Patent Application
20210298686
Embodiments described herein generally relate to systems and methods for improving patient care and, more particularly but not exclusively, to systems and methods for managing alerts during at least one of patient evaluation, diagnosis, and treatment.
Contextual and circumstantial information regarding a patient before, during, or after a clinical encounter are important contributors to clinical support decisions. However, this type of information is subjective in nature, and is usually not measured by clinical measurement devices. Accordingly, this information is often not recorded or otherwise stored in electronic databases.
Additionally, alerting a clinician to consider this type of information too often may overburden the clinician and lead to alert fatigue. This may cause a clinician to develop a reduced sensitivity to alarms and the inclination to act upon them. This in turn could have an adverse impact on the quality of care that a patient receives.
A need exists, therefore, for systems and methods that overcome the disadvantages of existing measurement and alerting techniques.
This summary is provided to introduce a selection of concepts in a simplified form that are further described below in the Detailed Description section. This summary is not intended to identify or exclude key features or essential features of the claimed subject matter, nor is it intended to be used as an aid in determining the scope of the claimed subject matter.
According to one aspect, embodiments relate to a method for managing alerts during at least one of patient evaluation, diagnosis, and treatment. The method includes receiving patient data using an interface; receiving, using the interface, a classification decision from a classifier executing a classification model, wherein the classification decision is based on the patient data; determining, using a processor executing instructions stored on a memory, whether the classification decision is a borderline classification decision; and applying, using the processor, at least one alert filter to the patient data upon determining the classification decision is a borderline classification decision, wherein the at least one alert filter is applied to determine whether to issue a contextual data alert to a clinician to prompt the clinician to consider contextual data related to the patient.
In some embodiments, the method further includes issuing a contextual data alert to the clinician to prompt the clinician to consider contextual data related to the patient.
In some embodiments, determining whether the classification decision is a borderline classification decision includes comparing the classification decision to a decision boundary, determining whether the classification decision is within a predetermined threshold from the decision boundary, and determining the classification decision is a borderline classification decision upon determining the classification decision is within the predetermined threshold from the decision boundary.
In some embodiments, applying the at least one alert filter includes analyzing the patient data to determine whether at least one alert filter condition contradicts the received borderline classification decision, and wherein the method further includes issuing a contextual data alert to the clinician upon the processor determining an alert filter condition contradicts the received borderline classification decision. In some embodiments, the at least one alert filter condition relates to at least one of patient visitation history, patient response to a treatment, a pattern of a clinically measured feature, a type of clinical measurement, and a change in a measured clinical feature.
In some embodiments, the at least one alert filter is selected based on the classification model.
In some embodiments, the contextual data related to the patient includes at least one of mood of the patient, physical appearance of the patient, emotional state of the patient, and agility of the patient.
According to another aspect, embodiments relate to a system for managing alerts during at least one of patient evaluation, diagnosis, and treatment. The system includes an interface for receiving patient data and a classification decision from a classifier executing a classification model, wherein the classification decision is based on the patient data; and a processor executing instructions stored on a memory to determine whether the classification decision is a borderline classification decision and apply at least one alert filter to the patient data upon determining the classification decision is a borderline classification decision, wherein the at least one alert filter is applied to determine whether to issue a contextual data alert to a clinician to prompt the clinician to consider contextual data related to the patient.
In some embodiments, the processor is further configured to issue, using a user interface, a contextual data alert to the clinician to prompt the clinician to consider contextual data related to the patient.
In some embodiments, the processor is further configured to determine whether the classification decision is a borderline classification decision by comparing the classification decision to a decision boundary, determining whether the classification decision is within a predetermined threshold from the decision boundary, and determining the classification decision is a borderline classification decision upon determining the classification decision is within a predetermined threshold from the decision boundary.
In some embodiments, the processor applies the at least one alert filter by analyzing the patient data to determine whether at least one alert filter condition contradicts the received borderline classification decision, and the processor is further configured to issue, using a user interface, a contextual data alert to the clinician upon determining an alert filter condition contradicts the received borderline classification decision. In some embodiments, the at least one alert filter condition relates to at least one of patient visitation history, patient response to a treatment, a pattern of a clinically measured feature, a type of clinical measurement, and a change in a measured clinical feature.
In some embodiments, the at least one alert filter is selected based on the classification model.
In some embodiments, the contextual data related to the patient includes at least one of mood of the patient, physical appearance of the patient, emotional state of the patient, and agility of the patient.
According to yet another aspect, embodiments relate to a non-transitory computer readable medium containing computer-executable instructions for performing a method for managing alerts during at least one of patient evaluation, diagnosis and treatment. The medium includes computer-executable instructions for receiving patient data using an interface, computer-executable instructions for receiving, using the interface, a classification decision from a classifier executing a classification model, wherein the classification decision is based on the patient data, computer-executable instructions for determining, using a processor executing instructions stored on a memory, whether the classification decision is a borderline classification decision, and computer-executable instructions for applying, using the processor, at least one alert filter to the patient data upon determining the classification decision is a borderline classification decision, wherein the at least one alert filter is applied to determine whether to issue a contextual data alert to a clinician to prompt the clinician to consider contextual data related to the patient.
Non-limiting and non-exhaustive embodiments of the embodiments herein are described with reference to the following figures, wherein like reference numerals refer to like parts throughout the various views unless otherwise specified.
Various embodiments are described more fully below with reference to the accompanying drawings, which form a part hereof, and which show specific exemplary embodiments. However, the concepts of the present disclosure may be implemented in many different forms and should not be construed as limited to the embodiments set forth herein; rather, these embodiments are provided as part of a thorough and complete disclosure, to fully convey the scope of the concepts, techniques and implementations of the present disclosure to those skilled in the art. Embodiments may be practiced as methods, systems or devices. Accordingly, embodiments may take the form of a hardware implementation, an entirely software implementation or an implementation combining software and hardware aspects. The following detailed description is, therefore, not to be taken in a limiting sense.
Reference in the specification to “one embodiment” or to “an embodiment” means that a particular feature, structure, or characteristic described in connection with the embodiments is included in at least one example implementation or technique in accordance with the present disclosure. The appearances of the phrase “in one embodiment” in various places in the specification are not necessarily all referring to the same embodiment. The appearances of the phrase “in some embodiments” in various places in the specification are not necessarily all referring to the same embodiments.
Some portions of the description that follow are presented in terms of symbolic representations of operations on non-transient signals stored within a computer memory. These descriptions and representations are used by those skilled in the data processing arts to most effectively convey the substance of their work to others skilled in the art. Such operations typically require physical manipulations of physical quantities. Usually, though not necessarily, these quantities take the form of electrical, magnetic or optical signals capable of being stored, transferred, combined, compared and otherwise manipulated. It is convenient at times, principally for reasons of common usage, to refer to these signals as bits, values, elements, symbols, characters, terms, numbers, or the like. Furthermore, it is also convenient at times, to refer to certain arrangements of steps requiring physical manipulations of physical quantities as modules or code devices, without loss of generality.
However, all of these and similar terms are to be associated with the appropriate physical quantities and are merely convenient labels applied to these quantities. Unless specifically stated otherwise as apparent from the following discussion, it is appreciated that throughout the description, discussions utilizing terms such as “processing” or “computing” or “calculating” or “determining” or “displaying” or the like, refer to the action and processes of a computer system, or similar electronic computing device, that manipulates and transforms data represented as physical (electronic) quantities within the computer system memories or registers or other such information storage, transmission or display devices. Portions of the present disclosure include processes and instructions that may be embodied in software, firmware or hardware, and when embodied in software, may be downloaded to reside on and be operated from different platforms used by a variety of operating systems.
The present disclosure also relates to an apparatus for performing the operations herein. This apparatus may be specially constructed for the required purposes, or it may comprise a general-purpose computer selectively activated or reconfigured by a computer program stored in the computer. Such a computer program may be stored in a computer readable storage medium, such as, but is not limited to, any type of disk including floppy disks, optical disks, CD-ROMs, magnetic-optical disks, read-only memories (ROMs), random access memories (RAMs), EPROMs, EEPROMs, magnetic or optical cards, application specific integrated circuits (ASICs), or any type of media suitable for storing electronic instructions, and each may be coupled to a computer system bus. Furthermore, the computers referred to in the specification may include a single processor or may be architectures employing multiple processor designs for increased computing capability.
The processes and displays presented herein are not inherently related to any particular computer or other apparatus. Various general-purpose systems may also be used with programs in accordance with the teachings herein, or it may prove convenient to construct more specialized apparatus to perform one or more method steps. The structure for a variety of these systems is discussed in the description below. In addition, any particular programming language that is sufficient for achieving the techniques and implementations of the present disclosure may be used. A variety of programming languages may be used to implement the present disclosure as discussed herein.
In addition, the language used in the specification has been principally selected for readability and instructional purposes and may not have been selected to delineate or circumscribe the disclosed subject matter. Accordingly, the present disclosure is intended to be illustrative, and not limiting, of the scope of the concepts discussed herein.
Embodiments described herein provide systems and methods for considering contextual and/or circumstantial (for simplicity, “contextual”) information related to a patient during at least one of evaluation, diagnosis, and/or treatment of the patient. For example, features such as the patient's mood, the patient's general appearance, how the patient moves and walks (e.g., whether they need support), and secondary responses to clinical encounters may be helpful in providing patient care.
However, this information is subjective in nature and not readily or automatically measured by sensor devices. Accordingly, this information is often not considered by machine learning procedures in providing classification decisions regarding a patient.
Classification models may therefore not have or otherwise consider all relevant information to make a confident classification decision regarding a patient. As a result, a classification model may output a “borderline” classification decision indicating that an accurate classification is difficult to achieve (e.g., because contextual data was not gathered/stored in a digital form and therefore was not considered). These borderline classification decisions may ultimately result in false positives and false negatives, and the patient being mistreated.
Accordingly, clinical care systems and methods can improve by considering contextual information related to a patient in these borderline situations. In other words, a clinician's decision may be more accurate if contextual information were considered as opposed to a clinician relying solely on the borderline classification decision.
Features of the present application may therefore alert clinicians of these borderline cases and prompt the clinician to further consider certain contextual information related to the patient. This information may come in the form of a clinician's inputs, and therefore the systems and methods may leverage the clinician's clinical experience to provide insights in addition to those provided by the classification model(s).
Features of the various embodiments described herein, however, also consider the undesirable potential to overwhelm the clinician with excessive alerts. Excess alerts may lead to alert fatigue in which clinicians develop a reduced sensitivity to alerts and the motivation to act upon them. In fact, issues associated with alerts have been ranked as a top technology health hazard.
Furthermore, alerts are often accompanied by sounds that can have a detrimental effect on patient comfort (e.g., by causing heightened stress, delirium, etc.). This is particularly true with vulnerable patients, such as infants or patients in intensive care units.
Accordingly, the features of various embodiments described herein may first consider whether a classification decision is a borderline classification decision. If so, one or more alert filters may be applied to patient data to determine one or more alert filter conditions. If an alert filter condition goes against or otherwise contradicts the borderline classification decision, then a contextual data alert may be issued to a clinician to prompt the clinician to consider certain contextual data related to the patient. If the alert condition agrees with or otherwise does not contradict the classification decision, no contextual data alert will be issued.
The processor 120 may be any hardware device capable of executing instructions stored on memory 130 or storage 160 or otherwise capable of processing data. As such, the processor 120 may include a microprocessor, field programmable gate array (FPGA), application-specific integrated circuit (ASIC), or other similar devices.
The memory 130 may include various memories such as, for example L 1 , L2, or L3 cache or system memory. As such, the memory 130 may include static random access memory (SRAM), dynamic RAM (DRAM), flash memory, read only memory (ROM), or other similar memory devices. The exact configuration of the memory 130 may vary as long as instructions for managing alerts can be executed.
The user interface 140 may execute on one or more devices for enabling communication with a user such as a clinician or other type of medical personnel. For example, the user interface 140 may include a display, a mouse, and a keyboard for receiving user commands. In some embodiments, the user interface 140 may include a command line interface or graphical user interface that may be presented to a remote terminal via the network interface 150.
The user interface 140 may execute on a user device such as a PC, laptop, tablet, mobile device, smartwatch, or the like. The exact configuration of the user interface 140 and the device on which it executes may vary as along as the features of various embodiments described herein may be accomplished.
The network interface 150 may include one or more devices for enabling communication with other hardware devices. For example, the network interface 150 may include a network interface card (NIC) configured to communicate according to the Ethernet protocol. Additionally, the network interface 150 may implement a TCP/IP stack for communication according to the TCP/IP protocols. Various alternative or additional hardware or configurations for the network interface 150 will be apparent.
The network interface 150 may be in operable communication with one or more sensor devices 151. In the healthcare context, these may include sensors configured as part of patient monitoring devices that gather various types of information regather patient's health.
The type of sensor devices 151 used may of course vary and may depend on the patient and context. Accordingly, any type of sensor device 151 may be used as long as they can gather or otherwise obtain the required data regarding the entity under analysis.
The sensor device(s) 151 may be in communication with the system 100 over one or more networks that may link the various components with various types of network connections. The network(s) may be comprised of, or may interface to, any one or more of the Internet, an intranet, a Personal Area Network (PAN), a Local Area Network (LAN), a Wide Area Network (WAN), a Metropolitan Area Network (MAN), a storage area network (SAN), a frame relay connection, an Advanced Intelligent Network (AIN) connection, a synchronous optical network (SONET) connection, a digital T1, T3, E1, or E3 line, a Digital Data Service (DDS) connection, a Digital Subscriber Line (DSL) connection, an Ethernet connection, an Integrated Services Digital
Network (ISDN) line, a dial-up port such as a V.90, a V.34, or a V.34 bis analog modem connection, a cable modem, an Asynchronous Transfer Mode (ATM) connection, a Fiber Distributed Data Interface (FDDI) connection, a Copper Distributed Data Interface (CDDI) connection, or an optical/DWDM network.
The network or networks may also comprise, include, or interface to any one or more of a Wireless Application Protocol (WAP) link, a Wi-Fi link, a microwave link, a General Packet Radio Service (GPRS) link, a Global System for Mobile Communication G(SM) link, a Code Division Multiple Access (CDMA) link, or a Time Division Multiple access (TDMA) link such as a cellular phone channel, a Global Positioning System (GPS) link, a cellular digital packet data (CDPD) link, a Research in Motion, Limited (RIM) duplex paging type device, a Bluetooth radio link, or an IEEE 802.11-based link.
The storage 160 may include one or more machine-readable storage media such as read-only memory (ROM), random-access memory (RAM), magnetic disk storage media, optical storage media, flash-memory devices, or similar storage media. In various embodiments, the storage 160 may store instructions for execution by the processor 120 or data upon which the processor 120 may operate. For example, the storage 160 may include borderline classification identification instructions 161, one or more classification models 162 and 163, clinician alert instructions 164, and one or more alert filters 165 and 166 applicable to one or more classification models.
In step 204, one or more classification models (such as the models 162 and 163) may analyze the patient data of step 202 to obtain a classification decision. The classification decision may relate to whether a patient tests positive for a condition, whether the patient should be discharged or admitted into a medical unit, or whether the patient should be prescribed certain medication, for example.
The one or more models may execute any type of suitable machine learning classification procedure. These may include, but are not limited to, logistic regression, random forests, neural networks, or the like.
Step 206 may involve receiving a classification decision from the classification model(s). In some embodiments, the output may include a probability of whether or not a patient should be admitted to a certain healthcare unit or the probability that a patient has some condition, for example.
Step 208 involves analyzing the classification decision to determine whether the classification decision is a borderline classification decision. This step may be performed by a physician/clinician alerting mechanism that executes the borderline identification instructions 161 of
The borderline identification instructions 161 may determine whether a classification decision is a borderline classification decision in accordance with the following. If the classification decision of the current encounter pi is within a threshold pthresh of a decision boundary pb, then the classification decision is a borderline classification decision. In other words, the borderline classification instructions 161 determine that a classification decision is a borderline classification upon determining that:
|pi−pb|<pthresh
These boundaries may be determined and/or presented in a number of ways. For example,
As seen in
Upon determining the classification decision is a borderline classification decision, the clinician alerting mechanism may in step 208 determine whether any applicable alert filter conditions contradict the classification decision and may present a display of patient information in step 210. If an alert filter condition contradicts the borderline classification decision, the clinician alerting mechanism may issue an alert to prompt the clinician to consider/provide inputs regarding contextual data related to the patient in step 212.
Step 420 involves applying the classification model selected in step 410 to patient data to receive a classification decision regarding the patient. The classification decision may relate to whether the patient should be admitted into the healthcare institution (such as a specific department thereof) or discharged and sent home. Or, in other embodiments, the classification decision may relate to whether or not the patient is diagnosed with a certain condition. In yet other embodiments, the classification decision may relate to whether or not the patient should be prescribed certain medication. These embodiments are merely exemplary and it is contemplated that the classification model may be configured to make other types of classification decisions.
Step 430 involves determining whether the classification decision is a borderline classification decision. As discussed above, the classification model may not be able to make a classification decision with a required or desired level of confidence.
For example, if the classification model is configured to output a classification in terms of a numeric value (such as on a sigmoid function scale as shown in
If the answer in step 430 is no (i.e., the classification decision is not a borderline classification decision), the method 400 may proceed to step 440. Step 440 specifies that if the classification decision is not a borderline classification, no contextual data alerts will be issued. This is essentially because there is not a requisite degree of uncertainty and therefore additional contextual data isn't needed to obtain an accurate classification decision.
If the answer in step 430 is yes (i.e., the classification decision is a borderline classification), the method 400 may proceed to step 450. Step 450 involves applying at least one alert filter. The alert filter(s) consider certain types of patient data to obtain a deeper view of the patient's overall state, as well as a clinician's knowledge of the patient's health.
The data subjected to the alert filters may include some of or all of the data upon which the classification is based. Additionally or alternatively, the data subjected to the alert filters may include data that is different than the data upon which the classification decision is based.
In some embodiments, the alert filter may consider patient visitation history. For example, if the patient visits the Emergency Department (ED) for the first time for a given symptom, they are often admitted for a more thorough examination. If the patient visits the ED repeatedly for the same condition(s) and in a short period of time, admitting the patient to the hospital (or a certain department thereof) will help determine if the condition is worsening and the condition's associated causes. On the other hand, if the repeated visits are spaced over a long period of time, efforts may be channeled towards stabilizing the patient in the ED and then sending them home.
In other words, if the classification decision was an instruction to admit the patient, and it is the patient's first visit to the hospital, a clinician would likely want to conduct a more thorough examination. Accordingly, an alert filter may be “is this the patient's first visit?” If the answer is “yes,” the method may ultimately be more inclined to issue a contextual data alert to prompt the clinician to input or otherwise consider contextual data related to the patient.
On the other hand, if the classification decision was an instruction to admit the patient but the patient has made multiple visits in the past, the clinician is likely already familiar with the patient. In this scenario, the method 400 may be more inclined to not issue a contextual data alert as this would likely only annoy the clinician. Or, another alert filter may be applied.
Another alert filter may look for patterns of measurements that are indicative of high acuity and therefore warrant admission of the patient to a higher level of care. These patterns may include, but are not limited to, the presence of certain measurements, the frequency of measurement, the method through which a measurement is obtained, or the like.
Certain measurements are only ordered by a clinician when the clinician is already concerned about the patient's condition. Therefore, the mere fact that a certain measurement has been ordered is indicative of the patient's deteriorating state (as well as the clinician's knowledge of the patient's deteriorating state). In these situations, the method 400 may be inclined to not issue a contextual data alert as the clinician is already familiar with the patient and the patient's state.
These features can be discovered by analyzing the features that are less common across a patient population. Additionally, if a measurement (e.g., vitals) is measured too frequently, it is often indicative of the care team's awareness of the patient's deteriorating state. Furthermore, the usage of invasive (as opposed to non-invasive) measurement methods, or higher sensitivity methods, to measure a certain clinical value is usually an indication of higher acuity.
For example, invasively measured blood pressure indicates a higher degree of clinician concern than non-invasive methods. In these instances, alerting the clinician of a borderline admission decision is redundant and should be avoided since the clinician already knows the patient is deteriorating and should be admitted.
Another alert filter may relate to patient data trends. For patients with acute exacerbation originating from an underlying chronic disease, patient history or the way the patient has changed over time is especially important in channeling clinical decisions. Trend analysis of past patient data can therefore provide valuable insight with respect to a patient's condition.
For example,
It may also be useful to carry out the above analysis over different temporal durations of past data. For instance, all available data may be considered or only data from a certain window is considered such as the past six months, the past month, the past 7 days, data from the past 2 days, the past 24 hours, during triage, etc. If the clinical value being accessed changes by a large extent as indicated by the aforementioned statistics, an alert filter condition may suggest the patient should be admitted.
Another alert filter may consider patient responsiveness to treatment and/or the aggressiveness of treatment provided to the patient in the ED. If a patient is responsive to treatment in the ED and is stabilized, then the clinician should consider possible discharge. If, however, the borderline classification decision was to admit the patient, and the patient has been responsive to treatment and is stabilized, an alarm should be issued to the clinician as the alert filter condition contradicts the classification decision. In other words, if the borderline classification decision is to admit the patient, but other patient data shows that the patient has been responsive to treatment and is stabilized, then an alert should be issued to the clinician prompting the clinician to consider contextual data related to the patient.
For instance, discharge is likely considered if the patient responded well to diuretics administered in the ED. A simple way to evaluate treatment responsiveness may be to estimate dosage and duration of medications. For example, one could calculate equivalent dosages of different types of diuretic drugs with different routes of administration (e.g., per oral drip IV, continuous IV pump, etc.). Escalating dosages generally indicate non-responsiveness and therefore admission to higher levels of care. More comprehensive but complicated assessment may be through analysis of changes in edema and fluid retention on chest X-ray images.
Referring back to
For example, the classification decision may have been to admit a patient to the emergency department. The alert filter condition (based on the alert filter(s)), may indicate that the patient has not responded to treatment. In this case, the alert condition may be said to agree with or otherwise not contradict the classification decision. The method 400 may then proceed to step 440 in which case no contextual data alert is issued.
If on the other hand, the alert condition indicates that the patient has responded to treatment, the alert filter condition may be said to contradict the classification decision of admitting the patient to the ED. In this case the method 400 of
Step 470 involves issuing a contextual data alert to a clinician. The contextual data alert may prompt the clinician to consider certain types of contextual data related to the patient so that the patient can receive more appropriate treatment. As discussed above, this information about the patient is usually unrecorded. However, it can be incorporated in the decision process to resolve difficult cases (i.e., cases with borderline classification decisions).
An alert may be issued to the clinician via any suitable user device. These alerts may include an audio-based alert, a visual-based alert, a haptic-based alert, or some combination thereof. For example, with reference back to
Specifically, the alert may involve a prompt for the clinician's attention, a presentation of more detailed clinical parameters and their temporal trends, and a presentation of the borderline clinical values. This information may be presented in any suitable forms such as text, graphics, audio messages, graphs, animations, or the like.
The clinician may then consider contextual information regarding the patient and/or the patient encounter that is not observable by the classification model(s). The clinician may enter this contextual information using input devices such as a keyboard, a mouse, a microphone, or by interacting with a touchscreen on a display, scanning a handwritten note to be analyzed by OCR techniques, or the like.
The patient data may be received via an interface such as the interface 150 of
Step 604 involves receiving, using the interface, a classification decision from a classifier executing a classification model, wherein the classification decision is based on the patient data. The classification decision may relate to a clinical care decision such as whether to admit the patient into a healthcare institution department or to treat the patient and send them home. The classification decision may be made by any suitable classification model executing one or more various machine learning procedures such as logistic regression, random forests, CNNs, RNNs, or the like.
Step 606 involves determining, using a processor executing instructions stored on a memory, whether the classification decision is a borderline classification decision. Oftentimes a classification decision may not have a desired confidence due to, for example, lack of certain types of patient data.
For example, a classification decision may be within a certain threshold of a decision boundary as described in conjunction with
Step 608 involves applying, using the processor, at least one alert filter to the patient data upon determining the classification decision is a borderline classification decision, wherein the at least one alert filter is applied to determine whether to issue a contextual data alert to a clinician to prompt the clinician to consider contextual data related to the patient. The at least one alert filter may be applied to certain types of patient data to determine one or more alert filter conditions. As discussed above, the alert filter condition(s) may relate to specific aspects regarding a patient's health, their personal history, their medical history, etc. This information may include whether the patient has responded well to previous treatment(s), the type of patient parameters previously measured, how these parameters have been previously measured, or the like.
Step 610 is optional and involves issuing a contextual data alert to the clinician to prompt the clinician to consider contextual data related to the patient. A processor such as the processor 120 of
The methods, systems, and devices discussed above are examples. Various configurations may omit, substitute, or add various procedures or components as appropriate. For instance, in alternative configurations, the methods may be performed in an order different from that described, and that various steps may be added, omitted, or combined. Also, features described with respect to certain configurations may be combined in various other configurations. Different aspects and elements of the configurations may be combined in a similar manner. Also, technology evolves and, thus, many of the elements are examples and do not limit the scope of the disclosure or claims.
Embodiments of the present disclosure, for example, are described above with reference to block diagrams and/or operational illustrations of methods, systems, and computer program products according to embodiments of the present disclosure. The functions/acts noted in the blocks may occur out of the order as shown in any flowchart. For example, two blocks shown in succession may in fact be executed substantially concurrent or the blocks may sometimes be executed in the reverse order, depending upon the functionality/acts involved. Additionally, or alternatively, not all of the blocks shown in any flowchart need to be performed and/or executed. For example, if a given flowchart has five blocks containing functions/acts, it may be the case that only three of the five blocks are performed and/or executed. In this example, any of the three of the five blocks may be performed and/or executed.
A statement that a value exceeds (or is more than) a first threshold value is equivalent to a statement that the value meets or exceeds a second threshold value that is slightly greater than the first threshold value, e.g., the second threshold value being one value higher than the first threshold value in the resolution of a relevant system. A statement that a value is less than (or is within) a first threshold value is equivalent to a statement that the value is less than or equal to a second threshold value that is slightly lower than the first threshold value, e.g., the second threshold value being one value lower than the first threshold value in the resolution of the relevant system.
Specific details are given in the description to provide a thorough understanding of example configurations (including implementations). However, configurations may be practiced without these specific details. For example, well-known circuits, processes, algorithms, structures, and techniques have been shown without unnecessary detail in order to avoid obscuring the configurations. This description provides example configurations only, and does not limit the scope, applicability, or configurations of the claims. Rather, the preceding description of the configurations will provide those skilled in the art with an enabling description for implementing described techniques. Various changes may be made in the function and arrangement of elements without departing from the spirit or scope of the disclosure.
Having described several example configurations, various modifications, alternative constructions, and equivalents may be used without departing from the spirit of the disclosure. For example, the above elements may be components of a larger system, wherein other rules may take precedence over or otherwise modify the application of various implementations or techniques of the present disclosure. Also, a number of steps may be undertaken before, during, or after the above elements are considered.
Having been provided with the description and illustration of the present application, one skilled in the art may envision variations, modifications, and alternate embodiments falling within the general inventive concept discussed in this application that do not depart from the scope of the following claims.
For example, the above-described systems and methods can be applied to emergency department triage of acute heart failure (AHF) patients. ED is the first in line of care for patients suffering from AHF, and a triage decision of “admit to hospital” or “send home” needs to be made quickly.
To alleviate the burden in the ED and the healthcare institution (e.g., a hospital) as a whole, the systems and methods described above may analyze patients suffering from heart failure to recognize those that should be admitted and those who can be treated and sent home. Accordingly, the systems and methods described above may help correctly identifier borderline cases by the upstream machine learning classification algorithm and prompt clinicians, when appropriate, to reconsider the models' classification/triage decision in the context of the current patient and patient encounter. As discussed above, this contextual information is generally not recorded and therefore not considered by the algorithm.
In fact, several types of contextual information contribute to acute exacerbations but are not recorded or stored in electronic databases. The systems and methods of various embodiments described above therefore improves the management of patients with, for example AHF.
In addition to AHF, the systems and methods of various embodiments described above can be generalized to other conditions, such as kidney disease, asthma, and COPD. Different version of the alert mechanisms may be used in conjunction with various different machine learning procedures. For example, the alert triggers used in conjunction with AHF will likely be different from that used for chronic kidney disease. Additionally, the framework proposed may be adapted to changing clinical practices and variation in clinical practice between clinicians, hospitals, geographical areas, or the like.
The systems and methods of various embodiments described above can also be generalized to transfer decisions in clinical settings other than the ED. For example, the systems and methods described above may be implemented in the general ward.
In fact, the processes outlined for prompting clinicians for contextual information can be generalized to any classification task. For example, the systems and methods described above may be applied to classification tasks involving financial market analysis, logistic operations, manufacturing, weather analysis, or the like.
Regardless of the exact application, the identification of borderline cases close to a decision boundary can be based on a classification probability of a distance measure (e.g., Euclidean distance). The alert mechanism(s) can be co-developed with the upstream machine learning classification model. Accordingly, specific alert filters may be applied based on the classification model used.
Additionally or alternatively, the alert mechanisms could be used with existing CDS algorithms such as those developed by the Applicant. Additionally, the alert systems and methods can be created using machine learning whose target population is the subpopulation of borderline cases identified by the upstream classification algorithm, thereby building a hierarchical pipeline of algorithms. For instance, alerts may be issued for a candidate case if the associated decision by the algorithm goes against the decision by any one of the conditions specified by the method. More complicated relations between the conditions can be learned if the machine learning is deployed.
Lastly, the alert fatigue prevention methods described above may be used for developing new CDS algorithms or refining existing ones.
| Filing Document | Filing Date | Country | Kind |
|---|---|---|---|
| PCT/EP2019/070557 | 7/30/2019 | WO | 00 |
| Number | Date | Country | |
|---|---|---|---|
| 62715825 | Aug 2018 | US |
