Depolarization identification and arrhythmia localization visualization based on machine learning

Abstract

Disclosed is a technique comprising applying a machine learning model to episode data stored by a medical device, including an electrocardiogram. In some instances, based on applying one or more machine learning models to the episode data, for each of a plurality of arrhythmia type classifications, a processing circuit system derives classification activation data indicating a likelihood of change in the classification over a time period associated with an episode. The processing circuit system may display a graph of the likelihood of change in the arrhythmia type classification over the time period. In some instances, the processing circuit system may use the arrhythmia type likelihood and the depolarization likelihood to identify a depolarization, such as a QRS complex, during the episode.

Description

Technical Field

[0001] The present disclosure relates generally to medical devices, and more particularly to analysis of signals sensed by medical devices. Background Art

[0002] Medical devices can be used to monitor the physiological signals of patients. For example, some medical devices are configured to sense electrocardiogram (EGM) signals, such as electrocardiogram (ECG) signals, which indicate the electrical activity of the heart via electrodes. Some medical devices are configured to detect the occurrence of arrhythmias, commonly referred to as attacks, based on the data from the heart EGM and in some cases from additional sensors. Example arrhythmia types include cardiac arrest, bradycardia, ventricular tachycardia, supraventricular tachycardia, wide complex tachycardia, atrial fibrillation, atrial flutter, ventricular fibrillation, atrioventricular block, ventricular premature beats and atrial premature beats. The medical device can store the heart EGM and other data collected during the time period including the attack as attack data. The medical device can also store the attack data for a period of time in response to user input, such as input from the patient.

[0003] The computing system may obtain episode data from the medical device to allow a clinician or other user to review the episode. The clinician may diagnose the patient's medical condition based on the occurrence of an arrhythmia during the identified episode. In some examples, the clinician or other reviewer may review the episode data to annotate the episode, including determining whether an arrhythmia detected by the medical device actually occurred, prioritizing the episode, and generating a report for further review by a clinician who prescribes a medical device for the patient or is otherwise responsible for the care of a particular patient. Summary of the invention

[0004] In general, the present disclosure describes techniques for classifying, annotating, and reporting on episode data including cardiac EGM data using one or more machine learning models. In some instances, the processing circuit system applies one or more arrhythmia classification machine learning models to the episode data. The one or more machine learning models output corresponding probability values ​​representing the likelihood of the corresponding arrhythmia type classification occurring at any point during the episode. The processing circuit system also derives classification activation data for each arrhythmia type classification, which indicates the likelihood of the classification changing over a time period, and a graph of the likelihood changing over a time period can be displayed. The displayed graph of likelihood over time can help a user understand the reasoning for any arrhythmia detection during an episode, particularly when displayed in conjunction with a basal cardiac EGM.

[0005] In some examples, the processing circuit system applies one or more depolarization detection machine learning models to the episode data. The one or more depolarization detection machine learning models are configured to output a set of depolarization likelihood values ​​that represent the likelihood of depolarization at different times during the episode. The processing circuit system can use the depolarization and arrhythmia type classification likelihood values ​​to improve its ability to detect depolarizations during an episode.

[0006] In one example, the present disclosure describes a computer-implemented method that includes receiving, by a processing circuit system of a medical device system, episode data of an episode stored by a medical device of a patient, wherein the episode is associated with a time period, and the episode data includes an electrocardiogram sensed by the medical device during the time period. The method also includes applying one or more machine learning models to the episode data by the processing circuit system, the one or more machine learning models being configured to output corresponding likelihood values ​​for each of a plurality of arrhythmia type classifications, each of the likelihood values ​​representing the likelihood of the corresponding arrhythmia type classification occurring at any point during the time period. The method also includes deriving, by the processing circuit system and for each of the arrhythmia type classifications, classification activation data indicating the likelihood of a classification changing within the time period based on applying the one or more machine learning models to the episode data. The method also includes displaying, by the processing circuit system and to a user, a graph of the likelihood of arrhythmia type classification changing within the time period.

[0007] In another example, a computer-implemented method includes receiving, by a processing circuit system of a medical device system, episode data of an episode stored by a medical device of a patient, wherein the episode is associated with a time period, and the episode data includes an electrocardiogram sensed by the medical device during the time period. The method also includes applying, by the processing circuit system, one or more arrhythmia classification machine learning models to the episode data, the one or more arrhythmia classification machine learning models being configured to output a set of corresponding arrhythmia type likelihood values ​​for each of a plurality of arrhythmia type classifications, each of the set of arrhythmia type likelihood values ​​representing a likelihood of the corresponding arrhythmia type classification occurring at a corresponding time during the time period. The method also includes applying, by the processing circuit system, one or more depolarization detection machine learning models to the episode data, the one or more depolarization detection machine learning models being configured to output a set of depolarization likelihood values, each of the set of depolarization likelihood values ​​representing a likelihood of a depolarization occurring at a corresponding time during the time period. The method also includes identifying, by the processing circuit system, one or more depolarizations during the episode based on the arrhythmia type likelihood values ​​and the depolarization likelihood values.

[0008] The present disclosure is intended to provide an overview of the subject matter described in the present disclosure. It is not intended to provide an exclusive or exhaustive explanation of the devices and methods described in detail in the following figures and description. Additional details of one or more examples are set forth in the following figures and description. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] Figure 1 A conceptual diagram of an example of a medical device system configured to detect cardiac depolarization and arrhythmias using a machine learning model according to the technical description of the present disclosure.

[0010] Figure 2 For illustration Figure 1 A block diagram of an example configuration of an implantable medical device (IMD).

[0011] Figure 3 For illustration Figure 1 and 2 Conceptual side view of an example configuration of an IMD.

[0012] Figure 4 For illustration Figure 1 A functional block diagram of an example configuration of a computing system.

[0013] Figure 5 A flow chart illustrating example operations for providing a visualization of the likelihood of each of a plurality of arrhythmia classifications over time during an episode.

[0014] Figure 6 A filtered graph is viewed to illustrate an example episode, including a visualization of the likelihood of each of a plurality of arrhythmia classifications over time during the episode.

[0015] Figure 7 A graph illustrating another example episode review filter includes a visualization of the likelihood of each of a plurality of arrhythmia classifications over time during an episode.

[0016] Figure 8 A flow chart illustrating example operations for identifying depolarizations during an episode based on arrhythmia classification and depolarization likelihood.

[0017] Fig. 9 A conceptual diagram illustrating an example technique for identifying depolarizations during an episode based on arrhythmia classification and depolarization likelihood.

[0018] Like reference numerals refer to like elements throughout the drawings and description. DETAILED DESCRIPTION

[0019] Various types of implantable and external medical devices detect arrhythmia episodes based on sensed cardiac EGMs and, in some cases, other physiological parameters. External devices that can be used to noninvasively sense and monitor cardiac EGMs include wearable devices, such as patches, watches, or necklaces, having electrodes configured to contact the patient's skin. An example of a wearable physiological monitor configured to sense cardiac EGMs is the SEEQ 100 from Medtronic plc, of Dublin, Ireland. TM Mobile cardiac telemetry systems. These external devices facilitate relatively long-term monitoring of patients during normal daily activities and can periodically transmit collected data, such as detected arrhythmia episodes, to a remote patient monitoring system such as Medtronic Carelink. TM network.

[0020] Implantable medical devices (IMDs) also sense and monitor the cardiac EGM and detect arrhythmia episodes. Examples of IMDs that monitor the cardiac EGM include pacemakers and implantable cardioverter defibrillators that can be connected to intravascular or extravascular leads, as well as pacemakers with housings configured for implantation in the heart, which can be leadless. Some IMDs that do not provide treatment, such as implantable patient monitors, sense the cardiac EGM. An example of such an IMD is the Reveal LINQ available from Medtronic. TM Insertable cardiac monitors (ICMs) are devices that can be inserted subcutaneously. These IMDs can facilitate relatively long-term monitoring of patients during normal daily activities and can periodically transmit collected data, such as the onset of detected arrhythmia episodes, to a remote patient monitoring system such as Medtronic Carelink. TM network.

[0021] Such a network service may support centralized or clinically based review, annotation, and reporting of arrhythmia episodes by uploading episode data from medical devices and distributing the episode data to various users. The episode data may include an indication of one or more arrhythmias detected by the medical device during the episode. The episode data may also include data collected by the medical device during a time period, the time period including the time before and after the moment when the medical device determines that one or more arrhythmias have occurred. The episode data may include a cardiac EGM digitized during the time period, a heart rate or other parameter derived from the EGM during the time period, and any other physiological parameter data collected by the medical device during the time period.

[0022] Figure 1A conceptual diagram illustrating an example of a medical device system 2 configured to detect cardiac depolarization and arrhythmias using a machine learning model according to the techniques of the present disclosure. The example techniques can be used with an IMD 10 that can wirelessly communicate with an external device 12. In some examples, the IMD 10 is implanted outside the chest of a patient 4 (e.g., subcutaneously). Figure 1 ). The IMD 10 may be positioned near the sternum near or just below the level of the heart of the patient 4, for example at least partially within the outline of the heart. The IMD 10 includes a plurality of electrodes ( Figure 1 10 is not shown in the figure) and is configured to sense cardiac EGM via a plurality of electrodes. In some examples, the IMD 10 employs LINQ TM Form of ICM. Although described primarily in the context of examples in which a medical device collecting seizure data takes the form of an ICM, the techniques of the present disclosure may be implemented in a system including any one or more implantable or external medical devices, including a monitor, pacemaker, or defibrillator.

[0023] External device 12 is a computing device configured for wireless communication with IMD 10. External device 12 may be configured to communicate with computing system 24 via network 25. In some examples, external device 12 may provide a user interface and allow a user to interact with IMD 10. Computing system 24 may include a computing device configured to allow a user to interact with IMD 10 or data collected from the IMD via network 25.

[0024] External device 12 may be used to retrieve data from IMD 10 and may transmit the data to computing system 24 via network 25. The retrieved data may include values ​​of physiological parameters measured by IMD 10, indications of an arrhythmia episode or other disease detected by IMD 10, episode data collected for the episode, and other physiological signals recorded by IMD 10. The episode data may include EGM segments recorded by IMD 10, e.g., because IMD 10 determined that an episode of an arrhythmia or another disease occurred during the segment, or in response to a request for a recorded segment from patient 4 or another user.

[0025] In some examples, computing system 24 includes one or more handheld computing devices, computer workstations, servers, or other networked computing devices. In some examples, computing system 24 may include one or more devices implementing monitoring system 450, including processing circuitry and storage devices. In some examples, computing system 24, network 25, and monitoring system 450 may be connected to a Medtronic Carelink ® computer system. TM Network or other patient monitoring system implementation.

[0026] Network 25 may include one or more computing devices (not shown), such as one or more non-edge switches, routers, hubs, gateways, security devices (such as firewalls), intrusion detection and / or intrusion prevention devices, servers, computer terminals, laptops, printers, databases, wireless mobile devices (such as cellular phones or personal digital assistants), wireless access points, bridges, cable modems, application accelerators, or other network devices. Network 25 may include one or more networks managed by a service provider, and thus may form part of a large-scale public network infrastructure (such as the Internet). Network 25 may provide access to the Internet for computing devices such as computing system 24 and IMD 10, and may provide a connectivity framework that allows computing devices to communicate with each other. In some instances, network 25 may be a dedicated network that provides a communication framework that allows computing system 24, IMD 10, and / or external device 12 to communicate with each other but isolates computing system 24, IMD 10, or external device 12 from devices outside network 25 for security purposes. In some instances, communications between computing system 24, IMD 10, and external device 12 are encrypted.

[0027] Monitoring system 450, implemented, for example, by processing circuitry of computing system 24, may implement techniques of the present disclosure, including applying machine learning models to episode data to detect cardiac depolarizations and arrhythmias. Monitoring system 450 may receive episode data for an episode from a medical device including IMD 10, which may store the episode data in response to its detection of an arrhythmia and / or user input. Based on application of one or more arrhythmia classification machine learning models, monitoring system 450 may determine a likelihood that one or more types of one or more arrhythmias occurred during an episode, in some instances, the episode including an arrhythmia identified by the medical device storing the episode data.

[0028] The monitoring system 450 may also derive and plot activation data indicating the likelihood of various arrhythmia type classifications within a time period of an episode, and apply one or more depolarization detection machine learning models to the episode data to identify the occurrence of depolarizations during an episode, such as an R-wave or a QRS complex. The monitoring system 450 may, for example, display one or more of the arrhythmia type classifications, arrhythmia type classification likelihood plots, landmarks indicating the time of the identified depolarizations, and a cardiac EGM of the episode in combination, which may facilitate a user to review and understand the classification(s) of the episode data by the monitoring system 450. Although the techniques are described herein as being performed by the monitoring system 450, and therefore by the processing circuitry of the computing system 24, the techniques may be performed by the processing circuitry of any one or more devices or systems of a medical device system, such as the computing system 24, the external device 12, or the IMD 10. The machine learning model may include as an example a neural network, a deep learning model, a convolutional neural network, or other type of predictive analysis system.

[0029] Figure 2 For explanation Figure 1 A block diagram of an example configuration of an IMD 10 is shown. Figure 2 As shown, IMD 10 includes processing circuitry 50, sensing circuitry 52, communication circuitry 54, memory 56, sensor 58, switch circuitry 60, and electrodes 16A, 16B (hereinafter "electrodes 16"), one or more of which may be disposed on a housing of IMD 10. In some examples, memory 56 includes computer-readable instructions that, when executed by processing circuitry 50, cause IMD 10 and processing circuitry 50 to perform various functions attributed to IMD 10 and processing circuitry 50 herein. Memory 56 may include any volatile, nonvolatile, magnetic, optical, or electrical media, such as random access memory (RAM), read-only memory (ROM), nonvolatile RAM (NVRAM), electrically erasable programmable ROM (EEPROM), flash memory, or any other digital media.

[0030] Processing circuit system 50 may include fixed function circuit system and / or programmable processing circuit system. Processing circuit system 50 may include any one or more of the following: a microprocessor, a controller, a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA), or an equivalent discrete or analog logic circuit system. In some examples, processing circuit system 50 may include multiple components (such as any combination of one or more microprocessors, one or more controllers, one or more DSPs, one or more ASICs, or one or more FPGAs) and other discrete or integrated logic circuit systems. The functions attributed to processing circuit system 50 herein may be embodied as software, firmware, hardware, or any combination thereof.

[0031] Sensing circuitry 52 may be selectively coupled to electrodes 16A, 16B via switching circuitry 60 as controlled by processing circuitry 50. Sensing circuitry 52 may monitor signals from electrodes 16A, 16B to monitor Figure 1 The processing circuit system 50 may sense the electrical activity of the heart of patient 4 and generate cardiac EGM data of patient 4. In some instances, the processing circuit system 50 may identify features of the sensed cardiac EGM to detect an arrhythmia episode in patient 4. The processing circuit system 50 may store the digitized cardiac EGM and features of the EGM used to detect an arrhythmia episode in the memory 56 as episode data of the detected arrhythmia episode. In some instances, the processing circuit system 50 stores one or more segments of the cardiac EGM data, features derived from the cardiac EGM data, and other episodes in response to instructions from the external device 12 (e.g., when the patient 4 experiences one or more symptoms of an arrhythmia and inputs a command to the external device 12 instructing the IMD 10 to upload data for analysis by a monitoring center or clinician).

[0032] In some examples, processing circuit system 50 transmits seizure data of patient 4 via communication circuit 54 to an external device, such as Figure 1 For example, the IMD 10 sends the digitized cardiac EGM and other episode data to the network 25 for Figure 1 The monitoring system 450 is processed.

[0033] The sensing circuit system 52 and / or the processing circuit system 50 may be configured to detect cardiac depolarization (e.g., a P-wave for atrial depolarization or an R-wave for ventricular depolarization) when the cardiac EGM amplitude crosses a sensing threshold. In some instances, for cardiac depolarization detection, the sensing circuit system 52 may include a rectifier, a filter, an amplifier, a comparator, and / or an analog-to-digital converter. In some instances, the sensing circuit system 52 may output an indication to the processing circuit system 50 in response to the sensing of cardiac depolarization. In this way, the processing circuit system 50 may receive detected cardiac depolarization indications corresponding to the occurrence of detected R-waves and P-waves in the corresponding chambers of the heart. The processing circuit system 50 may use the indications of the detected R-waves and P-waves to determine characteristics of the cardiac EGM, including depolarization intervals, heart rate, and detection of arrhythmias, such as tachyarrhythmias and cardiac arrest. The sensing circuitry 52 may also provide one or more digitized cardiac EGM signals to the processing circuitry 50 for analysis, such as for cardiac rhythm discrimination and / or identifying and characterizing characteristics of the cardiac EGM, such as QRS amplitude and / or width, or other morphological characteristics.

[0034] In some instances, IMD 10 includes one or more sensors 58, such as one or more accelerometers, microphones, optical sensors, and / or pressure sensors. In some instances, sensing circuit system 52 may include one or more filters and amplifiers for filtering and amplifying signals received from one or more of electrodes 16A, 16B and / or other sensors 58. In some instances, sensing circuit system 52 and / or processing circuit system 50 may include rectifiers, filters and / or amplifiers, sensing amplifiers, comparators, and / or analog-to-digital converters. Processing circuit system 50 may determine a physiological parameter value of patient 4 based on the signal from sensor 58, which may be used to identify an arrhythmia episode and stored as episode data in memory 56.

[0035] Communication circuit system 54 may include any suitable hardware, firmware, software, or any combination thereof, for communicating with another device, such as external device 12. Under the control of processing circuit system 50, communication circuit system 54 may receive downlink telemetry from external device 12 or another device via an internal or external antenna (e.g., antenna 26), and send uplink telemetry thereto. In some examples, processing circuit system 50 may communicate via an external device (e.g., external device 12) and a Medtronic® antenna, such as that developed by Medtronic plc of Dublin, Ireland. A computer network of networks communicates with networked computing devices.

[0036] Although described herein in the context of an example IMD 10, the techniques for arrhythmia detection disclosed herein may be used with other types of devices. For example, the techniques may be used with an additional cardiac defibrillator coupled to electrodes external to the cardiovascular system, a transcatheter pacemaker configured for implantation within the heart (such as the Micra 100 commercially available from Medtronic, Inc. of Dublin, Ireland), or a transcatheter pacemaker configured for implantation within the heart. TM transcatheter pacing systems), insertable cardiac monitors (such as RevealLINQ TM ICM, which is also commercially available from Medtronic), neurostimulators, drug delivery devices, medical devices external to the patient 4, wearable devices such as wearable cardioverter defibrillators, fitness trackers or other wearable devices, mobile devices such as mobile phones, "smart" phones, laptops, tablet computers, personal digital assistants (PDAs) or "smart" clothing such as "smart" glasses, "smart" patches or "smart" watches.

[0037] Figure 3 1 is a conceptual side view illustrating an example configuration of an IMD 10. Figure 3In the example shown, IMD 10 may include a leadless, subcutaneously implantable monitoring device having housing 14 and insulating cover 74. Electrodes 16A and 16B may be formed or disposed on an outer surface of cover 74. Figure 2 The described circuit systems 50-56 and 60 can be formed or placed on the inner surface of the cover 74 or within the housing 14. In the illustrated example, the antenna 26 is formed or placed on the inner surface of the cover 74, but in some examples, it can be formed or placed on the outer surface. In some examples, the sensor 58 can also be formed or placed on the inner or outer surface of the cover 74. In some examples, the insulating cover 74 can be positioned over the open housing 14 so that the housing 14 and the cover 74 enclose the antenna 26, the sensor 58, and the circuit systems 50-56 and 60, and protect the antenna and circuit systems from fluids such as body fluids.

[0038] One or more of antenna 26, sensor 58, or circuit system 50-56 may be formed on insulating cover 74, such as by using flip chip technology. Insulating cover 74 may be flipped onto housing 14. When flipped and placed onto housing 14, components of IMD 10 formed on the inner side of insulating cover 74 may be positioned in gap 76 defined by housing 14. Electrode 16 may be electrically connected to switch circuit system 60 by one or more through holes (not shown) formed through insulating cover 74. Insulating cover 74 may be formed of sapphire (i.e., corundum), glass, poly-p-xylene, and / or any other suitable insulating material. Housing 14 may be formed of titanium or any other suitable material (e.g., biocompatible material). Electrode 16 may be formed of any of stainless steel, titanium, platinum, iridium, or alloys thereof. In addition, electrode 16 may be coated with materials such as titanium nitride or fractal titanium nitride, but other suitable materials and coatings for such electrodes may also be used.

[0039] Figure 4 2 is a block diagram illustrating an example configuration of a computing system 24. In the illustrated example, the computing system 24 includes a processing circuit system 402 for executing an application 424 including a monitoring system 450 or any other application described herein. The computing system 24 may be any component or system including a processing circuit system or other suitable computing environment for executing software instructions, and need not include, for example, Figure 4 406; and in some instances, components such as storage device(s) 408 may not be co-located or located in the same rack as other components. In some instances, computing system 24 may be a cloud computing system distributed across multiple devices.

[0040] exist Figure 4In the example of , computing system 24 includes processing circuit system 402, one or more input devices 404, communication circuit system 406, one or more storage devices 408, user interface (UI) device (one or more) 410 and one or more output devices 412. In some examples, computing system 24 also includes one or more applications 424 that can be executed by computing system 24, such as monitoring system 450, and operating system 416. Each of components 402, 404, 406, 408, 410 and 412 can be connected (physically, communicatively and / or operatively) for inter-component communication. In some examples, communication channel 414 may include a system bus, a network connection, an inter-process communication data structure, or any other method for communicating data. As an example, components 402, 404, 406, 408, 410 and 412 can be connected by one or more communication channels 414.

[0041] In one example, processing circuitry 402 is configured to implement functions and / or processing instructions for execution within computing system 24. For example, processing circuitry 402 may be capable of processing instructions stored in storage device 408. Examples of processing circuitry 402 may include any one or more of a microprocessor, a controller, a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA), or equivalent discrete or integrated logic circuitry.

[0042] One or more storage devices 408 may be configured to store information within computing device 400 during operation. In some instances, storage device 408 is described as a computer-readable storage medium. In some instances, storage device 408 is a temporary memory, which means that the primary purpose of storage device 408 is not long-term storage. In some instances, storage device 408 is described as a volatile memory, which means that when the computer is turned off, storage device 408 does not maintain the stored content. Examples of volatile memory include random access memory (RAM), dynamic random access memory (DRAM), static random access memory (SRAM) and other forms of volatile memory known in the art. In some instances, storage device 408 is used to store program instructions executed by processing circuit system 402. In one instance, storage device 408 is used by software or application 424 running on computing system 24 to temporarily store information during program execution.

[0043] In some instances, storage device 408 also includes one or more computer-readable storage media. Storage device 408 may be configured to store a larger amount of information than volatile memory. Storage device 408 may additionally be configured for long-term storage of information. In some instances, storage device 408 includes a non-volatile storage element. Examples of such non-volatile storage elements include magnetic hard disks, optical disks, floppy disks, flash memory, or various forms of electrically programmable memory (EPROM) or electrically erasable and programmable memory (EEPROM).

[0044] In some examples, computing system 24 also includes communication circuitry 406 to communicate with other devices and systems, such as Figure 1 IMD 10 and external device 12. Communication circuit system 406 may include a network interface card, such as an Ethernet card, an optical transceiver, a radio frequency transceiver, or any other type of device that can send and receive information. Other examples of such network interfaces may include 3G and WiFi radios.

[0045] In one example, computing system 24 also includes one or more user interface devices 410. In some examples, user interface device 410 is configured to receive input from a user through tactile, audio, or video feedback. Examples of user interface device (one or more) 410 include a display that is sensitive to presence, a mouse, a keyboard, a voice response system, a camera, a microphone, or any other type of device for detecting commands from a user. In some examples, a display that is sensitive to presence includes a touch-sensitive screen.

[0046] One or more output devices 412 may also be included in the computing system 24. In some instances, the output device 412 is configured to provide output to the user using tactile, audio, or video stimulation. In one instance, the output device 412 includes a presence-sensitive display, a sound card, a video graphics adapter card, or any other type of device for converting signals into an appropriate form understandable to a human or machine. Additional instances of the output device 412 include a speaker, a cathode ray tube (CRT) monitor, a liquid crystal display (LCD), or any other type of device that can generate understandable output to the user.

[0047] Computing system 24 may include operating system 416. In some examples, operating system 416 controls the operation of components of computing system 24. For example, in one example, operating system 416 facilitates communication of one or more applications 424 and monitoring system 450 with processing circuitry 402, communication circuitry 406, storage devices 408, input devices 404, user interface devices 410, and output devices 412.

[0048] Applications 424 may also include program instructions and / or data that may be executed by computing device 400. Example application(s) 424 that may be executed by computing device 400 may include monitoring system 450. Other additional applications not shown may alternatively or additionally be included to provide other functionality described herein and are not depicted for simplicity.

[0049] In accordance with techniques of this disclosure, computing system 24 receives, via communications circuitry 406, episode data for an episode stored by a medical device, such as IMD 10. Storage device 408 may store the episode data for the episode in storage device 408. The episode data may have been collected by the medical device in response to the medical device detecting an arrhythmia and / or user input directing storage of the episode data.

[0050] Monitoring system 450, as implemented by processing circuitry 402, may review and annotate episodes, and after annotation, generate a report or other presentation of the episode for review by a clinician or other reviewer. Monitoring system 450 may utilize input device 404, output device 412, and / or communication circuitry 406 to display episode data, arrhythmia type classifications, plots, identified depolarizations, and any other information described herein to a user, and receive any annotations or other input regarding the episode data from a user.

[0051] To review and annotate episodes, monitoring system 450 may apply episode data as input to one or more machine learning models of choice. Figure 4 In the illustrated example, monitoring system 450 may apply episode data to one or more arrhythmia classification machine learning models 452 and / or one or more depolarization detection models 454. As examples, machine learning models 452 and 454 may include neural networks, such as deep neural networks, which may include convolutional neural networks, multilayer perceptrons, and / or echo state networks, as examples.

[0052] The arrhythmia classification machine learning model 452 can be configured to output a value for each of a plurality of arrhythmia type classifications indicating the likelihood that an arrhythmia of that type occurred at any point during an episode. The monitoring system 450 can apply a configurable threshold (e.g., 50%, 75%, 90%, 95%, 99%) to the likelihood value to annotate an episode as including one or more arrhythmia types, e.g., based on the likelihood that the classification meets or exceeds the threshold.

[0053] In some examples, the arrhythmia classification machine learning model 452 is trained with training data that includes cardiac EGM or other episode data for multiple patients labeled with descriptive metadata. For example, during the training phase, the monitoring system 450 processes multiple cardiac EGM waveforms. Typically, the multiple cardiac EGM waveforms are from multiple different patients, but may be from a single patient. Each cardiac EGM waveform is labeled with one or more episodes of one or more types of arrhythmias.

[0054] For example, a training cardiac EGM waveform may include multiple segments, each segment is labeled with a descriptor that specifies the absence of an arrhythmia or the presence of a specific classification of arrhythmia (e.g., bradycardia, pause, tachycardia, atrial fibrillation, atrial fibrillation, atrioventricular block, or ventricular fibrillation). In some instances, a clinician manually marks the presence of an arrhythmia in each cardiac EGM waveform. In some instances, the presence of an arrhythmia in each cardiac EGM waveform is based on a label classified by a cardiac EGM feature delimitation algorithm, such as a technique similar to that used by IMD 10 to identify arrhythmias based on rate, interval, and state features derived from the cardiac EGM.

[0055] The monitoring system 450 is operable to convert the training data into vectors and multi-dimensional arrays to which the monitoring system 450 can apply mathematical operations, such as linear algebra, nonlinear, or alternative computational operations. The monitoring system 450 uses the training data to teach one or more arrhythmia classification machine learning models 452 to weight different features depicted in the cardiac EGM data. In some examples, the monitoring system 450 uses the cardiac EGM data to teach the machine learning model to apply different coefficients that indicate that one or more features in the cardiac EGM are more or less important to the occurrence of a particular classification of arrhythmias. By processing a variety of such waveforms marked with the onset of an arrhythmia, the monitoring system 450 can establish and train one or more arrhythmia classification machine learning models 452 from patients, such as Figure 1 Patient 4 receives cardiac EGM data that has not been previously analyzed by the monitoring system 450, and processes such cardiac EGM data to detect the presence or absence of differently classified arrhythmia types in the patient with high accuracy. Generally, the larger the amount of cardiac EGM data upon which one or more arrhythmia classification machine learning models 452 are trained, the higher the accuracy of the machine learning model in detecting or classifying arrhythmias in new cardiac EGM data.

[0056] After the monitoring system 450 has trained one or more arrhythmia classification machine learning models 452, the monitoring system 450 may receive episode data, such as electrocardiogram (EGM) data, for a particular patient, such as patient 4. The monitoring system 450 applies the one or more trained arrhythmia classification machine learning models 452 to the episode data to determine whether one or more arrhythmia types occur at any point during the episode.

[0057] In some instances, the monitoring system 450 may process one or more features of the cardiac EGM data in lieu of or in addition to the raw cardiac EGM data itself. The one or more features may be obtained via feature delimitation performed by the IMD 10 and / or the monitoring system 450. These features may include, for example, heart rate, depolarization intervals, other intervals between features of the cardiac EGM, one or more amplitude, width or morphological features of the QRS wave or other features of the cardiac EGM, variability of any of these features, T-wave alternations, or one or more of other types of cardiac features not explicitly described herein. In such example embodiments, in lieu of or in addition to multiple cardiac EGM waveforms marked with the onset of an arrhythmia as described above, the monitoring system 450 may train one or more arrhythmia classification machine learning models 452 via multiple training cardiac features marked with the onset of an arrhythmia.

[0058] In another example, the monitoring system 450 may generate an intermediate representation of the cardiac EGM data from the cardiac EGM data. For example, the monitoring system 450 may apply one or more of signal processing, downsampling, normalization, signal decomposition, wavelet decomposition, filtering, noise reduction, or neural network-based feature representation operations to the electrocardiogram data to generate an intermediate representation of the electrocardiogram data. The monitoring system 450 may process such intermediate representations of the cardiac EGM data to detect and classify various types of arrhythmias of the patient 4. In addition, instead of multiple original cardiac EGM waveforms marked with the onset of arrhythmias as described above, the monitoring system 450 may train one or more arrhythmia classification machine learning models 452 via multiple training intermediate representations marked with the onset of arrhythmias. The use of such intermediate representations of cardiac EGM data may allow the training and development of low-weight, computationally less complex arrhythmia classification machine learning models 452 by the monitoring system 450. In addition, compared to training machine learning models using original cardiac EGM data, the use of such intermediate representations of electrocardiogram data may require fewer iterations and less training data to establish an accurate machine learning model.

[0059] In some examples, the monitoring system 450 may derive, for each of the arrhythmia type classifications, classification activation data indicating the likelihood of a change in the classification over a time period of the episode waveform based on applying the episode data to one or more arrhythmia classification machine learning models 452. For a given arrhythmia type, the magnitude of such likelihood values ​​at different times corresponds to the probability of the arrhythmia occurring at said time, with higher values ​​corresponding to higher probabilities.

[0060] The classification activation map can identify regions of the input time series, such as regions of cardiac EGM data, that constitute the reason why the time series was given a particular classification by one or more arrhythmia classification machine learning models 452. The classification activation map for a given classification can be a univariate time series, where each element (e.g., each timestamp at the sampling frequency of the input time series) can be a weighted sum or other value derived from the output of an intermediate layer of a neural network or other machine learning model. For each classification, the intermediate layer can be the last layer before the global average pooling layer and / or the output layer neurons.

[0061] The monitoring system 450 may display, for example, via output device(s) 412 and / or communicate to another device via communication circuit system 406, a graph of activation data over a time period of an episode. In some instances, the monitoring system 450 may display the classified activation data, for example, on the same screen and simultaneously in conjunction with the input cardiac EGM. While the one or more arrhythmia classification machine learning models 452 may be configured to provide outputs indicating the likelihood of different arrhythmia type classifications occurring throughout an episode, the classified activation data may allow the monitoring system 450 and / or a user to identify times during an episode and points during the cardiac EGM when one or more arrhythmias of one or more types are likely to occur.

[0062] Post-processing of episode data stored by the IMD 10 may include identifying the occurrence of depolarizations within the cardiac EGM data, such as R-waves and / or QRS complexes. The identification of depolarizations during post-processing may be different from that during the detection of the episode by the IMD 10 and storage of the episode data, providing evidence that the IMD 10 may have misclassified the episode. Annotating the cardiac EGM data with markers of identified depolarization occurrences may also facilitate review of the episode data by the user and / or monitoring system 450. Feature delimitation techniques for detecting depolarizations may include filtering the cardiac EGM data, feature extraction (e.g., using a rectified power signal), peak detection, and should not be processed analysis or other further processing. This type of feature delimitation may require feature engineering and detection rule development.

[0063] According to example techniques of the present disclosure, the monitoring system 450 may apply one or more depolarization detection machine learning models 454 to episode data, such as cardiac EGM data. The depolarization detection machine learning model 454 may be configured to output a set of depolarization likelihood values, each of which represents the likelihood of a depolarization occurring at a corresponding time during a time period. The set of depolarization probabilities may be expressed as a probability within a time period, such as p(QRS) in the case of detecting a QRS complex. (t) ).

[0064] The monitoring system 450 can train one or more depolarization detection machine learning models 454 using a training set of cardiac EGM data in a similar manner as described above with respect to training the arrhythmia classification machine learning model 452. The training set of cardiac EGM data is annotated, for example, by a user and / or a feature delimiting algorithm to indicate which portions of the cardiac EGM data are depolarized. When presented with new cardiac EGM data, such as episode data stored by the IMD 10 of the patient 4, the monitoring system 450 can identify one or more depolarizations in the cardiac EGM based on depolarization likelihood values ​​output by the one or more depolarization detection machine learning models 454. For example, the monitoring system 450 can compare the depolarization likelihood value to a configurable and / or adaptive threshold and identify a depolarization when the depolarization likelihood value reaches or exceeds the threshold. In some instances, the monitoring system 450 can further post-process the time of occurrence of the depolarization identified based on the likelihood threshold. For example, the monitoring system 450 can align the identified time with the R-wave peak identified in the cardiac EGM waveform. As another example, monitoring system 450 may apply a non-processable process to the identified depolarization times, such as to eliminate times that are no greater than a threshold time interval from another candidate time.

[0065] In some cases, the monitoring system 450 and one or more depolarization detection machine learning models 454 may have difficulty detecting a particular depolarization in the cardiac EGM data with a high probability, for example, due to a smaller training data set with depolarization annotations and / or the occurrence of arrhythmias, such as tachycardia, pauses, and atrial fibrillation, which may change the morphology and rate of the depolarization. In order to increase the ability of the monitoring system 450 to identify depolarizations, the monitoring system 450 may identify depolarizations based on a depolarization probability value and an arrhythmia type probability value at a time, such as activation data. For example, the arrhythmia type probability value indicates that the probability of atrial tachycardia at a time in the cardiac EGM data is relatively high, and the monitoring system 450 may take one or more actions that result in an effective increase in the depolarization probability value at the time or a reduction in the depolarization detection threshold compared to the depolarization probability value. Such actions may result in an effective adjustment of the probability of depolarization to the probability of an arrhythmia or certain types of arrhythmias. For example, if the arrhythmia classification machine learning model 452 provides the probability p1(X) of arrhythmia type classification X at a certain interval, such as one second during an episode, (t) ), and the depolarization detection machine learning model 454 provides the probability p2(QRS (t) ), then when considering the possibility of arrhythmia, the probability of QRS can be expressed as p3(QRS (t) |p1&p2).

[0066] Figure 5 A flowchart illustrating an example operation for providing a visualization of the likelihood of each of a plurality of arrhythmia classifications over time during an episode. For clarity of description, Figure 5-9 The illustrated operations and techniques are described as being performed by monitoring system 450, and therefore by processing circuitry 402 of computing system 24. However, these operations and techniques, as well as any other operations and techniques described herein, may be performed by processing circuitry of any one or more devices of medical device system 2, such as IMD 10, external device 12, and computing device 24.

[0067] According to Figure 5 In the illustrated example, monitoring system 450 receives episode data (500) for an episode recorded by IMD 10. The episode is associated with a time period, and the episode data includes the cardiac EGM sensed by IMD 10 during the time period. The episode data may include other signals sensed by IMD 10 during the time period, or data derived from the cardiac EGM or other signals.

[0068] The monitoring system 450 applies one or more arrhythmia classification machine learning models 452 (502) to episode data, such as cardiac EGM data. The one or more arrhythmia classification machine learning models 452 output a corresponding likelihood value for each of a plurality of arrhythmia type classifications, each of the likelihood values ​​representing the likelihood of the corresponding arrhythmia type classification occurring at any point during a time period. Based on applying the one or more arrhythmia classification machine learning models 452 to the episode data, for each arrhythmia type classification, the monitoring system 450 also derives classification activation data (506) indicating the likelihood of the classification changing over the time period. As described herein, the monitoring system 450 may derive activation data from an intermediate layer of the machine learning model 452, such as a final layer before a classification layer and / or an output of a global average pooling layer.

[0069] The monitoring system 450 plots activation data for various arrhythmia type classifications over time (506) and displays a graph of the activation data plot to the user (508). In some instances, the monitoring system 450 displays the plot in conjunction with the cardiac EGM of the episode, which may allow the user to associate times of relatively high likelihood of a particular arrhythmia type with portions of the cardiac EGM that caused the relatively high likelihood. In some instances, the monitoring system 450 may indicate, for example, annotate or highlight higher likelihood of at least one of the temporal arrhythmia type classifications relative to other times of the plot of at least one arrhythmia type classification. In some instances, when the monitoring system 450 indicates based on the output of one or more arrhythmia detection machine learning models 452 that a particular type of arrhythmia occurred at some unspecified point during the episode, the monitoring system 450 may further indicate the relatively high likelihood times of the arrhythmia type classification on the plot of the activation data so that the user may understand, for example, the reasoning of the classification by one or more arrhythmia detection machine learning models 452 by reference to corresponding portions of the cardiac EGM.

[0070] Figure 6 and 7 To illustrate an example episode, refer to the diagrams of screens 600, 700, which include visualizations of the likelihood of each of a plurality of arrhythmia classifications over time during the episode. Each of the screens 600 and 700 displays a corresponding cardiac EGM 602, 702 in conjunction with a graph 604, 704 of arrhythmia type classification activation data. Each of the graphs 604 and 704 includes a set of corresponding plots 606, 706 and corresponding keywords 608, 708. Each of the plots 606 and 706 is a plot of the likelihood values ​​over time for a corresponding arrhythmia type classification. The keywords 608 and 708 identify the arrhythmia type classification for each of the plots.

[0071] The graph 604 also includes an annotation 610 identifying an area of ​​the graph 604 where a plot 606 associated with one of the arrhythmia type classifications, such as bradycardia, has a relatively high likelihood value associated with the classification of the associated episode as having at least one arrhythmia of the type. Similarly, the graph 704 includes annotations 710, 712, and 714, each of which indicates that an area of ​​the graph 704 where a corresponding plot 606 associated with one of the arrhythmia type classifications, such as PVC, atrial fibrillation, and sinus tachycardia, has a relatively high likelihood value associated with the classification of the associated episode is at least one arrhythmia of the type. By combining the display of the cardiac EGM 602, 702, the graphs 604, 704, and the annotations 610, 710, 712, and 714 on the screens 600, 700, the monitoring system 450 can allow a user to more easily review, understand, and confirm (or reject) arrhythmia detection and classification of an episode performed by the monitoring system 450.

[0072] By combining the display of cardiac EGM 602, 702, graphs 604, 704 and annotations 610, 710, 712 and 714 on screens 600, 700, the monitoring system 450 can allow the user to quickly identify the portion of the cardiac EGM that is of interest when reviewing an episode to identify / confirm the presence or absence of an arrhythmia. In some instances, the monitoring system 450 can also use activation data to automatically identify segments of episode data from an episode, such as segments of the cardiac EGM, to be included in the report reviewed by the user. In some cases, the duration of the episode is relatively long (e.g., 1 minute) and contains multiple arrhythmias at different times during the episode. Although reports generated by such episodes may include a presentation of the entire cardiac EGM stored for the episode, such reports may additionally or alternatively include representative segments, each of which may include multiple consecutive seconds of cardiac EGM data.

[0073] The monitoring system 450 may identify one or more times (in episode duration) when activation data for a particular arrhythmia type is highest and / or exceeds a threshold. The monitoring system 450 may select a segment of the cardiac EGM to present to the user based on the identified times as representative of the cause of the detection of the arrhythmia type. For example, referring to Figure 7 , the highest probability value of the PVC classification activation map occurs at a peak approximately 18 seconds after the onset of the episode. For example, if a representative PVC segment of this episode needs to be reported, the monitoring system 450 can present a six-second segment of the cardiac EGM from 15 seconds to 21 seconds.

[0074] Figure 8 A flow chart illustrating example operations for identifying depolarizations during an episode based on arrhythmia classification and depolarization likelihood. Figure 8In the illustrated example, monitoring system 450 receives episode data (800) for an episode recorded by IMD 10. The episode is associated with a time period, and the episode data includes the cardiac EGM sensed by IMD 10 during the time period. The episode data may include other signals sensed by IMD 10 during the time period, or data derived from the cardiac EGM or other signals.

[0075] The monitoring system 450 applies one or more arrhythmia classification machine learning models 452 to episode data, such as cardiac EGM data (802). For each arrhythmia type classification, the monitoring system 450 may also derive classification activation data indicating the likelihood of a change in the classification over a time period. The monitoring system 450 also applies one or more depolarization detection machine learning models 454 to the episode data, the one or more depolarization detection machine learning models 454 being configured to output a set of depolarization likelihood values, each of the depolarization likelihood values ​​of the set representing the likelihood of a depolarization occurring at a corresponding time during a time period (804). The monitoring system 450 identifies one or more depolarizations during an episode based on the arrhythmia type likelihood values ​​and the depolarization likelihood values ​​(806).

[0076] The monitoring system 450 may use the arrhythmia type likelihood values ​​to identify depolarizations in a variety of ways. For example, the monitoring system may use a time series of arrhythmia type likelihood values, such as activation data, such as those generated by Figure 7 The illustrated activation data plots 606 and 706 illustrate that the activation data, along with episode data, such as cardiac EGM data, is applied as input to one or more depolarization detection machine learning models 454. The one or more depolarization detection machine learning models 454 can be trained to determine the likelihood of a depolarization at a given time of an episode based on the cardiac EGM input and likelihood values ​​for one or more arrhythmia type classifications. In some instances, the likelihood values ​​for certain arrhythmia type classifications are selected as inputs to the one or more depolarization detection machine learning models 454, rather than all arrhythmia type classifications for which the one or more arrhythmia detection machine learning models 452 output likelihood values.

[0077] In some examples, the monitoring system 450 may apply heuristic rules to one or more arrhythmia type classification probability values, such as selected arrhythmia type classifications. Based on the rules and the arrhythmia type classification probability values ​​at certain times during the episode, the monitoring system 450 may modify, such as increase or decrease, the depolarization probability values ​​provided by one or more depolarization detection models 454 at certain times during the episode. Additionally or alternatively, the monitoring system 450 may similarly modify the depolarization probability threshold based on the rules and the arrhythmia type classification probability values ​​at certain times during the episode, and the monitoring system compares the depolarization probability values ​​at certain times during the episode with the depolarization probability threshold. In either case, the modification of the depolarization probability value and / or the depolarization probability threshold may result in the detection of depolarizations that would otherwise not be detected by the monitoring system 450. The heuristic rules may be configured to modify the depolarization probability value and / or the depolarization probability threshold at a time after a relatively high probability time of a certain type of arrhythmia or at a specific time based on an understanding of the timing of depolarization during the type of arrhythmia. For example, a heuristic rule may be configured to modify a depolarization likelihood value and / or a depolarization likelihood threshold at a particular time after a time of relatively high likelihood of a tachyarrhythmia based on a depolarization rate or depolarization interval typical of a tachyarrhythmia.

[0078] Fig. 9 A conceptual diagram illustrating an example technique for identifying depolarizations during an episode based on arrhythmia classification and depolarization likelihood. Specifically, Fig. 9 An electrocardiogram 900 including depolarizations 902A-902C (collectively "depolarizations 902") is illustrated. Fig. 9 Also illustrated is a depolarization likelihood value plot 904. Depolarization likelihood value plot 904 includes regions 906A-906C (collectively, "regions 906") having relatively high depolarization likelihood values ​​and corresponding in time to depolarization 902. Also illustrated is a depolarization likelihood threshold 908. The depolarization likelihood values ​​of plot 904 in region 906B do not exceed depolarization likelihood threshold 908. Therefore, monitoring system 450 will not detect depolarization 902B.

[0079] Fig. 9 Also illustrated is a plot of an arrhythmia classification likelihood value 910 for a tachyarrhythmia relative to a baseline 912 and a threshold 914. The threshold 914 may be established by a heuristic rule applied by the monitoring system 450. Fig. 9 As illustrated, arrhythmia classification likelihood value 910 exceeds threshold 914 at time 916 .

[0080] Fig. 9Also illustrated is a depolarization likelihood value plot 920 that includes regions 922A-922C (collectively, "regions 922") having relatively high depolarization likelihood values ​​and corresponding in time to depolarization 902. Regions 922A and 922C are identical to regions 906A and 906C, i.e., have not been modified by the monitoring system 450 based on the arrhythmia classification likelihood value 910 heuristic rule. However, based on the arrhythmia classification likelihood value 910 and the heuristic rule, the monitoring system 450 has increased region 922B relative to region 906B based on the arrhythmia classification likelihood value 910 having exceeded threshold 908. Therefore, the monitoring system 450 can detect depolarization 902B.

[0081] In addition to using the depolarization and arrhythmia potential data used in this article, e.g. Figure 8 and 9 In addition to the described techniques identifying depolarizations within an episode, the monitoring system 450 can use depolarization and arrhythmia likelihood data to determine the type of depolarization within an episode. As an example, the type of depolarization may include normal, ventricular premature beats, atrial premature beats, and artifact / noise. One or more arrhythmia classification machine learning models 452 used by the monitoring system to classify episodes can be configured to classify episodes into ventricular premature beats, atrial premature beats, and artifact / noise, as well as other classifications. Therefore, the monitoring system 450 can be configured to derive classification activation values ​​over time for ventricular premature beats, atrial premature beats, and artifact / noise. The monitoring system 450 can also use the classification activation data of those arrhythmia types to determine multiple types, such as which depolarizations are normal, ventricular premature beats, atrial premature beats, and artifact / noise.

[0082] In some instances, the monitoring system 450 may mark a depolarization that occurs when the probability value of the noise / artifact classification reaches (e.g., exceeds) a threshold as artifact / noise. The monitoring system 450 may consider the depolarization intervals of the remaining (e.g., non-artifact / noise) depolarizations during the episode, and determine whether any of these remaining depolarizations is "premature" based on their depolarization intervals (e.g., RR intervals) reaching (e.g., less than) a threshold. The depolarization interval threshold may be determined based on the mean or median of the depolarization intervals during the episode, for example, the threshold may be a fraction or percentage of the mean or median of the depolarization intervals during the episode. The monitoring system 450 may mark a premature depolarization as a ventricular premature beat or an atrial premature beat based on a probability value corresponding in time to the premature depolarization reaching (e.g., exceeding) a threshold. The monitoring system 450 may mark a depolarization identified during an episode, but does not determine one of the ventricular premature beats, atrial premature beats, or artifact / noise as normal.

[0083] In some examples, the techniques of this disclosure include a system including an apparatus to perform any of the methods described herein. In some examples, the techniques of this disclosure include a computer-readable medium including instructions to cause a processing circuit system to perform any of the methods described herein.

[0084] It should be understood that the various aspects disclosed herein may be combined in combinations different from those specifically presented in the specification and drawings. It should also be understood that, depending on the example, certain actions or events in any of the processes or methods described herein may be performed in a different order, may be added, merged, or omitted entirely (e.g., all described actions or events may not be necessary to perform these techniques). In addition, although certain aspects of the present disclosure are described as being performed by a single module, unit, or circuit for clarity, it should be understood that the techniques of the present disclosure may be performed by a combination of units, modules, or circuit systems associated with, for example, a medical device.

[0085] In one or more instances, the described techniques may be implemented in hardware, software, firmware, or any combination thereof. If implemented in software, the functions may be stored in the form of one or more instructions or codes on a computer-readable medium and may be executed by a hardware-based processing unit. A computer-readable medium may include a non-transitory computer-readable medium, which corresponds to a tangible medium, such as a data storage medium (e.g., RAM, ROM, EEPROM, flash memory, or any other medium that can be used to store desired program code in the form of instructions or data structures and can be accessed by a computer).

[0086] Instructions may be executed by one or more processors, such as one or more digital signal processors (DSPs), general purpose microprocessors, application specific integrated circuits (ASICs), field programmable logic arrays (FPGAs), or other equivalent integrated or discrete logic circuit systems. Thus, the terms "processor" or "processing circuit system" as used herein may refer to any of the foregoing structures or any other physical structure suitable for implementing the described techniques. Moreover, the techniques may be fully implemented in one or more circuits or logic elements.

[0087] The following examples illustrate the techniques described herein.

[0088] Example 1: A computer-implemented method, comprising: receiving, by a processing circuit system of a medical device system, episode data of an episode stored in a medical device of a patient, wherein the episode is associated with a time period and the episode data comprises an electrocardiogram sensed by the medical device during the time period; applying, by the processing circuit system, one or more machine learning models to the episode data, the one or more machine learning models being configured to output corresponding likelihood values ​​for each of a plurality of arrhythmia type classifications, each of the likelihood values ​​representing the likelihood of the corresponding arrhythmia type classification occurring at any point during the time period; deriving, by the processing circuit system and for each of the arrhythmia type classifications, classification activation data indicating the likelihood of a change in the classification within the time period based on the application of the one or more machine learning models to the episode data; and displaying, by the processing circuit system and to a user, a graph of the likelihood of a change in the arrhythmia type classification within the time period.

[0089] Example 2: The method of Example 1, wherein displaying the graph comprises displaying the graph in conjunction with an electrocardiogram.

[0090] Example 3: The method of example 1 or 2 further comprising indicating, by the processing circuit system, on the graph, at least one time of higher likelihood of at least one arrhythmia type classification relative to other times of the graph of at least one of the arrhythmia type classifications.

[0091] Example 4: A method according to Example 3, further comprising indicating, by a processing circuit system and based on the output of one or more machine learning models, that at least one arrhythmia type classification occurs at any point during a time period, wherein indicating at least one higher likelihood time of at least one arrhythmia type classification on the curve graph includes indicating at least one higher likelihood time of at least one arrhythmia type classification on the curve graph in response to indicating that at least one arrhythmia type classification occurs at any point during the time period.

[0092] Example 5: A method according to any one of Examples 1 to 4, wherein the multiple arrhythmia type classifications include multiple bradycardias, pauses, ventricular tachycardias, ventricular fibrillation, supraventricular tachycardias, atrial fibrillation, atrial flutter, sinus tachycardia, ventricular premature beats, atrial premature beats, wide-complex tachycardias, and atrioventricular conduction blocks.

[0093] Example 6: A method according to any one of Examples 1 to 5, wherein each of the one or more machine learning models comprises multiple layers, and wherein deriving activation data comprises deriving an activation layer from an output of an intermediate layer of the multiple layers.

[0094] Example 7: The method according to Example 6, wherein the intermediate layer comprises a global average pooling layer.

[0095] Example 8: The method of any one of Examples 1 to 7, further comprising: selecting a segment from the electrocardiogram based on the classified activation data; and displaying the selected segment of the electrocardiogram.

[0096] Example 9: The method of Example 8, wherein selecting a segment comprises: identifying a time during an episode based on the classified activation data of an arrhythmia type classification; and selecting a segment of the electrocardiogram including the identified time.

[0097] Example 10: A computer-implemented method comprising: receiving, by a processing circuit system of a medical device system, episode data of an episode stored by a medical device of a patient, wherein the episode is associated with a time period and the episode data includes an electrocardiogram sensed by the medical device during the time period; applying, by the processing circuit system, one or more arrhythmia classification machine learning models to the episode data, the one or more arrhythmia classification machine learning models being configured to output a set of corresponding arrhythmia type likelihood values ​​for each of a plurality of arrhythmia type classifications, each of the arrhythmia type likelihood values ​​of the set representing the likelihood of a corresponding arrhythmia type classification occurring at a corresponding time during the time period; applying, by the processing circuit system, one or more depolarization detection machine learning models to the episode data, the one or more depolarization detection machine learning models being configured to output a set of depolarization likelihood values, each of the depolarization likelihood values ​​of the set representing the likelihood of a depolarization occurring at a corresponding time during the time period; and identifying, by the processing circuit system, one or more depolarizations during the episode based on the arrhythmia type likelihood values ​​and the depolarization likelihood values.

[0098] Example 11: A method according to Example 10, wherein each of the one or more arrhythmia classification machine learning models includes multiple layers, and the method further includes deriving a set of arrhythmia type likelihood values ​​from the output of an intermediate layer of the multiple layers through a processing circuit system.

[0099] Example 12: The method of Example 11, wherein the intermediate layer comprises a global average pooling layer.

[0100] Example 13: A method according to any one of Examples 10 to 12, wherein applying one or more depolarization detection machine learning models to the episode data and identifying one or more depolarizations based on the arrhythmia type likelihood value and the depolarization likelihood value includes applying one or more depolarization detection machine learning models to the episode data and the arrhythmia type likelihood value.

[0101] Example 14: A method according to any one of Examples 10 to 13, wherein identifying one or more depolarizations based on an arrhythmia type likelihood value and a depolarization likelihood value comprises: modifying one or more of the depolarization likelihood values ​​based on one or more of the arrhythmia type likelihood values; and identifying one or more depolarizations based on the one or more modified depolarization likelihood values.

[0102] Example 15: A method according to any one of Examples 10 to 14, wherein identifying one or more depolarizations based on an arrhythmia type likelihood value and a depolarization likelihood value comprises: modifying a depolarization likelihood threshold based on one or more of the arrhythmia type likelihood values; comparing the depolarization likelihood value with the modified depolarization likelihood threshold; and identifying one or more depolarizations based on the comparison.

[0103] Example 16: The method of any one of Examples 10 to 15, wherein the depolarization comprises at least one of an R-wave or a QRS complex.

[0104] Example 17: The method of any one of Examples 10 to 16, further comprising labeling each of the one or more identified depolarizations as one of a plurality of depolarization types based on the arrhythmia type likelihood value.

[0105] Example 18: The method of Example 17, wherein the plurality of depolarization types comprises a plurality of normal, premature ventricular beats, premature atrial beats, noise, or artifacts.

[0106] Example 19: A medical device system comprising a medical device configured to: sense an electrocardiogram of a patient via a plurality of electrodes; and store episode data of an episode, wherein the episode is associated with a time period and the episode data comprises the electrocardiogram sensed by the medical device during the time period. The medical device system further comprises a processing circuit system configured to perform the method according to any one of Examples 1 to 18.

[0107] Example 20: The medical device system of Example 19, wherein the processing circuit system comprises processing circuit system of a computing device.

[0108] Example 21: The medical device system of Example 19 or 20, wherein the medical device is implantable.

[0109] Example 22: A non-transitory computer-readable medium comprising instructions that, when executed by a processing circuit system of a computing system, cause the computing system to perform the method of any one of Examples 1-18.

[0110] Various examples have been described. These and other examples are within the scope of the following claims.

Claims

1. A medical device system, comprising: A medical device configured to: sensing the patient's electrocardiogram via a plurality of electrodes, and storing episode data for an episode, wherein the episode is associated with a time period and the episode data comprises the electrocardiogram sensed by the medical device during the time period; and A processing circuit system configured to: receiving the seizure data, applying one or more arrhythmia classification machine learning models to the episode data, the one or more arrhythmia classification machine learning models being configured to output a corresponding arrhythmia type likelihood value for each of a plurality of arrhythmia type classifications, each of the arrhythmia type likelihood values ​​representing a likelihood of the corresponding arrhythmia type classification occurring at any point during the time period, applying one or more depolarization detection machine learning models to the seizure data, the one or more depolarization detection machine learning models configured to output a set of depolarization likelihood values, each of the depolarization likelihood values ​​of the set representing a likelihood that a depolarization occurred at a corresponding time during the time period; identifying one or more depolarizations during the episode based on the arrhythmia type likelihood value and the depolarization likelihood value; and At least one of the arrhythmia type likelihood value and an indication of one or more identified depolarizations within the time period is displayed to a user.

2. The medical device system of claim 1, wherein the processing circuit system is configured to display the at least one of the arrhythmia type likelihood value and the indication of the one or more identified depolarizations in conjunction with the electrocardiogram.

3. The medical device system of claim 1 or 2, wherein the processing circuit system is configured to: deriving, for each of the arrhythmia type classifications based on applying the one or more arrhythmia classification machine learning models to the episode data, class activation data indicating a likelihood of change of the classification over the time period; displaying to the user a graph of the likelihood of change of at least one of the arrhythmia type classifications; and At least one higher likelihood time of at least one of the arrhythmia type classifications is indicated on the graph relative to other times on the graph for at least one of the arrhythmia type classifications.

4. The medical device system of claim 3, wherein the processing circuit system is configured to: indicating that at least one arrhythmia type classification occurs at any point during the time period based on the output of the one or more arrhythmia classification machine learning models; and At least one higher likelihood time of the at least one arrhythmia type classification is indicated on the graph in response to indicating that the at least one arrhythmia type classification occurs at any point during the time period.

5. The medical device system according to any one of claims 1 to 2, wherein the multiple arrhythmia type classifications include multiple bradycardias, pauses, ventricular tachycardias, ventricular fibrillation, supraventricular tachycardias, atrial fibrillation, atrial flutter, sinus tachycardia, ventricular premature beats, atrial premature beats, wide complex tachycardias, and atrioventricular conduction blocks.

6. A medical device system according to any one of claims 1 to 2, wherein each of the one or more arrhythmia classification machine learning models comprises multiple layers, and wherein deriving the classification activation data comprises deriving the classification activation data from an output of an intermediate layer of the multiple layers.

7. The medical device system of claim 6, wherein the intermediate layer comprises a global average pooling layer.

8. A medical device system according to claim 1, wherein each of the one or more arrhythmia classification machine learning models comprises multiple layers, and the processing circuit system is configured to derive the set of arrhythmia type likelihood values ​​from the output of an intermediate layer of the multiple layers.

9. A medical device system according to claim 1, wherein in order to identify the one or more depolarizations based on the arrhythmia type likelihood value and the depolarization likelihood value, the processing circuit system is configured to apply the one or more depolarization detection machine learning models to the episode data and the arrhythmia type likelihood value.

10. The medical device system of claim 1 , wherein to identify the one or more depolarizations based on the arrhythmia type likelihood value and the depolarization likelihood value, the processing circuitry is configured to at least one of: modifying one or more of the depolarization likelihood values ​​based on one or more of the arrhythmia type likelihood values; or A depolarization likelihood threshold is modified based on one or more of the arrhythmia type likelihood values.

11. A medical device system according to claim 1, wherein the processing circuit system is configured to mark each of the one or more identified depolarizations as one of a plurality of depolarization types based on the arrhythmia type likelihood value, wherein the plurality of depolarization types include a plurality of normal, ventricular premature beats, atrial premature beats, noise, or artifacts.

12. The medical device system of any one of claims 1-2, wherein the processing circuitry comprises processing circuitry of a computing device.

13. The medical device system of any one of claims 1 to 2, wherein the medical device is implantable.

14. A non-transitory computer-readable medium comprising instructions that, when executed by processing circuitry of a computing system, cause the computing system to: receiving episode data for an episode stored by a medical device of a patient, wherein the episode is associated with a time period and the episode data comprises an electrocardiogram sensed by the medical device during the time period; applying one or more arrhythmia classification machine learning models to the episode data, the one or more arrhythmia classification machine learning models configured to output a corresponding arrhythmia type likelihood value for each of a plurality of arrhythmia type classifications, each of the arrhythmia type likelihood values ​​representing a likelihood of the corresponding arrhythmia type classification occurring at any point during the time period; applying one or more depolarization detection machine learning models to the seizure data, the one or more depolarization detection machine learning models configured to output a set of depolarization likelihood values, each of the depolarization likelihood values ​​of the set representing a likelihood that a depolarization occurred at a corresponding time during the time period; identifying one or more depolarizations during the episode based on the arrhythmia type likelihood value and the depolarization likelihood value; and At least one of the arrhythmia type likelihood value and an indication of one or more identified depolarizations within the time period is displayed.

Images