Systems and methods for clustering physiological events

Abstract

Systems and methods for managing physiological events generated by a patient's medical device are described. An exemplary system receives information regarding a plurality of physiological events detected by a patient's medical device, uses the received information to generate a respective set of features for each of the plurality of physiological events, and includes a controller circuit that clusters the plurality of physiological events into a plurality of different event groups using values of temporal or morphological features of the generated sets of features. Each event group includes a respective set of physiological events. The controller circuit may identify a representative event that represents the physiological events of the event group from at least one of the event groups and output the representative event and a display indicating that the representative event has been determined to a user or process, the display indicating that the representative event represents the set of physiological events within the event group.

Description

[Technical Field]

[0001] This application claims the benefit of U.S. Provisional Patent Application No. 63 / 359,333, filed July 8, 2022, which is incorporated herein by reference in its entirety. [Background technology]

[0002] Implantable medical devices (IMDs) are used to monitor patient health or disease states and deliver therapy. For example, implantable cardioverter-defibrillators (ICDs) can be used to monitor certain abnormal heart rhythms (or cardiac arrhythmias) and deliver electrical energy to the heart to correct the abnormal rhythm. One of the most common cardiac arrhythmias is atrial fibrillation (AF), which causes an inappropriately fast, irregular heart rate when chaotic electrical impulses originating in or near the atria are irregularly conducted to the ventricles. AF can be paroxysmal, lasting from minutes to days before naturally terminating. Persistent AF can last for more than a week and typically requires drug or other treatment to return to normal sinus rhythm. AF is permanent if normal heart rhythm cannot be restored with treatment. Some IMDs can be used to monitor chronic cardiac diseases, such as worsening heart failure (WHF), which can be characterized by cardiac dyssynchrony within or between the ventricles. IMDs can also provide cardiac stimulation therapy, such as cardiac resynchronization therapy (CRT), to correct cardiac dyssynchrony.

[0003] Some IMDs may record physiological data of a particular medical event upon detection, such as a cardiac arrhythmia event, a WHF event, or other physiological event, and may generate a patient alert notification. Some IMDs may register the onset of a patient-initiated medical event and record physiological data in response to the patient trigger. The recorded physiological data and alert notification may be provided to a medical professional to inform them of the patient's health status. Using this notification, the medical provider may choose to review the recorded physiological data associated with the event onset, determine the physiological event detected by the device (e.g., determine the presence and possible cause of the physiological event), or evaluate whether a treatment resulted in a desired therapeutic outcome.

[0004] A patient management system may monitor a patient using an IMD interconnected to the patient management system via a data communications network, which may allow a healthcare provider to remotely follow up with the patient or periodically assess device functionality. Summary of the Invention

[0005] A patient management system may store and manage a large number of alert notifications corresponding to physiological events detected and reported by ambulatory medical devices (AMDs). For example, when managing a cohort of AMD patients in a clinic, the patient management system may frequently receive alert notifications regarding various cardiac arrhythmia episodes or worsening heart failure (WHF) events detected by implantable cardiac devices, such as cardiac monitors, pacemakers, implantable cardioverter-defibrillators, or cardiac resynchronization therapy devices. Some AMDs may register patient-initiated episodes, such as when a patient exhibits certain signs or symptoms or experiences a precursor event indicative of a medical event (e.g., cardiac arrhythmia, syncope, or WHF event). Physiological data associated with the device-detected physiological events or patient-initiated episodes may be transmitted to the patient management system and reviewed by a clinician, for example, to determine the device-detected physiological events, schedule a patient follow-up visit, or reprogram the AMD, among other purposes.

[0006] With multiple AMDs connected to a patient management system, reviewing device-detected physiological events (e.g., cardiac arrhythmia episodes) can require significant time and costly clinical, technical, and human resources for a medical facility. Physiological events with similar or similar underlying causes (e.g., certain types of arrhythmia episodes) may repeatedly trigger alert notifications when detected by an AMD. For example, a patient with a history of atrial fibrillation (AF) may frequently experience AF episodes that are properly detected (true positive, i.e., TP detection) and inappropriately detected (false positive, i.e., FP detection) by an implantable medical device. Excessive TP detection of AF episodes may reflect frequent recurrent AF events, be caused by the device being overly sensitive to noise, or be caused by temporary stabilization of heart rate during an underlying, long-lasting AF episode. Recurrent FP detection during AF episodes can be due to atrioventricular conduction abnormalities such as premature ventricular contractions (PVCs), premature atrial contractions (PACs), or Wenckebach atrioventricular block, among others.

[0007] Such excessive event occurrences (either TP or FP detections) and repeated alert notifications increase clinician workload and medical costs. For example, because event occurrences may be routinely reviewed by clinicians or other human experts, excessive event occurrences increase the human burden of reviewing and / or interpreting such occurrences. Furthermore, storing such a large number of physiological events (e.g., sensor data and seizure summaries) detected by a device can occupy a large amount of the device's computational and storage resources.

[0008] Embodiments of the systems, devices, and methods described herein may advantageously improve the functionality of physiological event management systems, including identifying and clustering physiological events with similar characteristics and prioritizing the storage, display (e.g., display for user review and determination), or transmission of physiological event data between devices (e.g., transmission from an implanted device to an external device). An exemplary system includes a controller circuit that receives information regarding a plurality of physiological events detected by a patient's medical devices, generates a respective feature set for each of the plurality of physiological events, and uses values ​​of temporal or morphological features of the feature set to cluster the plurality of physiological events into a plurality of distinct event groups. Each event group includes a respective set of physiological events. The controller circuit may determine a representative event from at least one event group that is representative of the physiological events in that event group. The representative event and an indication that the representative event has been determined and that the representative event represents the set of physiological events in the event group may then be provided to a user or process. The representative event may be automatically displayed and made available for user judgment, while other physiological events within the same event group are hidden and assigned the same judgment result as the representative event by default unless requested by the user, and may be displayed and / or judged on demand.

[0009] Example 1 is a system for managing a plurality of physiological events detected by a patient's medical device, the system comprising a controller circuit configured to receive information regarding a plurality of physiological events, use the received information to generate, for each of the plurality of physiological events, a respective feature set including values ​​of temporal or morphological features, use the values ​​of the temporal or morphological features in the generated feature set to cluster the plurality of physiological events into a plurality of different event groups, each of the plurality of different event groups including a respective set of physiological events, determine a representative event from at least one of the plurality of different event groups, the representative event representing the set of physiological events in the at least one event group, and output the representative event and an indication that the representative event has been determined, where the representative event represents the set of physiological events in the at least one event group, to a user or process of the system.

[0010] In Example 2, the subject matter of Example 1 optionally includes that the controller circuit is configured to initiate automatic display of information related to the determined representative event on a user interface, hide information related to the set of physiological events of the at least one event group on the user interface, and initiate on-demand display of information related to one or more physiological events in the set of physiological events of the at least one event group in response to a user command via the user interface.

[0011] In Example 3, the subject matter of Example 2 optionally includes: the user command includes a user selection of a physiological event from within the set of physiological events of the at least one event group; and the controller circuit is configured to display information regarding the selected physiological event on the user interface in response to the user selection.

[0012] In Example 4, the subject matter of any one or more of Examples 2 to 3 optionally includes a user interface configured to receive a determination result of the representative event from a user, and the controller circuit is configured to assign the received determination result of the representative event to each physiological event in the set of physiological events of the at least one event group.

[0013] In Example 5, the subject matter of any one or more of Examples 1 to 4 optionally includes that the controller circuit is configured to prioritize storage of information relating to the determined representative event in the memory device over storage of information relating to the set of physiological events of the at least one event group.

[0014] In Example 6, the subject matter of any one or more of Examples 1 to 5 optionally includes that the controller circuit is configured to prioritize transmission of information regarding the determined representative event over transmission of information regarding the set of physiological events of the at least one event group between different devices.

[0015] In Example 7, the subject matter of any one or more of Examples 1 to 6 optionally includes that the plurality of physiological events includes cardiac arrhythmia episodes detected from the patient, and the generated feature set includes at least one of a heart rate distribution feature, an arrhythmia duration, a morphological feature, or a time at which the plurality of physiological events are detected.

[0016] In Example 8, the subject matter of any one or more of Examples 1 to 7 optionally includes: to determine the representative event, the controller circuit is configured to determine a centroid of the set of physiological events of the at least one event group in a feature space using feature sets generated from at least some of the physiological events in the set of physiological events, respectively, and to determine the representative event of the at least one event group as the physiological event closest to the determined centroid.

[0017] In Example 9, the subject matter of any one or more of Examples 1 to 8 optionally includes: to cluster the plurality of physiological events, the controller circuit is configured to initialize cluster centers using a feature set of one physiological event among the plurality of physiological events, iteratively update the cluster centers using central tendencies of feature sets generated from physiological events in an event group respectively until a convergence criterion or an iterative stopping criterion is met, and assign the physiological events to the clustered event groups based on a similarity metric between the physiological events and the iteratively updated cluster centers.

[0018] In Example 10, the subject matter of Example 9 optionally includes the controller circuit being configured to calculate a respective similarity metric between the representative event and each physiological event in the set of physiological events of the at least one event group, select a subset of events from the set of physiological events of the at least one event group, each having the similarity metric less than a similarity threshold, and output an indication to a user or process of the system that the representative event represents the selected subset of events.

[0019] In Example 11, the subject matter of any one or more of Examples 1 to 10 optionally includes that the controller circuit is configured to calculate a respective similarity metric between each of the received plurality of physiological events and the user-specified physiological event using feature sets generated from the received plurality of physiological events and the user-specified physiological event, determine a set of the physiological events of the at least one event group having the similarity metric exceeding a similarity threshold, and determine the user-specified physiological event as the representative event of the at least one event group.

[0020] In Example 12, the subject matter of Example 11 optionally includes a user interface configured to display the received plurality of physiological events in a graphical representation of a feature space, receive user input to identify the user-specified physiological event, and display information about the user-specified physiological event and an indication that the representative event represents the set of physiological events in the at least one event group.

[0021] In Example 13, the subject matter of Example 12 optionally includes, wherein the user interface is configured to receive a user input of the similarity threshold or a user selection of a predetermined cluster density.

[0022] In Example 14, the subject matter of any one or more of Examples 11 to 13 optionally includes that the controller circuit is configured to generate an estimate of a workload reduction associated with assigning a user-determined result of the representative event to each physiological event in the set of physiological events of the at least one event group, and determine or adjust the similarity threshold based on the generated estimate of the workload reduction.

[0023] In Example 15, the subject matter of any one or more of Examples 11 to 14 optionally includes that the controller circuit is configured to generate an estimate of an error rate associated with assigning a user-determined result of the representative event to each physiological event in the set of physiological events of the at least one event group, and determine or adjust the similarity threshold based on the generated estimate of the error rate.

[0024] Example 16 is a method for managing a plurality of physiological events detected in a patient by a medical device, the method including: receiving information regarding the plurality of physiological events; using the received information to generate, for each of the plurality of physiological events, a respective feature set including temporal or morphological features having respective values; using the values ​​of the temporal or morphological features of the generated feature set to cluster the plurality of physiological events into a plurality of different event groups, each of the different event groups including a respective set of physiological events; determining a representative event from at least one of the plurality of different event groups, the representative event representing the set of physiological events in the at least one event group; and outputting the representative event and an indication that the representative event has been determined and that the representative event represents the set of physiological events in the at least one event group to a user or process of the system.

[0025] In Example 17, the subject matter of Example 16 optionally includes initiating automatic display of information regarding the determined representative event on a user interface, hiding information regarding the set of physiological events of the at least one event group on the user interface, and initiating on-demand display of information regarding one or more physiological events in the set of physiological events of the at least one event group in response to a user command via the user interface.

[0026] In Example 18, the subject matter of Example 17 optionally includes receiving a determination result of the representative event from a user via the user interface, and automatically assigning the received determination result of the representative event to each physiological event in the set of physiological events of the at least one event group.

[0027] In Example 19, the subject matter of any one or more of Examples 16-18 optionally includes prioritizing the storage or transmission of information relating to the determined representative event over the storage or transmission of information relating to the set of physiological events of the at least one event group.

[0028] In Example 20, the subject matter of any one or more of Examples 16 to 19 optionally includes that clustering the plurality of physiological events into a plurality of different event groups includes initializing cluster centers using a feature set of one physiological event among the plurality of physiological events, iteratively updating the cluster centers using central tendencies of feature sets generated from each physiological event in the event group until a convergence criterion or an iterative stopping criterion is met, and assigning the physiological events to the clustered event groups based on a similarity metric between the physiological events and the iteratively updated cluster centers.

[0029] In Example 21, the subject matter of any one or more of Examples 16 to 20 optionally includes that clustering the plurality of physiological events into a plurality of different event groups includes calculating a respective similarity metric between each of the received plurality of physiological events and the user-specified physiological event using feature sets generated respectively from the received plurality of physiological events and the user-specified physiological event, determining a set of the physiological events of the at least one event group having the similarity metric exceeding a similarity threshold, and identifying the user-specified physiological event as the representative event of the at least one event group.

[0030] In Example 22, the subject matter of Example 21 optionally includes, via a user interface, displaying the received plurality of physiological events in a graphical representation of a feature space, receiving the user-specified physiological events and user input of the similarity threshold or user selection of a predetermined cluster density, and displaying information about the user-specified physiological events and an indication that the representative event represents the set of physiological events in the at least one event group.

[0031] In Example 23, the subject matter of any one or more of Examples 21 to 22 optionally includes generating an estimate of a workload reduction amount or an estimate of an error rate associated with assigning the user judgment result of the representative event to each physiological event in the set of physiological events of the at least one event group, and determining or adjusting the similarity threshold based on the generated estimate of the workload reduction amount or the estimate of the error rate.

[0032] The systems, devices, and methods described herein may improve physiological event and alert management in patient monitoring systems. As mentioned above, one challenge in physiological event management is that clinicians typically need to respond to a large number of excessive physiological event or alert notifications, which can be time-consuming and costly. This specification provides a technical solution for simplifying the review and adjudication process by prioritizing multiple physiological events within the same group for review and adjudication, such as by clustering multiple physiological events into multiple groups, identifying a representative event in each event group, and automatically displaying only the representative event for adjudication, while hiding other events within the event group unless requested by the user (i.e., on-demand display of events other than the representative event). Compared to conventional physiological event and alert management systems and methods, the cluster-based event review and adjudication described herein according to various embodiments can substantially reduce the number of events or alert notifications presented to clinicians with little or no additional cost or system complexity. Therefore, the systems and methods described herein may better coordinate medical resources to address critically ill patients.

[0033] The cluster-based event review and adjudication described herein may also improve the functionality of computer systems for managing physiological events and alerts. Event clustering, presentation, and adjudication may be implemented within and performed by a communicator, mobile monitor, programmer, or remote patient management system that communicates with the AMD. Thus, in some cases, improved physiological event and alert management can be achieved without modifying existing patient AMDs or physiological event detectors. Because only a reduced set of physiological events (e.g., representative events within a clustered event group) is presented for adjudication, the complexity and operational costs of the patient management system can be reduced. Prioritized data storage, such as automatically storing only representative events within an event group and storing other events only upon user request (i.e., on-demand storage), may more efficiently use memory space. In some embodiments, physiological event detection implemented in the AMD may be updated based on user adjudication of physiological events. The updated detection helps reduce false positive or false negative detections, and may result in fewer unnecessary device therapies, medications, and procedures being scheduled, prescribed, or delivered, extending AMD battery life and longevity and saving overall system costs.

[0034] This Summary is an overview of some of the teachings of the present application and is not intended to be an exclusive or exhaustive treatment of the present subject matter. Further details about the present subject matter are found in the detailed description and claims. Other aspects of the present disclosure will become apparent to those skilled in the art upon reading and understanding the following detailed description and viewing the drawings that form a part thereof, each of which should not be construed in a limiting sense. The scope of the present disclosure is defined by the claims and their legal equivalents.

[0035] Various embodiments are illustrated by way of example in the accompanying drawings, and such embodiments are intended to be illustrative and not exhaustive or exclusive embodiments of the present subject matter. [Brief explanation of the drawings]

[0036] [Figure 1] FIG. 1 illustrates generally an example patient management system and part of the environment in which the system may operate. [Figure 2] FIG. 2 illustrates a schematic diagram of an example physiological event management system configured to cluster multiple physiological events to facilitate clinician review and interpretation. [Figure 3] FIG. 3 illustrates generally one example of a cardiac event management system configured to evaluate and prioritize cardiac events detected by medical devices associated with a patient. [Figure 4] Figures 4A and 4B show examples of clustering multiple physiological events in feature space using an unsupervised clustering algorithm (Figure 4A) or user-specified cluster centers and clustering criteria (Figure 4B). [Figure 5] FIG. 5 illustrates an example of at least a portion of a user interface for displaying information about a physiological event and receiving a user determination of the event. [Figure 6] FIG. 6 shows examples of workload reduction curves and decision error rate curves at different radius values ​​corresponding to user-defined clustering event regions. [Figure 7] FIG. 7 is a flow chart illustrating an example of a method for managing the presentation and adjudication of physiological events detected by a patient's medical device. [Figure 8] FIG. 8 illustrates generally a block diagram of an example machine capable of executing any one or more of the techniques (eg, methods) described herein. DETAILED DESCRIPTION OF THE INVENTION

[0037] Disclosed herein are systems, devices, and methods for managing multiple physiological events generated by a patient's medical device. An exemplary system includes a controller circuit that receives information about multiple physiological events detected by the patient's medical device, generates a respective feature set for each of the multiple physiological events using the received information, and can cluster the multiple physiological events into multiple different event groups using values ​​of temporal or morphological features in the generated feature set. Each of the clustered event groups includes a respective set of physiological events. The controller circuit can determine a representative event from at least one event group that represents the set of physiological events in the event group, and output the representative event and an indication that the representative event has been determined and that the representative event represents the set of physiological events in the at least one event group to a user or process. The representative event can be automatically displayed and made available for user evaluation, while other physiological events in the same event group are hidden unless requested by the user and are assigned the same evaluation result as the representative event by default, and can be displayed and / or evaluated upon request.

[0038] While the present specification focuses on the management of arrhythmia events, those skilled in the art will understand that this is by way of example and not limitation. The physiological event and alert management systems, devices, and methods according to various embodiments described herein may also be applicable to the management of other physiological events. The methods or techniques described herein may be implemented in a variety of portable (e.g., implantable, wearable, or retainable) or fixed devices or medical systems.

[0039] 1 illustrates an exemplary patient management system 100 and a portion of the environment in which the patient management system 100 may operate. The patient management system 100 may perform a range of activities, including remote patient monitoring and diagnosis of disease conditions. Such activities may be performed near the patient 101, such as in the patient's home or office, via a centralized server, such as in a hospital, clinic, or doctor's office, or via a remote workstation, such as a secure wireless mobile computing device.

[0040] The patient management system 100 may include one or more portable medical devices, an external system 105, and a communication link 111 that provides communication between the one or more portable medical devices and the external system 105. The one or more portable medical devices may include an implantable medical device (IMD) 102, a wearable medical device (WMD) 103, or one or more other implantable, leadless, subcutaneous, external, wearable, or portable medical devices configured to monitor, sense, or detect information from various conditions of the patient 101, such as one or more cardiac or non-cardiac conditions (e.g., dehydration, sleep-disordered breathing, etc.), determine physiological information related to those conditions, and provide one or more therapies to treat those conditions.

[0041] In one example, IMD 102 may include one or more conventional cardiac rhythm management devices implanted in the patient's chest with a lead system including one or more intravenous, subcutaneous, or non-invasive leads or catheters for placing one or more electrodes or other sensors (e.g., heart sound sensors) in, on, or around the heart, chest, abdomen, or neck of patient 101 at one or more locations. In another example, IMD 102 may include a monitor implanted, for example, subcutaneously in the chest of patient 101. IMD 102 includes a housing that houses circuitry and, in some examples, one or more sensors, such as a temperature sensor.

[0042] The IMD 102 may include evaluation circuitry configured to detect or determine certain physiological information of the patient 101, determine one or more conditions, provide information or alerts to the patient 101 (e.g., the patient), a clinician, or one or more other caregivers or one or more other processes, etc. The IMD 102 may alternatively or additionally be configured as a therapy device configured to treat one or more medical conditions of the patient 101. The therapy may be delivered to the patient 101 via a lead system and associated electrodes or using one or more other delivery mechanisms. The therapy may include delivery of one or more drugs to the patient 101, such as using one or more of the IMD 102 or other portable medical devices. In some examples, the therapy may include cardiac resynchronization therapy to correct dyssynchrony and improve cardiac function in a heart failure patient. In other examples, the IMD 102 may include a drug delivery system, such as an infusion pump, to deliver drugs to the patient to manage arrhythmia or complications of arrhythmia, high blood pressure, or one or more other physiological conditions. In other examples, the IMD 102 may include one or more electrodes configured to stimulate the patient's nervous system, to provide stimulation to muscles in the patient's airway, or the like.

[0043] WMD103 may include one or more wearable or external medical sensors or devices (e.g., an automated external defibrillator (AED), a Holter monitor, a patch-based device, a smart watch, a smart accessory, a wrist- or finger-worn medical device such as a finger-based photoplethysmography sensor, etc.).

[0044] The external system 105 may include a dedicated hardware / software system, such as a programmer, a remote server-based patient management system, or alternatively, a system defined primarily by software running on a standard personal computer. The external system 105 may manage the patient 101 through the IMD 102 or one or more other portable medical devices connected to the external system 105 via a communication link 111. In other examples, the IMD 102 may be connected to the WMD 103 via the communication link 111, or the WMD 103 may be connected to the external system 105 via the communication link 111. This may include, for example, programming the IMD 102 to perform one or more of acquiring physiological data, performing at least one self-diagnostic test (e.g., for device operation status), analyzing the physiological data, or optionally delivering or adjusting therapy for the patient 101. The external system 105 may also send information to or receive information from the IMD 102 or WMD 103 via the communication link 111. Examples of this information include real-time or stored physiological data from the patient 101, diagnostic data such as detection of the patient's hydration status, hospital admission information, response to therapy delivered to the patient 101, or device operational status of the IMD 102 or WMD 103 (e.g., battery status, lead impedance, etc.). The communication link 111 may be an inductive telemetry link, a capacitive telemetry link, a radio frequency (RF) telemetry link, or wireless telemetry based, for example, on the "strong" Bluetooth® or IEEE 802.11 wireless fidelity "Wi-Fi®" interface standards. Other configurations and combinations of patient data source interfaces are possible.

[0045] The external system 105 may include an external device 106 in proximity to one or more portable medical devices and a remote device 108 located relatively remotely from the one or more portable medical devices and communicating with the external device 106 via a communication network 107. An example of the external device 106 may include a medical device programmer. The remote device 108 may be configured to evaluate collected patient information and provide alert notifications, among other possible functions. In one example, the remote device 108 may include a centralized server that serves as a central hub for storage and analysis of collected data. The server may be configured as a single, multiple, or distributed computing and processing system. The remote device 108 may receive data from multiple patients. The data may be collected by one or more portable medical devices, among other data acquisition sensors or devices associated with the patient 101. The server may include a memory device for storing data in a patient database. The server may include alert analysis circuitry for evaluating the collected data to determine whether a particular alert condition is met. Satisfaction of an alert condition may trigger the generation of an alert notification, such as by one or more human-perceptible user interfaces. In some examples, the alert condition may alternatively or additionally be evaluated by one or more portable medical devices, such as an implantable medical device. By way of example, the alert notification may include a webpage update, a phone or pager call, an email, an SMS, a text, or an "instant" message, as well as a message to the patient and simultaneous direct notification to emergency services or a clinician. Other alert notifications are possible. The server may include an alert prioritization circuit configured to prioritize the alert notifications. For example, a similarity metric between physiological data associated with the detected physiological event and physiological data associated with previous alerts may be used to prioritize alerts for the detected physiological event.

[0046] The remote device 108 may further include one or more locally configured or remote clients securely connected to the server via the communications network 107. Examples of clients may include personal desktops, notebook computers, mobile devices, or other computing devices. System users, such as clinicians or other qualified medical professionals, may use the clients to securely access stored patient data compiled in a database in the server and select and prioritize patients and alerts for medical provisioning. In addition to generating alert notifications, the remote device 108, including the server and interconnected clients, may also execute a follow-up scheme by sending follow-up requests to one or more portable medical devices or by sending messages or other communications as compliance notifications to the patient 101 (e.g., the patient), a clinician, or an authorized third party.

[0047] Communications network 107 may provide wired or wireless interconnectivity. In one example, communications network 107 may be based on the Transmission Control Protocol / Internet Protocol (TCP / IP) network communications specification, although other types or combinations of networking implementations are possible. Similarly, other network topologies and configurations are possible.

[0048] One or more of the external device 106 or remote device 108 may output the detected physiological event to a system user, such as a patient or clinician, or to a process including, for example, an instance of a computer program executable within a microprocessor. In one example, the process may include the automatic generation of an antiarrhythmic therapy recommendation or a recommendation for further diagnostic testing or treatment. In one example, the external device 106 or remote device 108 may include a respective display unit for displaying a physiological or functional signal, an alert, an alarm, an emergency call, or other form of warning to indicate the detection of an arrhythmia. In some examples, the external system 105 may include an external data processor configured to analyze the physiological or functional signals received by one or more portable medical devices and confirm or reject the detection of an arrhythmia. Computationally intensive algorithms, such as machine learning algorithms, may be implemented in the external data processor to retrospectively process data to detect cardiac arrhythmias.

[0049] Portions of one or more portable medical devices or external systems 105 may be implemented using hardware, software, firmware, or a combination thereof. Portions of one or more portable medical devices or external systems 105 may be implemented using application-specific circuitry that can be built or configured to perform one or more functions, or may be implemented using general-purpose circuitry that can be programmed or configured to perform one or more functions. Such general-purpose circuitry may include a microprocessor or portion thereof, a microcontroller or portion thereof, programmable logic circuitry, memory circuitry, a network interface, and various components for interconnecting these components. For example, a “comparator” may include an electronic circuit comparator that can be built to perform, among other things, the specific function of comparing two signals. Alternatively, a comparator may be implemented as part of a general-purpose circuit that can be driven by code that instructs the portion of the general-purpose circuit to perform a comparison between two signals. A “sensor” may include an electronic circuit configured to receive information and provide an electronic output representative of the received information.

[0050] The therapy device 110 may be configured to send information to or receive information from one or more of the portable medical devices or external systems 105 using the communication link 111. In one example, one or more portable medical devices, external devices 106, or remote devices 108 may be configured to control one or more parameters of the therapy device 110. The external system 105 may be capable of programming one or more portable medical devices and may receive information regarding one or more signals acquired by the one or more portable medical devices, such as received via the communication link 111. The external system 105 may include a local external implantable medical device programmer. The external system 105 may include a remote patient management system that may monitor a patient's condition, adjust one or more therapies, or the like, from a remote location.

[0051] 2 schematically illustrates an example of a physiological event management system 200 configured to cluster physiological events to improve event presentation and adjudication. At least a portion of system 200 may be implemented within an external system 105, such as one or more of external device 106 or remote device 108, or distributed between external system 105 and one or more of IMD 102 or WMD 103.

[0052] System 200 may include one or more of a memory circuit 210, a controller circuit 220, a user interface 230, and an optional therapy circuit 250. Memory circuit 210 may be included in a storage device of external system 105, such as in external device 106 or remote device 108. Alternatively, memory circuit 210 may be included in an electronic medical record (EMR) system. Memory circuit 210 may store patient physiological event occurrences detected and recorded by IMD 102 or WMD 103 in a physiological event bank 212. In one example, memory circuit 210 may be communicatively coupled to IMD 102 or WMD 103 and may receive physiological event occurrences from IMD 102 or WMD 103 via communication link 115, etc., as described with reference to FIG. 3 .

[0053] In one example, the physiological event occurrences stored in the physiological event bank 212 may include cardiac events, such as cardiac arrhythmia episodes, detected and recorded by the IMD 102 or WMD 103. Examples of cardiac arrhythmia episodes may include, among others, atrial arrhythmia episodes, supraventricular arrhythmia episodes, or ventricular arrhythmia episodes. The cardiac arrhythmia episodes may include respective physiological data sensed from one or more physiological sensors during the detected arrhythmia event, or additional physiological data sensed before and / or after the detected arrhythmia event. The physiological data associated with the arrhythmia episode may include cardiac electrical signals, such as one or more electrocardiogram (ECG) or electrogram (EGM) signals. The cardiac electrical signals are sensed at various cardiac sites using different electrode combinations, such as one or more atrial EGMs or one or more ventricular EGMs. Additionally or alternatively, the physiological data may include cardiac mechanical or hemodynamic signals, such as cardiac pressure signals, impedance signals, heart sound signals, among others. In various examples, each cardiac arrhythmia episode may further include an arrhythmia detection or classification generated by a medical device, such as the IMD 102 or WMD 103. The arrhythmia detection or classification is a designation of a particular arrhythmia type, such as atrial fibrillation, atrial flutter, ventricular tachycardia, or ventricular fibrillation, among others. Other information about the cardiac arrhythmia episode, such as measurements or signal metrics obtained from physiological data (e.g., atrial rate, ventricular rate, atrial or ventricular rate variability), may also be associated with each episode and stored in the physiological event bank 212.

[0054] In some examples, the physiological event bank 212 may include patient-triggered bouts, which include physiological data sensed from one or more physiological sensors in response to a patient trigger, such as when a patient experiences a physiological event. Other information, such as patient input about the presence of a physiological event and symptom severity, symptom timing information, such as start and end times of the patient-triggered bout, may also be associated with the patient-triggered bouts and included in the physiological event bank 212.

[0055] Although the description of physiological event management herein focuses on arrhythmia episodes, this is meant to be exemplary and not restrictive in nature, and episodes of other types of physiological events, such as syncope, heart failure worsening events, or heart failure decompensation events, may also be stored, analyzed, and provided to a clinician for assessment using the systems, devices, and methods described herein.

[0056] The controller circuit 220 may be implemented as part of a microprocessor circuit, which may be a special-purpose processor such as a digital signal processor, an application specific integrated circuit (ASIC), a microprocessor, or other type of processor for processing information, including physical activity information. Alternatively, the microprocessor circuit may be a general-purpose processor that can receive and execute sets of instructions that implement the functions, methods, or techniques described herein.

[0057] The controller circuit 220 may include a circuit set with one or more other circuits or subcircuits, including a feature extraction circuit 221, a clustering circuit 222, a display control circuit 225, and a determination circuit 226. The circuits or subcircuits, alone or in combination, may implement the functions, methods, or techniques described herein. In one example, the hardware of the circuit set may be invariably designed (e.g., hardwired) to perform a specific operation. In one example, the hardware of the circuit set may include variably connected physical components (e.g., execution units, transistors, simple circuits, etc.) that include computer-readable media physically modified (e.g., magnetically, electrically, a movable arrangement of particles of invariable mass, etc.) to encode instructions for a specific operation. When connecting the physical components, the basic electrical properties of the hardware components are changed, for example, from insulator to conductor or vice versa. The instructions enable the embedded hardware (e.g., an execution unit or loading mechanism) to create the members of the circuit set within the hardware through the variable connections to perform some of the specific operations during operation. Thus, the computer-readable medium is communicatively coupled to other components of the circuit set members when the device is operating. In one example, any of the physical components may be used in more than one member of more than one circuit set. For example, during operation, an execution unit may be used by a first circuit of a first circuit set at one time and reused by a second circuit of the first circuit set, or reused by a third circuit of the second circuit set at a different time.

[0058] The feature extraction circuit 221 extracts N events {X1, X2, . . . , X N}, and extracting a respective feature set (e.g., event X) from each of the received events. k For a set of M features {S1,S2,...,S M}) may be extracted by the feature extraction circuit 221 from each received arrhythmia episode. In one example, the received information of multiple physiological events may include physiological data (e.g., ECG, EGM, or other sensor signals) recorded during (and optionally before and / or after) a cardiac arrhythmia episode detected by a portable device such as the IMD 102 or WMD 103. Features (e.g., M features {S1, S2, . . . , S M The feature extraction circuit 221 may extract one or more of the following: a heart rate distribution feature, an arrhythmia duration, or a morphology feature. The heart rate distribution feature is an important characterization of an atrial or ventricular tachyarrhythmia and may be represented by the amount of heart beats (or the cumulative duration of those beats) at each of multiple heart rates or multiple heart rate ranges during a specified period, or a percentage of the total heart beats (or a percentage of the total heart beat duration). The arrhythmia duration represents the duration from the start to the end of the arrhythmia detected by the IMD 102 or WMD 103. The morphology feature includes signal waveform characteristics such as slope, peak amplitude, curvature, area under the curve, or a similarity metric (e.g., correlation) between a portion of the episode signal and a predetermined morphology template. In some examples, the features extracted by the feature extraction circuit 221 from each received arrhythmia episode may include time information of the arrhythmia episode, such as the time the arrhythmia episode was detected and recorded by the IMD 102 or WMD 103.

[0059] The clustering circuit 222 clusters N events {X1, X2, . . . , X N} for each of M features {S1, S2, . . . , S M}, the received physiological events may be clustered into different event groups 223. The extracted feature set may be a set of M features {S1, S2, . . . , S M} may form a feature space S, such as an M-dimensional space spanned by the physiological events. Each physiological event may be represented as a data point in the feature space S. In one example, the clustering may be performed using an unsupervised clustering algorithm. To cluster multiple physiological events, the clustering circuit 222 may initialize one or more clusters, each including, for example, randomly selected physiological events. k ) may be assigned to one of a plurality of event groups 223 based on a similarity metric between the physiological event and the cluster center. Examples of similarity metrics may include a distance measure (e.g., Euclidean distance), connectivity, or strength between the feature set of each physiological event and the cluster center in the feature space S. A larger distance indicates less similarity between the physiological event and the cluster center. The cluster center may be iteratively updated using the central tendency (e.g., mean or median) of the feature sets extracted from each physiological event assigned to that event group until a convergence criterion or iterative stopping criterion is met.

[0060] Various clustering algorithms may be used to cluster the received physiological events. Clustering algorithms may differ in the set of rules for defining similarity between data points in feature space. One category of clustering algorithms is based on a centroid model, which iteratively establishes clustered event groups using a similarity metric derived by the proximity of data points to the cluster's centroid. In one example, the clustering circuit 222 may use a k-means clustering method that minimizes the intra-cluster variance within a set of physiological events within the same clustered event group. When N events are clustered into k groups, the k-means clustering method minimizes the sum of the intra-cluster variances of all k groups. In some examples, a variant of k-means clustering may be used to cluster the N physiological events. In one example, k-medians clustering uses the median (instead of the mean) in each feature dimension to determine the centroid or cluster center of each cluster. In another example, k-medoids uses actual data points (observations) as cluster centers instead of the calculated mean of the data points in the cluster, which is not necessarily the actual input data points. Furthermore, k-medoids minimizes the sum of pairwise dissimilarities instead of the sum of squared Euclidean distances, which may be more robust to noise and outliers than k-means clustering. In yet another example, fuzzy C-means clustering may be used to cluster multiple received physiological events. In contrast to hard clustering (such as k-means clustering), in which a data point is either assigned to a cluster or not, fuzzy C-means clustering is a soft clustering method that assigns each data point a fuzzy degree representing the probability or likelihood that the data point belongs to each of two or more clusters. Fuzzy C-means is a soft clustering technique.In addition to or instead of k-means clustering or its variants, other clustering methods may be used, such as clustering based on statistical distribution models (e.g., Gaussian mixture models), connectivity-based clustering, which defines clusters by the maximum distance required to connect different parts of the cluster, density-based clustering methods, which define clusters as areas of higher density than the rest of the dataset, among others.

[0061] Each of the one or more clustered event groups 223 is a cluster of N received events {X1, X2, . . . , X N}. From each of the clustered event groups 223, the clustering circuit 222 may determine a respective representative event 224 that represents a physiological event in the corresponding event group. In one example, the representative event 224 may be identified as the physiological event in the event group that has the closest distance (e.g., Euclidean distance in the feature space S) to the centroid of the physiological events in that event group. For example, for P physiological events {Y1, Y2, . . . , Y P}, the physiological events {Y1,Y2,···,Y P}, respectively extracting feature sets (e.g., M-dimensional feature sets) from cluster centers Y * can be determined from this cluster center Y * The physiological event in the group with the closest distance to (e.g., Y k ) may be identified as a representative event. An example of clustering physiological events to determine a representative event using a supervised clustering algorithm is described below with respect to FIG. 4A.

[0062] In addition to or as an alternative to automatic, unsupervised learning-based clustering, such as k-means clustering (or variations thereof), in some examples, the clustering circuit 222 may cluster the received plurality of physiological events according to a user-specified set of cluster centers and clustering rules. In one example, a user may specify a set of physiological events (e.g., N events {X1, X2, . . . , X N} selected X k ) as cluster centers. Clustering circuit 222 may use each feature set to calculate a similarity metric (e.g., Euclidean distance in feature space S) between the user-specified cluster centers and each of the N physiological events. Clustering circuit 222 may then establish clustered event groups by assigning to those physiological events respective similarity metrics that meet a similarity criterion (e.g., respective Euclidean distances to the cluster centers being below a threshold). An example of clustering physiological events using user-specified cluster centers and clustering criteria is described below with respect to FIG. 4B.

[0063] The display control circuitry 225 may prioritize the display of the set of physiological events within each event group 223 on the display unit of the user interface 230. In one example, at least a portion of the user interface 230 may be implemented within the external system 105. The displayed physiological event information may include physiological data (e.g., ECG, EGM, or other sensor signals) recorded during (and optionally before and / or after) the physiological event, device parameters used to detect the physiological event, and a summary of the detection results, including, for example, the detected event type and statistics. In one example, the prioritized display of event information may include automatically displaying information about a representative event of an event group while hiding information about other physiological events (other than the representative event) in the event group until or unless requested by the user. In one example, the display control circuitry 225 may provide a notification to the user (e.g., a message displayed on the user interface 230) indicating that the representative event represents a set of physiological events in the event group, without automatically displaying information about individual physiological events other than the representative event, and may prompt the user to provide a command to display information about any of the physiological events in the event group. In response to a user command, display control circuitry 225 may initiate an on-demand display of information related to one or more physiological events within the event group. In one example, the user command may include a user selection of a physiological event from within the event group. In response to this user selection, display control circuitry 225 may display information related to the selected physiological event.

[0064] In some examples, the control circuitry 220 can prioritize the storage of the set of physiological events within each event group 223 to storage (e.g., the internal memory 314 or the external memory 322 shown in FIG. 3 ) so that higher priority events can be stored in storage before lower priority events within the same group and / or so that more system resources (e.g., memory space) can be allocated to storing higher priority events than lower priority events. In one example, the representative event of each event group can be given higher priority to be stored in storage than other events within each event group. To conserve memory, in some examples, only the representative event of each event group can be stored in storage, and other events within each event group can be discarded (i.e., not stored). In some examples, within a clustered event group, events that are dissimilar to the representative event (e.g., “dissimilar” events that have a similarity value below a threshold or are further away from the representative event or cluster center in the feature space S) can have a higher priority than events that are similar to the representative event (e.g., “similar” events that have a similarity value above a threshold or are closer to the representative event or cluster center in the feature space S). Because “similar” events are better represented by a representative event than “dissimilar” events, prioritizing “dissimilar” events over “similar” events allows the “dissimilar” events to be saved, presented to a user, reviewed, and judged, thereby reducing the error rate associated with assigning the same judgment result of a representative event to all other events in the same event group. In some examples, high-priority events may be automatically stored in storage, while lower-priority events may be stored on-demand (i.e., upon user request). In some examples, prioritization of events across different clustered event groups is performed based on the number of events in each event group, allowing various event types to be reviewed and judged. For example, between a “dense” event group having a large number of events and a “sparse” event group having a relatively small number of events, the more events in the “dense” event group may be assigned a higher priority than those in the “sparse” event group.The cluster-based prioritized display and storage of physiological events described herein advantageously makes more efficient use of system resources and improves the functionality of computer systems for managing physiological events.

[0065] The determination circuit 226 may receive a user determination of a physiological event within the clustered event group 223. The user may provide the determination via the user interface 230. The determination may include a user designation of the event type of the physiological event being presented to the user. In an example of arrhythmia determination, the user may identify an arrhythmia type, such as atrial fibrillation, atrial flutter, supraventricular tachycardia, ventricular tachycardia, or ventricular fibrillation. The determination circuit 226 may compare the user-identified event type with the device detection results generated by the IMD 102 or WMD 103 and identify the episode as a true positive (TP) (or appropriate) detection if the user's determination matches the device detection result, or as a false positive (FP) (or inappropriate) detection if the user's determination differs from the device detection result. If a TP or FP determination cannot be made, the episode may be identified as indeterminate. The user may also provide annotations regarding the episode characteristics or diagnostic information of the physiological event episode. For example, when assessing an AF episode detected by a device, a user (e.g., a clinician) may specify one of a plurality of episode characteristics as the basis for forming the user's assessment decision. An example of a user interface for receiving a physiological event assessment or annotation is described below with reference to FIG.

[0066] In some examples, the determination circuit 226 may determine a physiological event in a clustered event group when information about the event (e.g., ECG, EGM, or other sensor data) is displayed on a user interface. According to the priority display of physiological events in the clustered event group, the determination circuit 226 may prioritize event determination in the order of the displayed events. For example, if a representative event is automatically displayed before other events in the same event group, the representative event may be determined before other physiological events displayed on demand. In some examples, upon determining the representative event, the determination circuit 226 may assign the same determination result of the representative event to each of the physiological events in the event group. Thus, by default, all physiological events in the event group may have the same event type as the determined representative event unless otherwise modified (e.g., by a user). In one example, in response to on-demand display of a physiological event selected from a clustered event group, the determination circuit 226 may initiate on-demand determination of the physiological event, allowing the user to confirm, reject, or modify the default determination result. The cluster-based priority display and priority determination described herein not only improves the functionality of computer systems for managing physiological events, but also reduces clinician workload and improves the efficiency of physiological event review and determination.

[0067] In some examples, as more additional physiological events are detected from the patient and added to the event bank 212, the controller circuit 220 may update previously established clusters using information about those additional physiological events. Cluster updates may be performed periodically or in response to the amount of received additional events reaching a threshold. In some examples, the controller circuit 220 may initiate a process to update the clusters based on decision performance. Decision performance may be measured by an estimate of the workload reduction associated with assigning the same decision result of a representative event to all other physiological events in the same event group and an estimate of the decision error rate. In one example, the controller circuit 220 may generate an estimate of the decision error rate based on on-demand decisions of physiological events (other than the representative event) in the clustered event group. If the estimated error rate exceeds an error tolerance level, the controller circuit 220 may update the clusters, for example, by restarting the clustering process using an updated set of received arrhythmia episodes or by adjusting one or more clustering criteria (e.g., a similarity metric or distance threshold that defines the clusters). In another example, the controller circuit 220 may determine a performance metric proportional to the amount of workload reduction and inversely proportional to the decision error rate, such that a higher workload reduction increases the performance metric and a higher decision error decreases the performance metric. The controller circuit 220 may update the clusters when the performance metric falls below a performance threshold. Updating the clusters may include updating the representative events of each cluster. In some examples, the aggressiveness of adjusting one or more clustering criteria (e.g., a similarity metric threshold that defines the size or boundaries of clustered event groups) may be determined based on the deviation of the performance metric from the performance threshold. A more significant decrease in decision performance may result in a more aggressive adjustment (e.g., a larger reduction in the similarity threshold).

[0068] The user interface 230 may include a display unit controllable by the display control circuitry 225 to display information about the physiological event, as described above, and a user input device for receiving user input, such as a determination of the physiological event displayed on the display unit. The user interface 230 may be coupled to a printer for printing a hard copy of the detected information. This information may be presented in a table, chart, diagram, or any other type of textual, tabular, or graphical presentation format. The presentation of output information may include audio or other media formats. In one example, the output unit may generate an alert, alarm, emergency call, or other form of warning to notify the user about the detected physiological event. In some examples, the user interface 230 may further generate recommendations for adjusting the operation of the IMD 102 or WMD 103 based on the determination of the physiological event, such as recommended adjustments to parameters for sensing physiological signals or parameters for detecting the physiological event.

[0069] Optional therapy circuitry 250 may be configured to deliver therapy to the patient in response to detection of a physiological event of interest. Examples of therapy may include electrical stimulation therapy delivered to the heart, neural tissue, other target tissue, cardioversion therapy, defibrillation therapy, or drug therapy including delivering a drug to a tissue or organ. In some examples, therapy circuitry 250 may modify an existing therapy, such as adjusting stimulation parameters or drug dosage.

[0070] FIG. 3 schematically illustrates an example of a cardiac event management system 300 configured to evaluate and prioritize cardiac events detected by a medical device associated with a patient. The system 300 includes an ambulatory medical device (AMD) 310 and an external system 320 communicatively coupled to each other via a communication link 115. The AMD 310, which is an embodiment of the IMD 102 or WMD 103 shown in FIG. 1, may include a sensor circuit 311 for sensing one or more physiological signals from a subject. The physiological signals may be sensed via one or more implantable, wearable, or portable sensors or electrodes associated with the patient. The sensors may be built into or otherwise associated with the portable device, such as the IMD 102 or WMD 103. Examples of physiological signals may include, among others, a surface electrocardiogram (ECG) sensed from electrodes placed on the body surface, a subcutaneous ECG sensed from electrodes placed under the skin, an intracardiac electrogram (EGM) sensed from one or more electrodes on a lead system, a thoracic or cardiac impedance signal, an arterial pressure signal, a pulmonary artery pressure signal, a left atrial pressure signal, an RV pressure signal, an LV coronary artery pressure signal, a coronary artery blood temperature signal, a blood oxygen saturation signal, a heart sound signal sensed by an ambulatory accelerometer or acoustic sensor, a physiological response to activity, an apnea-hypopnea index, one or more respiratory signals such as a respiratory rate signal or a tidal volume signal, brain natriuretic peptide (BNP), a blood panel, sodium and potassium levels, glucose levels, and other biomarkers and biochemical markers. The sensor circuit 311 may include one or more subcircuits for performing digitization, filtering, or other signal conditioning operations on the received physiological signals. In some examples, the sensor circuit 311 may register a patient-triggered seizure. When a patient exhibits a particular sign or symptom or experiences a precursor event indicative of a target physiological event, a trigger may be generated and detected by the patient trigger detector. When a patient trigger is detected, the sensor circuit 311 is activated, causing the sensor circuit 311 to register a patient-triggered seizure and acquire physiological data, such as one or more physiological signals.

[0071] The arrhythmia detection circuit 312 may be configured to detect cardiac arrhythmia episodes using one or more sensed physiological signals. Examples of cardiac arrhythmias include atrial fibrillation, atrial flutter, supraventricular tachycardia, ventricular tachycardia, or ventricular fibrillation. Detection may be based on temporal or morphological features extracted from one or more physiological signals. In one example, the arrhythmia detection circuit 312 may detect a particular cardiac arrhythmia using a configurable detection algorithm, such that one or more detection algorithm features may be added or modified via a programming device. Detected arrhythmia episodes may be stored in the internal memory 314, including the detection results along with physiological data collected during detection or, in addition, physiological data collected before and / or after detection.

[0072] The communications circuit 315 may transmit the detected arrhythmia episodes (including the physiological data and device-generated detection results) to the external system 320 via the communications link 115. This transmission may be performed continuously, periodically at scheduled times, or in response to a data query command sent from the external system 320 to the AMD 310. The external system 320, an embodiment of the external system 105 shown in FIG. 1, may receive the arrhythmia episodes via the communications circuit 325 and store the received arrhythmia episodes in an external memory 322. The external memory 322, an embodiment of the memory circuit 210, may store the arrhythmia episodes in the physiological event bank 212. The external system 320 may also include a controller circuit 220 and a user interface 230. The controller circuit 220 may schedule the presentation of one or more arrhythmia episodes for determination. As described above with reference to FIG. 2 , the controller circuit 220 can cluster the received arrhythmia episodes into multiple different event groups, each including a set of episodes represented by a representative episode in the corresponding event group, and prioritize the display of the arrhythmia episodes within the clustered event groups. The user interface 230 can initiate automatic display of information (e.g., physiological data and AMD detection results) about the representative episode in the clustered event group and, in response to a user command, initiate on-demand display of information about other arrhythmia episodes in the clustered event group. The user interface 230 can also receive a user-determined arrhythmia type for the representative episode. By default, all arrhythmia episodes within an event group will have the same user-determined arrhythmia type as the representative event unless modified by the user through the on-demand determination process, as described above with reference to FIG. 2 . In some examples, the controller circuit 220 can assign an arrhythmia episode received from the AMD 310 to one of the established clustered event groups, for example, based on its similarity (e.g., distance in the feature space S) to each of the representative episodes in the respective clustered event group.An assigned arrhythmia episode may by default have the same determined arrhythmia type as the representative episode unless or until it is modified by the user via an on-demand display and determination process.

[0073] The external system 320 includes a programmer 324 that can generate commands for programming the AMD 310. The commands may include, among other things, recommended adjustments to one or more detection parameters of the arrhythmia detection circuit 312 or data collection parameters of the sensor circuit 311. The recommended adjustments may be confirmed or modified by a user (e.g., a clinician) via the user interface 230 and transferred to the AMD 310 via the communication link 115. The AMD 310 may use the modified algorithm to detect physiological events of interest.

[0074] 3 shows the controller circuit 220 included in the external system 320, but this is by way of example and not limitation. In some examples, at least a portion of the controller circuit 220 (such as the feature extraction circuit 221 and the clustering circuit 222) may be implemented within the AMD 310, while some other portions of the controller circuit 220 (such as the display control circuit 225 and the determination circuit 226) may be implemented within the external system 320. In the AMD 310, arrhythmia episodes detected by the arrhythmia detection circuit 312 and / or stored in the internal memory 314 may be processed by the feature extraction circuit 221 and the clustering circuit 222 and clustered into one or more clustered event groups. The clustered arrhythmia episodes may be stored in the internal memory 314 and transmitted to the external system 320 via the communication link 115, either automatically or upon user request.

[0075] Storage and / or transmission of seizure data may be prioritized according to seizure clustering. Higher priority seizures may be stored in storage (e.g., internal memory 314 or external memory 322) and / or transmitted between devices (e.g., from portable medical device 310 to external system 320) before lower priority seizures within the same group. Additionally or alternatively, more system resources (e.g., memory space and / or communications bandwidth) may be allocated to storing and / or transmitting higher priority seizures before storing and / or transmitting lower priority seizures. In one example, a representative seizure of each event group may have a higher priority than other seizures within each event group. In some examples, within clustered event groups, a higher priority may be assigned to “dissimilar” seizures (having a similarity value below a threshold) than to “similar” events (those with a similarity value above a threshold). Prioritizing "dissimilar" events relative to "similar" events allows the "dissimilar" events to be saved, presented to the user, reviewed, and judged, thereby reducing the error rate associated with assigning the same judgment of a representative event to all other events in the same event group. In some examples, prioritization of seizures across different clustered event groups is based on the number of events in each event group, allowing various event types to be reviewed and judged. In some examples, high-priority events may be automatically stored in storage and / or transmitted between devices, while low-priority events may be stored and / or transmitted on-demand (i.e., when requested by the user). To conserve system resources (e.g., memory space and / or communication bandwidth), in some examples, only representative seizures from each event group may be stored and / or transmitted, and other events within each event group may be discarded.

[0076] In the external system 320, the display control circuitry 225 may prioritize display of arrhythmia episodes of clustered event groups, such as automatically displaying information about the representative episode, while hiding other events in the event group until or unless requested by the user (i.e., on-demand display of other events). The determination circuitry 226 may receive a user-determined arrhythmia type for the representative episode of the event group. The programmer 324 may generate commands for programming the AMD 310.

[0077] 4A and 4B illustrate an example of clustering physiological events in a feature space S. Each physiological event is represented by a data point in a graphical representation of the feature space S. The coordinates of the data point correspond to the respective values ​​of a set of features extracted from the corresponding physiological event. By way of example and not limitation, each physiological event is represented by a data point in a two-dimensional feature space spanned by features S1 and S2, as shown in FIGS. 4A and 4B. In an example of clustering cardiac arrhythmia episodes detected by a medical device from a patient, the feature set may include heart rate distribution features, arrhythmia duration, morphological features, or temporal features such as the time at which the arrhythmia episode was detected. The diagrams shown in FIGS. 4A and 4B may each be displayed on a display unit of the user interface 230.

[0078] FIG. 4A illustrates automated clustering of physiological events using an unsupervised clustering algorithm, such as k-means clustering or a variation thereof, as described above with reference to FIG. 2. As shown in diagram 400A, three clustered event groups 410, 420, and 430 are formed, each including a respective set of physiological events and a cluster center (e.g., centroid) representing all physiological events in that event group. For example, clustered event group 420 has cluster center 421. Clustering circuit 222 may determine a representative event 422 from event group 420 that is closest to cluster center 421. Representative event 422 and an indication that representative event 422 represents a physiological event in event group 420 may then be displayed on user interface 230.

[0079] Information about the representative event 422 may be automatically displayed on the user interface 230, and a user determination of the representative event 422 may be received from a user. FIG. 5 illustrates at least a portion of a user interface 500 for displaying information about a physiological event and receiving a user determination of the event, including an example of an embodiment of a display unit of the user interface 230. By way of example and not limitation, the information may include physiological data 512 and trends in measurements 514 obtained from the physiological data 512. Information regarding patient identification and stroke number may also be displayed, and this information may be selected using, among other input control elements on the user interface 400, such as drop-down lists, check boxes, radio buttons, list boxes, buttons, toggles, text fields, etc. The physiological data 512 may be sensed using electrodes or physiological sensors in communication with a medical device and collected during, before, or after the detected physiological event. By way of non-limiting example, as shown in FIG. 5, the physiological data 512 may include an ECG or EGM sensed at a cardiac site, such as a ventricle. In some examples, two or more physiological signals may be displayed, including EGMs from multiple cardiac sites or via different sensing electrode configurations, cardiac mechanical signals, or hemodynamic signals sensed from one or more sensors. Trends in measurements 514 may include intra-channel or inter-channel timing information, such as heart rate trends, heart rate variability trends, or atrioventricular interval trends. The displayed information may include a detection summary 520. By way of example, the detection summary 520 may include physiological event types detected by the AMD and measurements from the sensed physiological signals acquired by the AMD. In the example shown in FIG. 5, the displayed episode is detected as an AF episode with a ventricular rate of 170 bpm and a duration of 140 seconds.

[0080] The user interface section 500 may include a display zone for receiving user annotations 530 of the displayed episode, such as the determination of the event type (e.g., a particular arrhythmia type), i.e., a designation of a proper (e.g., true positive) detection indicating a user match with the arrhythmia type detected by the device, or an inappropriate (e.g., false positive) detection indicating a user mismatch with the arrhythmia type detected by the device.

[0081] 4A , in addition to the automatic display of information about representative event 422, information about other events in the same event group, such as event 423, may be displayed in response to a user command. As shown, a user may use a pointing device to select an event (e.g., event 423) in the clustered event group, as indicated by mouse pointer 440 on the user interface. In response to that user selection, information about the selected event 423 may be displayed in a manner similar to the automatic display of representative event 422 as shown in FIG. 5. In response to the on-demand display of event 423, the user may be prompted to provide an on-demand judgment or annotation about the selected event 423.

[0082] As described above with reference to FIG. 2 , once the representative event 422 is determined, the determination circuit 226 may assign the same determination result of the representative event to each physiological event in the event group. The cluster-based prioritized display and prioritized determination of physiological events described herein advantageously utilizes system resources more efficiently and improves the computer system's ability to manage physiological events. It may also reduce clinician workload and improve the clinician's efficiency in reviewing and determining physiological events. In some examples, the user may further identify a subset of the physiological events in the event group that is less than all of the physiological events. The identified subset includes physiological events whose respective similarity metrics are less than a similarity threshold. The physiological events in the identified subset may be assigned the same event type as the representative event. For example, the user may identify a boundary region 425 in the feature space S that encompasses the representative event 422. The boundary region 425 may have a user-defined boundary. In the example shown in FIG. 4A , the user may define the boundary region 425 with a radius “r” centered on the representative event 422. The radius "r" may be adjustable by the user via a user interface. In one example, the user may select from multiple predefined regions of different sizes or shapes centered on the representative event 422. In one example, the user may select from multiple predetermined radii, such as one of a "large," "medium," or "small" radius having a predetermined value. Alternatively, the user may select from multiple predefined regions each representing a different density level of physiological events in the respective clustered event group. In one example, the predefined regions may include a "dense cluster" (a small region with densely clustered events in feature space), a "sparse cluster" (a large region with sparsely distributed events in feature space), or any intermediate density level (e.g., a moderately dense cluster, a moderately sparse cluster).

[0083] Because the radius "r" represents a similarity threshold for the representative event 422, the selection of the radius "r" can affect decision performance, as measured by the workload reduction and decision error rate associated with assigning the same decision result of the representative event to all other physiological events within the user-defined subgroup. FIG. 6 shows examples of workload reduction curves 610 and decision error rate curves 620 for different radius values ​​corresponding to user-defined cluster event regions. The workload reduction curve 610 shows the relationship between the workload reduction (for reviewing and deciding on events) and the radius value. The decision error rate curve 620 shows the decision error rate at different radius values. When a small radius "r" is used, more physiological events are excluded from the boundary region 425 and are not assigned the default decision (the same decision result as the representative event 422). Decisions of physiological events outside such regions increase the clinician's workload (i.e., the workload reduction is small). On the other hand, events within the boundary region 425 defined by a smaller radius "r" are more similar (or closer) to the representative event 422, so the default decision (the same decision result as the representative event 422) is more likely to be accurate, and a lower decision error rate can be expected. Conversely, a larger radius "r" results in a larger boundary region 425, and more physiological events are assigned the default decision (the same decision result as the representative event 422). This may reduce the workload but potentially increase the decision error rate. As an example shown in FIG. 6, when the radius "r" increases from 20 to 30, the workload reduction increases from approximately 20% (data point 612) to 35% (data point 614), while the decision error rate increases from approximately 1% (data point 622) to 3% (data point 624).

[0084] In one example, a user may provide an expected amount of workload reduction or error tolerance level associated with assigning the same decision outcome as representative event 422 to all events within boundary region 425, and controller circuit 220 may determine radius "r" based on the expected amount of workload reduction or error tolerance level. In another example, controller circuit 220 may determine a performance metric proportional to the amount of workload reduction and inversely proportional to the decision error rate, such that higher workload reductions increase the performance metric and higher decision errors decrease the performance metric. Controller circuit 220 may determine an "optimal" radius corresponding to a performance metric that exceeds a performance threshold.

[0085] FIG. 4B illustrates clustering physiological events using user-specified cluster centers and clustering criteria. As shown in diagram 400B, a user may designate a physiological event (e.g., event 452) as the center or representative event of a clustered event group that has not yet been established. In one example, a user may determine the designated representative event (e.g., event 452) before clustering the events. To establish the clustered event group, the clustering circuit 222 may establish the clustered event group by calculating a similarity metric between each of the physiological events and representative event 452 using the respective feature sets and assigning the physiological events respective similarity metrics that meet the similarity criteria. In one example, the similarity metric may include Euclidean distance on feature S, and the clustered event group may be established to include physiological events within a bounding region 455 of radius “r” centered on representative event 452. The radius “r” may be adjustable by the user via a user interface. In one example, a user may select one of a plurality of predetermined radii, such as one of a "large," "medium," or "small" radius having a predetermined value. Alternatively, a user may select from a plurality of predefined regions identified as a "dense cluster" (a small region with densely clustered events in feature space), a "sparse cluster" (a large region with sparsely distributed events in feature space), or any intermediate density level (e.g., a moderately dense cluster, a moderately sparse cluster).

[0086] Information about representative event 452 may be automatically displayed on user interface 230. As shown in Figure 5, a user judgment of representative event 452 may be received from a user. As described above with reference to Figure 4A, information about other events in the same event group, such as event 453, may similarly be displayed in response to a user command. In response to the on-demand display of event 453, the user may be prompted to provide an on-demand judgment or annotation about the selected event 453.

[0087] Once the representative event 452 is determined, the determination circuit 226 may assign the same determination result of the representative event to each physiological event within the boundary region 455. As described above with reference to FIG. 4A , the radius “r” defining the boundary region 455 may affect the workload reduction and determination error rate associated with assigning the determination result of the representative event 452 to all other physiological events within the user-defined subgroup. As shown in FIG. 6 , a smaller radius “r” results in a smaller workload reduction and a smaller determination error. Conversely, a larger radius “r” results in a greater workload reduction and a higher determination error rate. The user may provide an expected amount of workload reduction or error tolerance level associated with assigning the same determination result of the representative event 452 to all events within the boundary region 455, and the controller circuit 220 may determine the radius “r” based on the expected amount of workload reduction or error tolerance level.

[0088] In one example, the controller circuit 220 determines a value for radius "r" for a given amount of workload reduction (e.g., a 50% reduction in event data to be reviewed) or a set review and decision time (e.g., 15 minutes), and determines an estimated error rate based on the value of radius "r." When the actual error rate (for the set review and decision time) exceeds a threshold error rate, a notification or alert may be generated and presented to the user via the user interface 230. In another example, the controller circuit 220 may determine a value for radius "r" for a given error tolerance rate (e.g., 20%), and determine an estimated amount of workload reduction or an estimated time for review and decision based on the value of radius "r." When the actual review and decision time (for a given error rate) exceeds the threshold review and decision time, a notification or alert may be generated and presented to the user via the user interface 230.

[0089] In another example, the controller circuit 220 may determine a performance metric proportional to the amount of workload reduction and inversely proportional to the decision error rate, such that a higher workload reduction increases the performance metric and a higher decision error decreases the performance metric. The controller circuit 220 may determine an “optimal” radius that corresponds to a performance metric that exceeds a performance threshold.

[0090] 7 is a flowchart illustrating an example of a method 700 for managing the presentation and adjudication of multiple physiological events detected in a patient by a medical device. The physiological events may include, for example, cardiac arrhythmia episodes detected and recorded by an IMD 102 or a WMD 103. The method 700 may be implemented and performed in a portable medical device, such as an implantable or wearable device, or in a remote patient management system. In one example, the method 700 may be implemented in and performed by the IMD 102, the WMD 103, or an external system 105.

[0091] At 710, information regarding multiple physiological events detected from the patient may be received, for example, from the memory circuit 210 that stores patient physiological event episodes detected by the IMD 102 or WMD 103. The cardiac arrhythmia episode may include respective physiological data sensed from one or more physiological sensors during the detected arrhythmia event or additional physiological data sensed before and / or after the detected arrhythmia event. The physiological data associated with the arrhythmia episode may include one or more cardiac electrical signals, such as one or more ECG or EGM signals. In some examples, the physiological event may include a patient-triggered episode.

[0092] At 720, a respective feature set may be generated from the received information for each of the plurality of physiological events, for example, using feature extraction circuit 221. In examples where the received information includes physiological data (e.g., ECG, EGM, or other sensor signals) recorded during a cardiac arrhythmia episode, the feature set may include one or more heart rate distribution features, arrhythmia duration features, morphology features, or temporal features extracted from the arrhythmia episode.

[0093] At 730, the physiological events may be clustered into different event groups based on the feature sets of the physiological events, using, for example, the clustering circuitry 222. As described above with reference to FIG. 2, the extracted feature sets may form a feature space S. Each physiological event may be represented as a data point in the feature space S. The clustering may be based on an unsupervised learning algorithm, such as a k-means cluster, k-medoids, or fuzzy C-means clustering algorithm. The received physiological events may then be clustered into one or more clustered event groups, each including a respective set of physiological events.

[0094] At 740, a respective representative event may be identified from each of the clustered event groups to represent all physiological events in the corresponding event group. In one example, the representative event may be identified as the physiological event in the event group that has the closest distance (e.g., Euclidean distance in feature space S) to the centroid of the physiological events in that event group, as shown in FIG. 4A . In some examples, as an alternative to unsupervised learning-based clustering, the received physiological events may be clustered at 730 according to a set of user-specified cluster centers and clustering rules. For example, a user may designate a physiological event as a cluster center. A similarity metric (e.g., Euclidean distance in feature space S) between the user-specified cluster center and each of the physiological events may be determined using the respective feature sets. The clustered event groups may be established by assigning to the physiological events respective similarity metrics that meet a similarity criterion, such as respective distances to the user-specified cluster center less than a similarity threshold, as described above with reference to FIG. 4B . At 740, the user-specified physiological event may be identified as the representative event. The similarity threshold may be defined or adjusted by a user, for example, via the user interface 230. In one example, a user may define a bounding region in the feature space S around a user-specified physiological event. Physiological events having respective graphically represented feature sets that fall within the user-defined region belong to the same clustered event group. In one example, the user-defined region may have a user-defined radius "r" centered on the user-specified physiological event, as shown in FIG. 4B. The radius "r," which represents the similarity threshold for the representative event, may be user-defined or adjustable. In one example, a user may select from multiple predefined regions of different sizes or shapes centered on the representative event 422, or from multiple predefined radii (e.g., "large," "medium," and "small" radii), each having a predefined value.Alternatively, the user may select from multiple predefined regions identified as "dense clusters" (small regions with tightly clustered events in feature space), "sparse clusters" (large regions with sparsely distributed events in feature space), or any intermediate density level (e.g., moderately dense clusters, moderately sparse clusters).

[0095] The clustering results, including the representative event and an indication that the representative event represents a set of physiological events in the clustered event group, may be output to a user or a process. For example, at 752, the physiological events in the clustered event group may be displayed in prioritized order. For example, information about the representative event (e.g., ECG, EGM, or other recorded sensor data) may be automatically displayed on a user interface, while information about other physiological events in the same event group is hidden until or unless requested by the user. In one example, the event group (e.g., N events) represented by the representative event may be notified to the user. The user may be prompted to select a hidden physiological event in the clustered event group for display. In response to a use selection, information about the selected event may be displayed.

[0096] At 754, a user determination of the displayed physiological events may be received from a user via a user interface. This determination may include a user designation of an event type for the physiological events, as shown in FIG. 5 . Furthermore, the determination of physiological events within the same group may be prioritized according to the prioritized display of the physiological events. For example, an automatically displayed representative event may be determined before other physiological events within the event group that are displayed on demand (e.g., selected by the user). In some examples, upon determination of the representative event, the same determination result as the representative event may be assigned to each physiological event within the event group. Thus, unless otherwise modified (e.g., by the user), all physiological events within the event group will, by default, have the same event type as the determined arrhythmia type of the representative event. The default determination result may be confirmed, rejected, or modified by the user.

[0097] The cluster-based prioritized display and prioritized decision described herein can reduce a clinician's workload and improve the efficiency of physiological event review and decision. The amount of workload reduction can be directly related to clustering criteria, such as a similarity threshold, the size or shape of a user-defined region, or a user-defined radius "r" in feature space, as shown in FIGS. 4A and 4B. On the other hand, assigning the same decision result to all physiological events within the same clustered event group can result in a decision error rate. As shown in FIG. 6, a smaller radius "r" results in less workload reduction and a lower decision error rate, while a larger radius "r" results in more workload reduction but a higher decision error rate. The similarity threshold, or radius "r," can be determined or adjusted to meet user-specified requirements for the amount of workload reduction and / or error tolerance level. In some examples, an "optimal" similarity threshold or radius "r" can be determined to optimize a performance metric proportional to the amount of workload reduction and inversely proportional to the decision error rate.

[0098] 8 illustrates a block diagram of an example machine 800 on which any one or more of the techniques (e.g., methods) described herein may be implemented. Portions of this description may apply to the computing framework of various parts of an LCP device, an IMD, or an external programmer.

[0099] In alternative embodiments, machine 800 may operate as a standalone device or may be connected (e.g., networked) to other machines. In a networked configuration, machine 800 may operate as a server machine, a client machine, or both in a server-client network environment. In one example, machine 800 may operate as a peer machine in a peer-to-peer (P2P) (or other distributed) network environment. Machine 800 may be a personal computer (PC), tablet PC, set-top box (STB), personal digital assistant (PDA), mobile phone, web appliance, network router, switch, or bridge, or any machine capable of executing instructions (sequential or otherwise) that specify operations to be performed by that machine. Furthermore, although only a single machine is shown, the term “machine” should also be interpreted to include any collection of machines that individually or collectively execute a set (or sets) of instructions to perform any one or more of the methodologies described herein, such as cloud computing, software as a service (SaaS), other computer cluster configurations, etc.

[0100] An embodiment may include or operate by logic or several components or mechanisms as described herein. A circuit set is a collection of circuits implemented in tangible entities including hardware (e.g., simple circuits, gates, logic, etc.). Circuit set components may flexibly change over time and in response to variations in the underlying hardware. A circuit set includes components that, when operational, can perform specified operations, either alone or in combination. In one example, the hardware of a circuit set may be invariably designed (e.g., hardwired) to perform specific operations. In one example, the hardware of a circuit set may include variably connected physical components (e.g., execution units, transistors, simple circuits, etc.) including computer-readable media physically modified (e.g., magnetically, electrically, movable arrangements of particles of invariable mass, etc.) to encode instructions for specific operations. When connecting the physical components, the basic electrical properties of the hardware components are changed, for example, from insulator to conductor or vice versa. The instructions enable embedded hardware (e.g., an execution unit or a loading mechanism) to create circuit set members within the hardware via variable connections to perform certain portions of the operations during operation. Thus, the computer-readable medium is communicatively coupled to other components of the circuit set members when the device is operating. In one example, any of the physical components may be used in more than one member in more than one circuit set. For example, during operation, an execution unit may be used by a first circuit of a first circuit set at one time and reused by a second circuit of the first circuit set, or reused by a third circuit of the second circuit set at a different time.

[0101] The machine (e.g., a computer system) 800 may include a hardware processor 802 (e.g., a central processing unit (CPU), a graphics processing unit (GPU), a hardware processor core, or any combination thereof), a main memory 804, and a static memory 806, some or all of which may communicate with each other via an internal link (e.g., a bus) 808. The machine 800 may further include a display unit 810 (e.g., a raster display, a vector display, a holographic display, etc.), an alphanumeric input device 812 (e.g., a keyboard), and a user interface (UI) navigation device 814 (e.g., a mouse). In one example, the display unit 810, the input device 812, and the UI navigation device 814 may be touchscreen displays. The machine 800 may further include a storage device (e.g., a drive unit) 816, a signal generating device 818 (e.g., a speaker), a network interface device 820, and one or more sensors 821, such as a global positioning system (GPS) sensor, a compass, an accelerometer, or other sensor. The machine 800 may include an output controller 828, such as a serial (e.g., universal serial bus (USB)), parallel, or other wired or wireless (e.g., infrared (IR), near field communication (NFC), etc.) connection, for communicating with or controlling one or more peripheral devices (e.g., a printer, a card reader, etc.).

[0102] Storage device 816 may include machine-readable medium 822 having stored thereon one or more sets of data structures or instructions 824 (e.g., software) that embody or are utilized by any one or more of the techniques or functions described herein. Also, instructions 824 may reside, completely or at least partially, within main memory 804, static memory 806, or hardware processor 802 during execution thereof by machine 800. In one example, one or any combination of hardware processor 802, main memory 804, static memory 806, or storage device 816 may constitute a machine-readable medium.

[0103] Although the machine-readable medium 822 is shown as a single medium, the term "machine-readable medium" may include a single medium or multiple media (e.g., a centralized or distributed database and / or associated caches or servers) configured to store one or more instructions 824.

[0104] The term "machine-readable medium" may include any medium that can store, encode, or carry instructions for execution by machine 800, cause machine 800 to perform any one or more of the techniques of this disclosure, or store, encode, or carry data structures used by or related to such instructions. Non-limiting examples of machine-readable media may include solid-state memory, as well as optical and magnetic media. In one example, a high-capacity machine-readable medium includes a machine-readable medium having a plurality of particles with an unchanging (e.g., stationary) mass. Thus, a high-capacity machine-readable medium is not a transitory, propagating signal. Specific examples of high-capacity machine-readable media may include non-volatile memory such as semiconductor memory devices (e.g., electrically programmable read-only memory (EPROM), electrically erasable and programmable read-only memory (EEPROM)) and flash memory devices, magnetic disks such as internal hard disks and removable disks, magneto-optical disks, and CD-ROM and DVD-ROM disks.

[0105] The instructions 824 may further be transmitted or received over a communications network 826 using a transmission medium via a network interface device 820 utilizing any one of several transport protocols (e.g., frame relay, Internet Protocol (IP), Transmission Control Protocol (TCP), User Datagram Protocol (UDP), Hypertext Transfer Protocol (HTTP), etc.). Exemplary communications networks may include local area networks (LANs), wide area networks (WANs), packet data networks (e.g., the Internet), mobile telephone networks (e.g., cellular networks), plain old telephone service (POTS) networks, and wireless data networks (e.g., the IEEE 802.11 family of standards known as Wi-Fi®, the IEEE 802.16 family of standards known as WiMax®), the IEEE 802.15.4 family of standards, peer-to-peer (P2P) networks, etc. In one example, the network interface device 820 may include one or more physical jacks (e.g., Ethernet, coaxial, or telephone jacks) or one or more antennas for connecting to the communications network 826. In one example, network interface device 820 may include multiple antennas for wireless communication using at least one of single-input multiple-output (SIMO), multiple-input multiple-output (MIMO), or multiple-input single-output (MISO) technologies. The term "transmission medium" shall be taken to include any intangible medium capable of storing, encoding, or carrying instructions for execution by machine 800, including digital or analog communication signals or other intangible media for facilitating communication of such software.

[0106] Various embodiments are illustrated in the figures described above, and one or more features from one or more of these embodiments may be combined to form other embodiments. The example methods described herein may be at least partially machine- or computer-implemented. Some examples may include computer-readable or machine-readable media encoded with instructions operable to configure an electronic device or system to perform methods such as those described in the examples above. Implementations of such methods may include code, such as microcode, assembly language code, high-level language code, etc. Such code may include computer-readable instructions for performing various methods. The code may form part of a computer program product. Furthermore, the code may be tangibly stored on one or more volatile or non-volatile computer-readable media during execution or at other times.

[0107] The foregoing detailed description is intended to be illustrative, and not limiting. Accordingly, the scope of the present disclosure should be determined with reference to the appended claims, along with the full scope of equivalents to which such claims are entitled.

Claims

1. 1. A system for managing multiple physiological events detected by a patient's medical device, comprising: a controller circuit, the controller circuit comprising: receiving information regarding the plurality of physiological events; generating, for each of the plurality of physiological events, a respective feature set using the received information, the feature set including values ​​of temporal or morphological features; clustering the plurality of physiological events into a plurality of distinct event groups, each group comprising a respective set of physiological events, using the values ​​of the temporal or morphological features of the generated feature set; determining a representative event from at least one of the plurality of different event groups, the representative event being representative of the set of physiological events within the at least one event group; outputting the representative event and an indication that the representative event has been determined and that the representative event represents the set of physiological events within the at least one event group to a user or process of the system; The system is configured as follows:

2. the controller circuit Initiating automatic display of information about the determined representative event on a user interface; hiding information about the set of physiological events of the at least one event group on the user interface; Initiating an on-demand display of information regarding one or more physiological events in the set of physiological events of the at least one event group in response to a user command via the user interface. The system of claim 1 , configured to:

3. 3. The system of claim 2, wherein the user command includes a user selection of a physiological event from within the set of physiological events of the at least one event group, and the controller circuit is configured to display information about the selected physiological event on the user interface in response to the user selection.

4. a user interface configured to receive a determination result of the representative event from a user; The system of claim 2 , wherein the controller circuit is configured to assign a determination result of the received representative event to each physiological event in the set of physiological events of the at least one event group.

5. 5. The system of claim 1, wherein the controller circuit is configured to prioritize storage of information relating to the determined representative event in a memory device over storage of information relating to the set of physiological events of the at least one event group.

6. 5. The system of claim 1, wherein the controller circuit is configured to prioritize transmission of information relating to the determined representative event over transmission of information relating to the set of physiological events of the at least one event group between different devices.

7. the plurality of physiological events includes cardiac arrhythmia episodes detected in the patient, and the generated feature set comprises: Heart rate distribution features, arrhythmia duration, Morphological characteristics, or the times at which the plurality of physiological events were detected; The system according to any one of claims 1 to 4, comprising at least one of:

8. To determine the representative event, the controller circuit determining a centroid of the set of physiological events of the at least one event group in a feature space using feature sets generated from at least some of the physiological events in the set of physiological events; determining the representative event of the at least one event group as the physiological event closest to the determined centroid; The system according to any one of claims 1 to 4, configured as follows:

9. To cluster the plurality of physiological events, the controller circuitry initializing cluster centers using a feature set of a physiological event among the plurality of physiological events; iteratively updating the cluster centers using central tendencies of feature sets generated from each physiological event in the event group until a convergence criterion or an iteration stopping criterion is met; assigning physiological events to clustered event groups based on a similarity metric between the physiological events and the iteratively updated cluster centers; The system according to any one of claims 1 to 4, configured as follows:

10. the controller circuit calculating a respective similarity metric between the representative event and each physiological event in the set of physiological events of the at least one event group; selecting a subset of events from the set of physiological events of the at least one event group, the subset of events each having the similarity metric less than a similarity threshold; outputting an indication to a user or process of the system that the representative event represents a subset of the selected events; The system of claim 9 , configured to:

11. the controller circuit calculating a respective similarity metric between each of the received physiological events and the user-specified physiological event using feature sets generated from the received physiological events and the user-specified physiological event, respectively; determining a set of the physiological events of the at least one event group having the similarity metric exceeding a similarity threshold; determining the user-specified physiological event as the representative event of the at least one event group; The system according to any one of claims 1 to 4, configured as follows:

12. A user interface, comprising: displaying the received plurality of physiological events in a graphical representation of a feature space; receiving a user input identifying the user-specified physiological event; displaying information about the user-specified physiological event and an indication that the representative event represents the set of physiological events in the at least one event group; The system of claim 11 , wherein the user interface is configured to:

13. The system of claim 12 , wherein the user interface is configured to receive a user input of the similarity threshold or a user selection of a predetermined cluster density.

14. the controller circuit generating an estimate of a workload reduction associated with assigning the user-determined result of the representative event to each physiological event in the set of physiological events of the at least one event group; Determining or adjusting the similarity threshold based on the generated workload reduction estimate. The system of claim 11 configured to:

15. the controller circuit generating an estimate of an error rate associated with assigning a user determination result of the representative event to each physiological event in the set of physiological events of the at least one event group; Determining or adjusting the similarity threshold based on the generated error rate estimate. The system of claim 11 configured to:

Images