal sensor code sensitivity can be used to normalize the Sc values. A plot similar to plot 700 in FIG. 7 can then be obtained. Using this inference, a risk distribution can be obtained, where, for any given calibration, the computed sensitivity is compared to the nominal sensor code sensitivity, and a correction factor is applied to the computed sensitivity based on this comparison. In this example, the correction factor table can be obtained a priori from a database of sensor use. The correction factor table can be periodically reviewed offline to determine whether any adjustment is necessary to the algorithm. Another example would be to compute the median sensitivity based on all past calibration attempts plus the nominal sensor code sensitivity to revise the correction factor table previously described in real time during each sensor wear. The present disclosure uses information regarding the effect of over and under estimation of computed sensitivity with respect to common performance metrics such as, but not limited to low end MRD, full MRD, MARD, Clarke Grid Percent A, best fit line intercept, and root mean squared error (RMSE).

FIG. 8 is a graph 800 depicting an example correlation between an accuracy metric and the ratio of calibration sensitivity (Sc) to ideal sensitivity (Sn). FIG. 9 is a graph 900 depicting an example of a calibration adjustment factor derived by inverting the relative accuracy metric of FIG. 8. Turning to FIG. 10, a flowchart detailing an example embodiment of a method 1000 for determining an adjustment to a calibration based on the correlation of calibration error, calibrated sensitivity, and ideal sensitivity according to the present disclosure, is shown. Paired reference BG data and calibrated sensor data are collected (1002). The data is paired (1004) and grouped based on the calibration used (1006). Each “calibration” set contains applicable paired data, the calibration sensitivity, and the ideal sensitivity. Calibration sensitivity may be either immediate or composite sensitivity. Ideal sensitivity may require raw sensor data and reference values from the past. One or more accuracy metrics are selected to be used as a basis of the adjustment (1008). For example, mean relative difference (MRD) of paired values below 100 mg/dL. A correlation is found between the accuracy metric and the ratio of calibration sensitivity (Sc) to ideal sensitivity (Sn) (1010). For example, the ideal sensitivity can be the sensor-code-based sensitivity as depicted in FIG. 8. Based on the correlation, the calibration adjustment is computed, either as a lookup table (for different ranges of Sc/Sn values) or a continuous function (1012). For example, given the relative accuracy metric above, the calibration adjustment becomes a relative value. So that adjusted glucose Ga is a function of raw glucose Gr, calibration sensitivity Sc, and the calibration adjustment Kc (labeled as Glucose Correction Factor in FIG. 9).Ga=Kc Gr/Sc, where Kc is the inverse of the Accuracy Metric.

For an additive accuracy metric such as Mean Difference, the adjustment is:Ga=(Gr/Sc)+Kc, where Kc is the negative of the Accuracy Metric.

For subsequent calibrations, the latest function or lookup table is used to determine the necessary calibration correction (1014). The method 1000 repeats if new data becomes available (1016) or otherwise uses the latest function or lookup table (1018).

FIG. 11 is a flowchart detailing an example embodiment of a method 1100 for applying an adjustment to a calibration in real-time according to the present disclosure. Once the in vivo sensor of an analyte monitoring system has been started and an initial calibration completed, the calibration sensitivity is compared to the sensor-code-based sensitivity (1102). The sensor-code-based sensitivity is used as the ideal sensitivity. The correction function or lookup table is next used to find out the required correction given the latest calibration sensitivity, ideal sensitivity, and elapsed time since sensor started (1104). From the time of calibration to the scheduled time to the next calibration, the amount of adjustment is gradually increased from 0 to the maximum as determined by the previous process (1106). The adjustment modifies how the calibration sensitivity transforms raw signal into glucose concentration units. If the sensor is still active (1108), the method 1100 waits for the next calibration (1110) and then repeats. Otherwise, the method 1100 ends.

FIG. 12 is a flowchart detailing an example embodiment of an alternative method 1200 for applying an adjustment to a calibration in real-time according to the present disclosure. Once the in vivo sensor of an analyte monitoring system has been started and an initial calibration completed, the calibration sensitivity is compared to the sensor-code-based sensitivity (1202). In this embodiment, the median of the sensitivities from prior calibrations is used as the ideal sensitivity. The correction function or lookup table is then used to determine the required correction given the latest calibration sensitivity, ideal sensitivity, and elapsed time since sensor started (1204). From the time of calibration to the scheduled time to the next calibration, the amount of adjustment is gradually increased from 0 to the maximum as determined by the previous process (1206). The adjustment modifies how the calibration sensitivity transforms raw signal into glucose concentration units. If the sensor is still active (1208), the method 1200 waits for the next calibration (1210) and then repeats. Otherwise, the method 1200 ends.

Accordingly, in certain embodiments, a computer-implemented method includes receiving a signal representative of sensor data from an analyte sensor configured to monitor an analyte level over time, calibrating the analyte sensor if attenuation of the received signal from the analyte sensor is not detectable, deferring calibration of the analyte sensor if attenuation of the received signal from the analyte sensor is detectable when calibration is requested, wherein the calibration is deferred until the attenuation of the received signal is no longer detected, storing the signal from the analyte sensor for a period of time spanning from before the calibration to after the calibration, determining whether a previously undetected signal attenuation has occurred during the calibration based on the stored signal after the calibration has been completed, performing reconstruction of the received signal based on the stored signal from the analyte sensor for the period of time spanning from before the calibration to after the calibration, when it is determined that the previously undetected signal attenuation has occurred, to generate a reconstructed signal, updating the calibration of the analyte sensor based on the reconstructed signal, and invalidating the calibration of the analyte sensor and requesting a new calibration of the analyte sensor if performing the reconstruction of the received signal does not generate the reconstructed signal.

Certain embodiments further include displaying an analyte concentration value corresponding to the monitored analyte level if the previously undetected signal attenuation did not occur during calibration based on the stored signal, or if performing the reconstruction of the received signal generated the reconstructed signal.

Certain embodiments also include suspending displaying the analyte concentration value if the calibration of the analyte sensor is invalidated and the previously undetected signal attenuation is greater than a predefined threshold.

Certain embodiments further include displaying an analyte concentration value when the new calibration is performed and the previously undetected signal attenuation is less than a predefined threshold.

Certain embodiments further include delaying invalidating the calibration if the signal attenuation is determined to be continuing to occur.

In certain embodiments, deferring calibration if attenuation of the signal from the analyte sensor is detectable further includes detecting signal attenuation from at least one of a hardware system condition, an environmental condition, and a physiological condition.

In certain embodiments, deferring calibration if attenuation of the signal from the analyte sensor is detectable further includes detecting signal attenuation from at least one of early signal attenuation and a dropout.

A system for monitoring an analyte using an in vivo sensor in accordance with an embodiment of the present disclosure includes a processor, a memory operatively coupled to the processor, the memory storing instructions which, when executed by the processor, causes the processor to: receive a signal representative of sensor data from an analyte sensor configured to monitor an analyte level over time, calibrate the analyte sensor if attenuation of the received signal from the analyte sensor is not detectable, defer calibration of the analyte sensor if attenuation of the received signal from the analyte sensor is detectable when calibration is requested, wherein the calibration is deferred until the attenuation of the received signal is no longer detected, store the signal from the analyte sensor for a period of time spanning from before the calibration to after the calibration, determine whether a previously undetected signal attenuation has occurred during the calibration based on the stored signal after the calibration has been completed, perform reconstruction of the received signal based on the stored signal from the analyte sensor for the period of time spanning from before the calibration to after the calibration, when it is determined that the previously undetected signal attenuation has occurred, to generate a reconstructed signal, update the calibration of the analyte sensor based on the reconstructed signal, and invalidate the calibration of the analyte sensor and request a new calibration of the analyte sensor if performing the reconstruction of the received signal does not generate the reconstructed signal.

A computer program product stored on a computer-readable medium in accordance with one embodiment of the present disclosure includes instructions to: receive a signal representative of sensor data from an analyte sensor configured to monitor an analyte level over time, calibrate the analyte sensor if attenuation of the received signal from the analyte sensor is not detectable, defer calibration of the analyte sensor if attenuation of the received signal from the analyte sensor is detectable when calibration is requested, wherein the calibration is deferred until the attenuation of the received signal is no longer detected, store the signal from the analyte sensor for a period of time spanning from before the calibration to after the calibration, determine whether a previously undetected signal attenuation has occurred during the calibration based on the stored signal after the calibration has been completed, perform reconstruction of the received signal based on the stored signal from the analyte sensor for the period of time spanning from before the calibration to after the calibration, when it is determined that the previously undetected signal attenuation has occurred, to generate a reconstructed signal, update the calibration of the analyte sensor based on the reconstructed signal, and invalidate the calibration of the analyte sensor and request a new calibration of the analyte sensor if performing the reconstruction of the received signal does not generate the reconstructed signal.

Certain embodiments include a computer-implemented method comprising defining a set of system checks associated with an analyte monitoring system, receiving a signal representative of sensor data from an analyte monitoring system related to an analyte level of a patient measured over time, receiving a reference analyte concentration value, calculating a sensitivity value based on the signal and the reference analyte concentration value, performing the system checks on the sensitivity value, determining if the signal is a cause of the sensitivity value being invalid if the system checks indicate the sensitivity value is invalid, determining if the reference analyte concentration value is a cause of the sensitivity value being invalid if the system checks indicate the sensitivity value is invalid, recalculating the sensitivity value using new sensor data after a delay if the signal was a cause of the sensitivity value being invalid, recalculating the sensitivity value using a new reference analyte concentration value if the reference analyte concentration value was a cause of the sensitivity value being invalid, and displaying the analyte level of the patient based on the recalculated sensitivity value if the system checks indicate the sensitivity value is invalid.

In certain embodiments, determining if the reference analyte concentration value is a cause of the sensitivity value being invalid includes determining if a difference between the reference analyte concentration value and a prior reference analyte concentration value is larger than a predefined threshold amount wherein the prior reference analyte concentration value was received within a predefined time period of the reference analyte concentration value.

In certain embodiments, determining if the reference analyte concentration value is a cause of the sensitivity value being invalid includes determining if a difference between a standard deviation of the reference analyte concentration value and a standard deviation of a predicted reference analyte concentration is larger than a predefined threshold amount, wherein the standard deviation of the predicted reference analyte concentration is determined based on a plurality of prior reference analyte concentration values.

In certain embodiments, determining if the signal is a cause of the sensitivity value being invalid includes determining if a difference between the signal and a prior signal received during a prior calculation of a sensitivity value is larger than a predefined threshold amount wherein the prior signal was received within a predefined time period of the signal.

In certain embodiments, determining if the signal is a cause of the sensitivity value being invalid includes determining if the signal correlates with prior signals received during prior calculations of sensitivity values to within a predetermined amount.

In certain embodiments, determining if the signal is a cause of the sensitivity value being invalid includes determining if including the signal in a moving average population of prior signals received during prior calculations of sensitivity values changes a rate of change distribution of the moving average population by more than a predetermined threshold.

Certain embodiments may further comprise calculating a weighting of the sensitivity value indicative of a relative confidence rating of the accuracy of the sensitivity value, wherein the weighting is determined based on the signal and the reference analyte concentration value compared to prior signals and the reference analyte concentration values received during prior calculations of sensitivity values.

Certain embodiments include a computer-implemented method comprising providing an analyte monitoring system configured to request a reference analyte concentration value for use in calibrating an in vivo sensor of the analyte monitoring system, wherein the request may be satisfied by a user initiated supply of the reference analyte concentration value before the request or by a user response to the request, receiving a signal representative of sensor data from the analyte monitoring system related to an analyte level of a patient measured over time, providing a first method of computing a sensitivity value for calibrating the sensor wherein the sensitivity value is computed based on an amount of sensor data that is less than a predetermined amount, providing a second method of computing a sensitivity value for calibrating the sensor wherein the sensitivity value is computed based on an amount of sensor data that is more than a predetermined amount, calibrating the sensor using the first method when the request is satisfied by a user response to the request, calibrating the sensor using the second method when the request is satisfied by a user initiated supply of the reference analyte concentration before the request, and displaying the analyte level of the patient based on the calibration of the sensor.

In certain embodiments, if the user initiated supply of the reference analyte concentration occurs within a predetermined time before the request, the sensor is calibrated using the first method.

In certain embodiments, if the user supplies a reference analyte concentration before the request and a second reference analyte concentration in response to the request, a sensitivity value is computed for each analyte concentration and a weighted average of the sensitivity values is used for calibration.

In certain embodiments, if the user supplies two or more reference analyte concentrations before the request, a sensitivity value is computed for each analyte concentration and a weighted average of the sensitivity values is used for calibration.

In certain embodiments, if the user initiated supply of the reference analyte concentration occurs within a predetermined time before the request and a sensitivity value based on the second method passes an acceptance criteria, the system cancels the request and the sensor is calibrated using the second method.

Certain embodiments may further comprise defining a set of acceptance criteria for using a sensitivity value to calibrate the sensor, wherein if the user supplies two or more reference analyte concentrations, a sensitivity value is computed for each analyte concentration and the sensor is calibrated based on only sensitivity values that pass the acceptance criteria.

Certain embodiments include a computer-implemented method comprising providing an analyte monitoring system configured to receive a reference analyte concentration value for use in calibrating an in vivo sensor of the analyte monitoring system, wherein the reference analyte concentration value is calculated based on a test strip inserted in a port of an in vitro analyte meter for a predetermined period of time, receiving a signal representative of sensor data from the analyte monitoring system related to an analyte level of a patient measured over time, providing a first method of calculating the reference analyte concentration value wherein the reference analyte concentration value is calculated based on data collected from the test strip within the predetermined period of time, providing a second method of calculating the reference analyte concentration value wherein the reference analyte concentration value is calculated based on data collected from the test strip within the predetermined period of time and data collected from the test strip after the predetermined period of time, computing a sensitivity value for calibrating the sensor based on the reference analyte concentration value computed using the first method, re-computing the sensitivity value for calibrating the sensor based on the reference analyte concentration value computed using the second method if the test strip remains in the port of the in vitro analyte meter beyond the predetermined period, calibrating the sensor using the sensitivity value, and displaying the analyte level of the patient based on the calibration of the sensor.

Certain embodiments may further comprise determining a first value indicative of a variability of the reference analyte concentration value; determining a second value indicative of a variability of the sensitivity value based on the first value; and verifying validity of a subsequent calibration attempt using a subsequently computed sensitivity value based on the second value.

Certain embodiments may further comprise recursively re-computing the reference analyte concentration value based on data collected from the test strip within the predetermined period of time and data collected from the test strip after each additional passage of the predetermined period of time or fraction thereof.

In certain embodiments, re-computing the sensitivity value includes using a weighted average of the reference analyte concentration values computed using the first method and the second method.

In certain embodiments, the weighting is determined based upon historical differences between reference analyte concentration values computed using the first method and the second method.

In certain embodiments, the historical differences between reference analyte concentration values computed using the first method and the second method include reference analyte concentration values computed using the second method wherein the test strip remains in the port of the in vitro analyte meter for varying amounts of time.

Certain embodiments include a computer-implemented method comprising providing an analyte monitoring system configured to receive a reference analyte concentration value for use in calibrating an in vivo sensor of the analyte monitoring system, wherein the reference analyte concentration value is calculated based on a test strip inserted in a port of an in vitro analyte meter for a first predetermined period of time, receiving a signal representative of sensor data from the analyte monitoring system related to an analyte level of a patient measured over time, pairing the reference analyte concentration value with the sensor data in real time during the first predetermined period of time, calculating a first sensitivity value based on the reference analyte concentration value from the in vitro analyte meter during the first predetermined period of time, recursively updating the first sensitivity value for as long as the test strip remains in the port and can provide updated reference analyte concentration values, calculating a final sensitivity value after a second predetermined period of time based on sensor data up to the second predetermined time, paired with an updated reference analyte concentration value from a most recent recursive update.

In certain embodiments, the first predetermined period of time is approximately 3 seconds to approximately 10 seconds and wherein the second predetermined period of time is approximately 3 minutes to approximately 15 minutes.

Certain embodiments include a computer-implemented method comprising providing an analyte monitoring system configured to receive a reference analyte concentration value for use in calibrating an in vivo sensor of the analyte monitoring system, providing a nominal estimate of a sensitivity of the sensor based on at least one of past calibrations and sample lot statistics, adjusting one or more statistical properties of the nominal estimate based on a risk distribution, receiving a signal representative of sensor data from the analyte monitoring system related to an analyte level of a patient measured over time, computing a current sensitivity based on the reference analyte concentration value and the sensor data, determining a table of correction factors for each past calibration based on a comparison of a current sensitivity with prior sensitivity values, applying a correction factor from the table of correction factors to the computed current sensitivity to determine a corrected sensitivity, and displaying an analyte level of a patient based on the sensor being calibrated using the corrected sensitivity.

In certain embodiments, the risk distribution is derived by stratifying paired sensor data and reference analyte concentration values plotted on a Clarke Grid based on a ratio of sensor sensitivity during calibration and an estimate of true sensor sensitivity.

Certain embodiments include a computer-implemented method comprising providing an analyte monitoring system configured to receive a reference analyte concentration value for use in calibrating an in vivo sensor of the analyte monitoring system, receiving a signal representative of sensor data from the analyte monitoring system related to an analyte level of a patient measured over time, storing sensor data, calibrated sensor data, calibration information and reference analyte concentration values, pairing reference analyte concentration values with calibrated sensor data to form paired data, grouping paired data into calibration sets based on calibration sensitivity used for each paired data, computing one or more calibration error metrics for each calibration set, determining a correlation between the one or more calibration error metrics of the calibration sets and one or more measureable factors, determining a correction function that maps the determined correlation between the one or more calibration error metrics of the calibration sets and one or more measureable factors, computing a current sensitivity based on the reference analyte concentration value and the sensor data, applying a correction factor based on the correction function to the computed current sensitivity to determine a corrected sensitivity, and displaying an analyte level of a patient based on the sensor being calibrated using the corrected sensitivity.

In certain embodiments, the correction function is implemented as a look-up table.

In certain embodiments, the measureable factors include at least one of time of day of calibration, elapsed time since sensor start of calibration, and the ratio of calibration sensitivity to ideal sensitivity.

Certain embodiments may further comprise updating the correction function based on at least one of subsequently received sensor data, calibrated sensor data, calibration information and reference analyte concentration values.

Various other modifications and alterations in the structure and method of operation of the embodiments of the present disclosure will be apparent to those skilled in the art without departing from the scope and spirit of the present disclosure. Although the present disclosure has been described in connection with certain embodiments, it should be understood that the present disclosure as claimed should not be unduly limited to such embodiments. It is intended that the following claims define the scope of the present disclosure and that structures and methods within the scope of these claims and their equivalents be covered thereby.

Claims

1. A method for calibrating a signal generated by a glucose sensor indicative of a glucose level, comprising: storing, by at least one processor, a first sensitivity for the glucose sensor having an in vivo portion configured to be positioned in contact with an interstitial fluid of a user and an ex vivo portion;receiving, by the at least one processor, a prior calibrated glucose level data from a sensor electronics of the glucose sensor, wherein the calibrated glucose level data is based upon a signal generated by the glucose sensor over a first time period and indicative of the glucose level of the user;receiving, by the at least one processor, a reference blood glucose measurement of the user;pairing, by the at least one processor, the reference blood glucose measurement with the calibrated glucose level data to form a paired data;determining, by the at least one processor, an adjustment for the first sensitivity of the glucose sensor, wherein the adjustment includes a scaling factor determined based on the paired data, a distribution of sensor sensitivity, and a sensor start time;calibrating, by the at least one processor, the signal generated by the glucose sensor over a second time period after the first time period to corresponding glucose level data using the stored first sensitivity and the adjustment after the glucose sensor is positioned in contact with the interstitial fluid of the user; andtransmitting, by the at least one processor, the glucose level data to a display unit.

2. The method of claim 1, wherein the first sensitivity includes a median sensitivity.

3. The method of claim 1, wherein the first sensitivity includes a mean sensitivity.

4. The method of claim 1, wherein the calibrating includes determining a second sensitivity for the glucose sensor using the adjustment and the first sensitivity.

5. The method of claim 1, further comprising storing the adjustment in a non-transitory memory of a sensor electronics for the glucose sensor.

6. The method of claim 5, wherein storing the adjustment comprises storing one or more correction factors in a table.

7. The method of claim 1, further comprising updating the adjustment in a non-transitory memory of the sensor electronics.

8. The method of claim 1, further comprising updating the adjustment in a non-transitory memory of the sensor electronics in real time.

9. The method of claim 1, wherein the first sensitivity includes a lot sensitivity of a plurality of glucose sensor of a manufacturing lot, and the plurality of glucose sensors includes the glucose sensor.

10. The method of claim 9, wherein the adjustment is based on a plurality of prior sensor uses of a subset of the plurality of glucose sensors of the manufacturing lot.

11. The method of claim 9, wherein the lot sensitivity includes a standard deviation.

12. The method of claim 1, further comprising: receiving, by the at least one processor, a second calibrated glucose level data from a sensor electronics of the glucose sensor, wherein the second calibrated glucose level data is based upon a second signal generated by the glucose sensor over a second time period and indicative of a second glucose level of the user;receiving, by the at least one processor, a second reference blood glucose measurement of the user;pairing, by the at least one processor, the second reference blood glucose measurement with the second calibrated glucose level data to form a second paired data; anddetermining, by the at least one processor, a second adjustment for the first sensitivity of the glucose sensor, wherein the second adjustment includes a second scaling factor determined based on the second paired data, the distribution of sensor sensitivity, and the sensor start time.

13. The method of claim 1, wherein the adjustment is determined based further on a correlation between a ratio of calibration sensitivity to ideal sensitivity and an accuracy metric.

14. The method of claim 1, wherein the reference blood glucose measurement comprises a user-initiated reference blood glucose measurement.

15. The method of claim 1, wherein the distribution of sensor sensitivity is a risk distribution.

16. The method of claim 1, wherein the adjustment comprises a function.

17. A device for calibrating a signal generated by a glucose sensor indicative of a glucose level, comprising: the glucose sensor having an in vivo portion configured to be positioned in contact with an interstitial fluid of a user;a non-transitory memory; andat least one processor communicatively coupled to the non-transitory memory and the glucose sensor and configured to: store a first sensitivity for the glucose sensor;receive a prior calibrated glucose level data from a sensor electronics of the glucose sensor, wherein the calibrated glucose level data is based upon a signal generated by the glucose sensor over a first time period and indicative of the glucose level of the user;receive a reference blood glucose measurement of the user;pair the reference blood glucose measurement with the calibrated glucose level data to form a paired data;determine an adjustment for the first sensitivity of the glucose sensor, wherein the adjustment includes a scaling factor determined based on the paired data, a distribution of sensor sensitivity, and a sensor start time;calibrate the signal generated by the glucose sensor over a second time period after the first time period to corresponding glucose level data using the received first sensitivity and the determined adjustment after the glucose sensor is positioned in contact with the interstitial fluid of the user; andtransmit the glucose level data to a display unit.

18. The device of claim 17, wherein the first sensitivity includes a median sensitivity.

19. The device of claim 17, wherein the first sensitivity includes a mean sensitivity.

20. The device of claim 17, wherein the at least one processor is configured to calibrate by determining a second sensitivity for the glucose sensor using the adjustment and the first sensitivity.

21. The device of claim 17, wherein the at least one processor is further configured to store the adjustment in the non-transitory memory.

22. The device of claim 21, wherein the at least one processor is configured to store the adjustment by at least storing one or more correction factors in a table.

23. The device of claim 17, wherein the at least one processor is further configured to update the adjustment in the non-transitory memory of the sensor electronics.

24. The device of claim 17, wherein the at least one processor is further configured to update the adjustment in the non-transitory memory of the sensor electronics in real time.

25. The device of claim 17, wherein the first sensitivity includes a lot sensitivity of a plurality of glucose sensor of a manufacturing lot, and the plurality of glucose sensors includes the glucose sensor.

26. The device of claim 25, wherein the adjustment is based on a plurality of prior sensor uses of a subset of the plurality of glucose sensors of the manufacturing lot.

27. The device of claim 25, wherein the lot sensitivity includes a standard deviation.

28. The device of claim 17, wherein the at least one processor is further configured to: receive a second calibrated glucose level data from a sensor electronics of the glucose sensor, wherein the second calibrated glucose level data is based upon a second signal generated by the glucose sensor over a second time period and indicative of a second glucose level of the user;receive a second reference blood glucose measurement of the user;pair the second reference blood glucose measurement with the second calibrated glucose level data to form a second paired data; anddetermine a second adjustment for the first sensitivity of the glucose sensor, wherein the second adjustment includes a second scaling factor determined based on the second paired data, the distribution of sensor sensitivity, and the sensor start time.

29. The device of claim 17, wherein the adjustment is determined based further on a correlation between a ratio of calibration sensitivity to ideal sensitivity and an accuracy metric.

30. The device of claim 17, wherein the reference blood glucose measurement comprises a user-initiated reference blood glucose measurement.

31. The device of claim 17, wherein the distribution of sensor sensitivity is a risk distribution.

32. The device of claim 17, wherein the adjustment comprises a function.