Electrocardiogram analysis device, electrocardiogram analysis method, and program

Abstract

To facilitate identification of whether or not a subject of electrocardiogram analysis has a paroxysmal arrhythmia. [Solution] An electrocardiogram analysis device 3 according to one embodiment of the present invention comprises a machine learning unit 33 having a machine learning model trained by machine learning using teacher electrocardiogram data from a patient with episodic arrhythmia during a non-attack period when the patient is not experiencing an episode of episodic arrhythmia, an input processing unit 341 that inputs the electrocardiogram data of the subject to be analyzed into the machine learning model, and an output control unit 343 that outputs abnormality information from the machine learning model regarding whether the subject has episodic arrhythmia to an information terminal.

Description

Technical Field

[0001] The present invention relates to an electrocardiogram analysis apparatus, an electrocardiogram analysis method, and a program for analyzing an electrocardiogram.

Background Art

[0002] Conventionally, a Holter electrocardiograph that can be worn on a patient's body to measure an electrocardiogram over a long period of time is known (see, for example, Patent Document 1).

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] Waveform abnormalities that can be evaluated by an electrocardiogram include arrhythmias that occur episodically. Episodic arrhythmias occur, for example, at low frequencies such as once a day to several months. Specifically, episodic arrhythmias include paroxysmal tachycardias such as paroxysmal atrial fibrillation, ventricular tachycardia, supraventricular tachycardia, and atrial flutter, and ventricular fibrillation. When an episode does not occur, the electrocardiogram shows a sinus rhythm electrocardiogram waveform. Therefore, even if the electrocardiogram of a subject for electrocardiogram analysis is measured by a Holter electrocardiograph for a long time (for example, 24 hours), the electrocardiogram may not include the waveform of an episode, and it may be difficult to determine whether the subject has an episodic arrhythmia.

[0005] Therefore, the present invention has been made in view of these points, and an object thereof is to make it easier to determine whether a subject for electrocardiogram analysis has an episodic arrhythmia.

Means for Solving the Problems

[0006] The electrocardiogram analysis device according to the first aspect of the present invention includes a machine learning unit having a machine learning model obtained by machine learning using teacher electrocardiogram data during a non-attack period when an attack of the paroxysmal arrhythmia does not occur in a patient having a paroxysmal arrhythmia, an input processing unit that inputs electrocardiogram data of an analyzer to be analyzed to the machine learning model, and an output control unit that outputs abnormal information regarding whether or not the analyzer has the paroxysmal arrhythmia output from the machine learning model to an information terminal.

[0007] The machine learning unit may have the machine learning model obtained by machine learning using the teacher electrocardiogram data during the non-attack period, which is at least one predetermined period before and / or after the attack period in which the paroxysmal arrhythmia is identified as having occurred.

[0008] The machine learning unit may have the machine learning model obtained by machine learning using the teacher electrocardiogram data of the patient having at least one of paroxysmal atrial fibrillation, ventricular tachycardia, supraventricular tachycardia, atrial flutter, and ventricular fibrillation as the paroxysmal arrhythmia.

[0009] The input processing unit may input the electrocardiogram data measured at the same sampling rate as the teacher electrocardiogram data to the machine learning model.

[0010] The output control unit may output the abnormal information indicating the probability that the analyzer has the paroxysmal arrhythmia to the information terminal.

[0011] The input processing unit may input a plurality of electrocardiogram data measured in a plurality of different periods to the machine learning model, and the output control unit may output the abnormal information regarding the change in the probability in the plurality of different periods to the information terminal.

[0012] The input processing unit may input the electrocardiogram data measured by an electrocardiograph worn by the analyzer during daily life to the machine learning model.

[0013] The output control unit may output the abnormality information representing at least one of a value, degree, or score of the probability that the subject has the arrhythmia that occurs episodically, a value, degree, or score of the probability that the subject does not have the arrhythmia that occurs episodically, the presence or absence of the arrhythmia that occurs episodically in the subject, and the presence or absence of signs of the arrhythmia that occurs episodically in the subject, to the information terminal.

[0014] The machine learning unit may receive determination information indicating whether or not the subject has the arrhythmia that occurs episodically from the information terminal where the abnormality information is output, and regenerate the machine learning model by performing machine learning using the determination information and the electrocardiogram data.

[0015] The electrocardiogram analysis method according to the second aspect of the present invention includes steps of: acquiring a machine learning model obtained by performing machine learning using teacher electrocardiogram data of a patient having an arrhythmia that occurs episodically during a non-episode period in which the episode of the arrhythmia that occurs episodically does not occur, which is executed by a computer; inputting electrocardiogram data of a subject to be analyzed into the machine learning model; and outputting abnormality information regarding whether or not the subject has the arrhythmia that occurs episodically, which is output from the machine learning model, to an information terminal.

[0016] The program according to the third aspect of the present invention causes a computer to function as a machine learning unit having a machine learning model obtained by performing machine learning using teacher electrocardiogram data of a patient having an arrhythmia that occurs episodically during a non-episode period in which the episode of the arrhythmia that occurs episodically does not occur, an input processing unit that inputs electrocardiogram data of a subject to be analyzed into the machine learning model, and an output control unit that outputs abnormality information regarding whether or not the subject has the arrhythmia that occurs episodically, which is output from the machine learning model, to an information terminal.

Advantages of the Invention

[0017] According to the present invention, there is an effect that it is possible to easily identify whether or not a subject of electrocardiogram analysis has an arrhythmia that occurs episodically.

Brief Description of the Drawings

[0018]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Embodiments for Carrying Out the Invention

[0019] [Outline of Electrocardiogram Analysis System S] FIG. 1 is a diagram for explaining an outline of an electrocardiogram analysis system S according to the present embodiment. The electrocardiogram analysis system S includes an electrocardiograph 1, a doctor terminal 2, and an electrocardiogram analysis apparatus 3. A plurality of electrocardiographs 1 and doctor terminals 2 may be provided respectively. The electrocardiogram analysis system S may include other devices such as servers and terminals.

[0020] The electrocardiograph 1 is an electrocardiograph worn by a human. For example, it is an electrocardiograph measuring device that generates electrocardiogram data indicating a human's electrocardiogram by measuring the potential while being worn on a human's wrist, palm, chest, etc. That is, the electrocardiograph 1 is a Holter electrocardiograph (also referred to as a wearable electrocardiograph or a continuous wear electrocardiograph). The electrocardiograph 1 transmits the generated electrocardiogram data to the electrocardiogram analyzer 3 via a network N including a wireless communication line. The electrocardiogram data generated by the electrocardiograph 1 may be delivered to the electrocardiogram analyzer 3 using, for example, a storage medium without passing through the network N.

[0021] The doctor terminal 2 is an information terminal used by medical staff such as doctors who examine humans, and includes, for example, a display (display device) and a computer. The doctor terminal 2 is pre-associated with the medical staff using the doctor terminal 2 based on an ID or the like given to the medical staff. The doctor terminal 2 outputs the abnormality information output by the electrocardiogram analyzer 3 based on the electrocardiogram data generated by the electrocardiograph 1. The doctor terminal 2 may display the abnormality information on the display or output a sound indicating the abnormality information from the speaker.

[0022] The electrocardiogram analyzer 3 is a device that outputs abnormality information regarding whether or not the subject has an arrhythmia occurring episodically based on the electrocardiogram data generated by the electrocardiograph 1, and is, for example, a server. The abnormality information is information representing a probability value, degree, or score of the subject having an arrhythmia occurring episodically, or information on the presence or absence of an arrhythmia occurring episodically or a sign in the subject. Further, the abnormality information may be information representing a probability value, degree, or score of the subject not having an arrhythmia occurring episodically. Further, the abnormality information may be information indicating the presence or absence of an arrhythmia occurring episodically in the subject. Further, the abnormality information may be other information regarding the occurrence of an arrhythmia occurring episodically. The arrhythmia occurring episodically is at least one of paroxysmal tachycardias such as paroxysmal atrial fibrillation, ventricular tachycardia, supraventricular tachycardia, and atrial flutter, and ventricular fibrillation.

[0023] Figure 2 is a block diagram of the electrocardiogram analysis system S according to the present embodiment. In Figure 2, the arrows indicate the main data flow, and there may be a data flow not shown in Figure 2. In Figure 2, each block represents a configuration in terms of functional units, not hardware (device) units. Therefore, the blocks shown in Figure 2 may be implemented within a single device or may be divided and implemented in a plurality of devices. The transfer of data between the blocks may be performed via any means such as a data bus, a network, a portable storage medium, etc.

[0024] The electrocardiogram analysis device 3 includes a communication unit 31, a storage unit 32, a machine learning unit 33, and a control unit 34. The control unit 34 includes an input processing unit 341, a result acquisition unit 342, and an output control unit 343.

[0025] The communication unit 31 has a communication controller for transmitting and receiving data between the electrocardiograph 1 and the doctor terminal 2 via the network N. The communication unit 31 notifies the control unit 34 of the data received from the electrocardiograph 1 and the doctor terminal 2 via the network N. Also, the communication unit 31 transmits the data output from the control unit 34 to the doctor terminal 2 via the network N.

[0026] The storage unit 32 is a storage medium including a ROM (Read Only Memory), a RAM (Random Access Memory), a hard disk drive, etc. The storage unit 32 stores in advance the programs executed by the control unit 34. The storage unit 32 may be provided outside the electrocardiogram analysis device 3, and in that case, data transfer may be performed between the storage unit 32 and the control unit 34 via the network N.

[0027] The machine learning unit 33 generates a machine learning model that outputs abnormality information regarding whether or not the input electrocardiogram data has an arrhythmia that occurs episodically by learning based on teacher electrocardiogram data used as teacher data, and holds the generated machine learning model. The machine learning model is a model generated by performing machine learning using normal electrocardiogram data (i.e., electrocardiogram data of a person who has not developed an arrhythmia that occurs episodically) and abnormal electrocardiogram data (i.e., electrocardiogram data of a person who has developed an arrhythmia that occurs episodically) as teacher data. Also, the machine learning unit 33 may hold a machine learning model generated externally.

[0028] The internal configuration of the machine learning model is arbitrary. For example, it may be composed of a CNN (Convolutional Neural Network) or an RNN (Recurrent Neural Network). The machine learning unit 33 includes, for example, a processor that executes various operations using a CNN, and a memory that stores the coefficients of the CNN. The machine learning model possessed by the machine learning unit 33 outputs abnormality information regarding whether or not the subject has developed an arrhythmia that occurs episodically based on the input electrocardiogram data. The machine learning unit 33 may include only a memory that stores a machine learning model generated externally. At least a part of the functions of the machine learning unit 33 may be incorporated in the control unit 34.

[0029] The control unit 34 is a processor such as a CPU (Central Processing Unit), and functions as an input processing unit 341, a result acquisition unit 342, and an output control unit 343 by executing a program stored in the storage unit 32. At least a part of the functions of the control unit 34 may be executed by an electric circuit. Also, at least a part of the functions of the control unit 34 may be realized by executing a program that the control unit 34 executes via a network.

[0030] The input processing unit 341 inputs the electrocardiogram data of the subject to be analyzed into the machine learning model of the machine learning unit 33. The result acquisition unit 342 acquires the information output by the machine learning model of the machine learning unit 33. The output control unit 343 outputs abnormal information corresponding to whether or not the subject output from the machine learning model of the machine learning unit 33 has an arrhythmia occurring episodically to the doctor terminal 2. The detailed processing executed by the input processing unit 341, the result acquisition unit 342, and the output control unit 343 will be described later.

[0031] The electrocardiogram analysis system S according to the present embodiment is not limited to the specific configuration shown in FIG. 2. The electrocardiograph 1, the doctor terminal 2, and the electrocardiogram analyzer 3 may be configured by connecting two or more physically separated devices by wire or wirelessly. The electrocardiogram analyzer 3 may be configured by a single computer, may be configured by a plurality of computers cooperating with each other, or may be configured by a cloud which is a set of computer resources. Two or more of the electrocardiograph 1, the doctor terminal 2, and the electrocardiogram analyzer 3 may be configured as one device.

[0032] [Explanation of Electrocardiogram Analysis Method] Hereinafter, the electrocardiogram analysis method executed by the electrocardiogram analysis system S according to the present embodiment will be described in detail. The electrocardiogram analyzer 3 generates a machine learning model by performing machine learning using teacher electrocardiogram data in advance. FIG. 3 is a schematic diagram for explaining the method by which the electrocardiogram analyzer 3 performs machine learning. The electrocardiograph 1 measures the electrocardiogram of a patient. The patient is a person diagnosed as having an arrhythmia (for example, at least one of paroxysmal tachycardia such as paroxysmal atrial fibrillation, ventricular tachycardia, supraventricular tachycardia, atrial flutter, etc. and ventricular fibrillation) occurring episodically by a doctor. The electrocardiograph 1 generates electrocardiogram data H0 indicating the measured electrocardiogram and transmits it to the electrocardiogram analyzer 3 via the network N.

[0033] In the electrocardiogram analysis device 3, the input processing unit 341 acquires the electrocardiogram data H0 transmitted by the electrocardiograph 1. For a patient with paroxysmal arrhythmia, there are a non-attack period T0 during which no arrhythmia attack occurs and an attack period T1 during which an arrhythmia attack occurs. A doctor can identify the attack period T1 by looking at the electrocardiogram. The non-attack period T0 is at least one predetermined period before and / or after the attack period T1 identified by the doctor as a period during which an arrhythmia attack occurs, and is a period not identified by the doctor as a period during which an arrhythmia attack occurs. The non-attack period T0 is preferably 7 days before or after the attack period T1, more preferably 24 hours before or after the attack period T1.

[0034] The input processing unit 341 inputs, as teacher electrocardiogram data for a human, the portion of the electrocardiogram data H0 during the non-attack period T0 to the machine learning unit 33 for a human having paroxysmal arrhythmia. In addition, the input processing unit 341 inputs, as teacher electrocardiogram data for a human, the electrocardiogram data of a human diagnosed by a doctor as not having paroxysmal arrhythmia to the machine learning unit 33 as teacher electrocardiogram data for a human without paroxysmal arrhythmia.

[0035] The input processing unit 341 preferably uses, as teacher electrocardiogram data, electrocardiogram data during a predetermined period (for example, during sleep) when a human is at rest. In addition, the input processing unit 341 preferably uses, as teacher electrocardiogram data, electrocardiogram data during a period identified by a doctor as a period during which no symptoms other than the paroxysmal arrhythmia to be analyzed occur. Thereby, the electrocardiogram analysis system S can exclude data that can become noise and improve the accuracy of machine learning.

[0036] The machine learning unit 33 generates a machine learning model that outputs abnormality information regarding whether the input electrocardiogram data is electrocardiogram data of a human having paroxysmal arrhythmia by performing known machine learning (for example, CNN or RNN) using the input teacher electrocardiogram data. In addition, the machine learning unit 33 may acquire a machine learning model generated by an external device (such as a server) by the above-described machine learning method.

[0037] Next, the electrocardiogram analysis device 3 analyzes the electrocardiogram data of the subject to be analyzed using the machine learning model possessed by the machine learning unit 33. FIG. 4 is a schematic diagram for explaining the method by which the electrocardiogram analysis device 3 analyzes electrocardiogram data. The electrocardiograph 1 measures the electrocardiogram of the subject to be analyzed. The electrocardiograph 1 generates electrocardiogram data H1 indicating the measured electrocardiogram and transmits it to the electrocardiogram analysis device 3 via the network N.

[0038] In the electrocardiogram analysis device 3, the input processing unit 341 acquires the electrocardiogram data H1 transmitted by the electrocardiograph 1. It is desirable for the input processing unit 341 to sequentially acquire the electrocardiogram data H1 measured by the electrocardiograph 1 (i.e., a Holter electrocardiograph or a patch-type electrocardiograph) worn by the subject during daily life. Thereby, the electrocardiogram analysis system S can promptly notify medical staff of information regarding whether the subject has paroxysmal arrhythmia. Further, the input processing unit 341 may acquire electrocardiogram data indicating the electrocardiogram of the subject measured in advance at a hospital or the like, electrocardiogram data acquired from an electrocardiograph mounted on the steering wheel of an automobile, electrocardiogram data acquired from a 12-lead electrocardiograph, electrocardiogram data acquired from an electrocardiograph mounted on a smartwatch, and the like.

[0039] The input processing unit 341 inputs the acquired electrocardiogram data H1 as electrocardiogram data to be analyzed into the machine learning model possessed by the machine learning unit 33. Here, it is desirable for the input processing unit 341 to input the electrocardiogram data to be analyzed measured at the same sampling rate (for example, 1000 Hz) as the teacher electrocardiogram data into the machine learning model. Thereby, the electrocardiogram analysis device 3 can suppress the analysis result from being affected by the difference in sampling rate between electrocardiogram data. When the sampling rate of the teacher electrocardiogram data and the sampling rate of the electrocardiogram data to be analyzed are different, the input processing unit 341 may perform a process of converting the sampling rate on the electrocardiogram data to be analyzed and then input it into the machine learning model.

[0040] When the electrocardiogram data is input, the machine learning model of the machine learning unit 33 outputs abnormality information regarding whether the input electrocardiogram data is electrocardiogram data of a person suffering from paroxysmal arrhythmia. For example, as the abnormality information, the machine learning model outputs abnormality information regarding whether the person (i.e., the subject) from whom the input electrocardiogram data is measured is suffering from paroxysmal arrhythmia. The machine learning model outputs, as the abnormality information, at least one of a value of the probability that the subject is suffering from paroxysmal arrhythmia, the degree of the probability of suffering from paroxysmal arrhythmia, a score of the probability that the subject is suffering from paroxysmal arrhythmia, the presence or absence of paroxysmal arrhythmia in the subject, and the presence or absence of signs of paroxysmal arrhythmia in the subject.

[0041] The result acquisition unit 342 acquires, as an analysis result, the abnormality information regarding whether the subject is suffering from paroxysmal arrhythmia, which is output by the machine learning model of the machine learning unit 33. The output control unit 343 outputs the abnormality information acquired by the result acquisition unit 342 to the doctor terminal 2. The abnormality information is represented by a value, degree, or score of the probability that the subject is suffering from paroxysmal arrhythmia. Further, the abnormality information may be information representing a value, degree, or score of the probability that the subject is not suffering from paroxysmal arrhythmia. The degree of probability is represented by, for example, characters or symbols such as high, low, ○, ×, etc. associated with each range of the probability value. The probability score is represented by, for example, a score obtained by converting the probability by a predetermined formula.

[0042] Further, the abnormality information may be information representing the presence or absence of paroxysmal arrhythmia or the presence or absence of signs of arrhythmia in the subject. The presence or absence of arrhythmia is represented by, for example, characters or symbols indicating the determination result of the presence or absence of arrhythmia or the presence or absence of signs of arrhythmia by the machine learning model. Further, the abnormality information may be output for the entire electrocardiogram data, or may be output in association with each of a plurality of periods constituting the electrocardiogram data. Further, the abnormality information may be other information regarding the occurrence of paroxysmal arrhythmia.

[0043] The doctor terminal 2 may display the abnormal information on a display, or may output a voice indicating the abnormal information from a speaker. Further, the doctor terminal 2 may directly display the abnormal information, or may output the abnormal information after converting it into other information (such as "Re-examination is required" or "Medication is required" when the probability is equal to or greater than a predetermined value).

[0044] FIG. 5 is a schematic diagram of an analysis result screen displayed by the doctor terminal 2. The analysis result screen includes an electrocardiogram H and abnormal information P. The electrocardiogram H represents at least a part of the electrocardiogram data used for the analysis. The abnormal information P represents, for example, a value or degree of the probability that the analyzed person has an arrhythmia occurring episodically, output from a machine learning model.

[0045] In this way, the electrocardiogram analysis system S analyzes the electrocardiogram using machine learning, detects signs of arrhythmia occurring episodically during a non-episode period, which are difficult to detect by a doctor's eyes, and can display it as information regarding whether the analyzed person has an arrhythmia occurring episodically. By referring to the electrocardiogram H and the abnormal information P displayed on the analysis result screen, the doctor can easily identify whether the analyzed person has an arrhythmia occurring episodically, and can determine whether to perform additional examinations or the like as necessary.

[0046] The electrocardiogram analysis device 3 may cause the doctor terminal 2 to display changes in probability over a plurality of different periods. In this case, in the electrocardiogram analysis device 3, the input processing unit 341 inputs each of a plurality of electrocardiogram data indicating the electrocardiogram of the analyzed person measured by the electrocardiograph 1 over a plurality of different periods (for example, specific days of each of a plurality of months) as electrocardiogram data to be analyzed into the machine learning model of the machine learning unit 33. The result acquisition unit 342 acquires abnormal information regarding whether the analyzed person has developed an arrhythmia occurring episodically, output by the machine learning model of the machine learning unit 33, in association with each of the plurality of different periods.

[0047] Then, the output control unit 343 causes the doctor terminal 2 to display the abnormality information regarding the change in probability for a plurality of different periods acquired by the result acquisition unit 342. The abnormality information represents the change in probability, for example, by a graph of the change over time or the degree of change in probability (difference, rate of change, etc.). By referring to the change in probability displayed on the analysis result screen, the doctor can determine the risk of the subject developing paroxysmal arrhythmia.

[0048] The analysis result screen may include an area (e.g., buttons or selection boxes) for the doctor to input a diagnosis as to whether the subject has paroxysmal arrhythmia. In this case, the doctor terminal 2 transmits the input content to the electrocardiogram analysis device 3 as determination information as to whether the subject has paroxysmal arrhythmia.

[0049] The electrocardiogram analysis device 3 receives the determination information transmitted by the doctor terminal 2. The machine learning unit 33 uses the received determination information and the electrocardiogram data of the subject to perform the above-described machine learning, thereby regenerating a machine learning model that outputs abnormality information regarding whether the input electrocardiogram data is electrocardiogram data of a person suffering from paroxysmal arrhythmia. As a result, the electrocardiogram analysis system S can receive feedback on the diagnosis result by the doctor and improve the accuracy of the abnormality information output by the machine learning model.

[0050] [Flowchart of electrocardiogram analysis method] FIG. 6 is a diagram showing a flowchart of an electrocardiogram analysis method executed by the electrocardiogram analysis system S according to the present embodiment. The present embodiment includes the electrocardiogram analysis method illustrated in FIG. 6, a program for executing the electrocardiogram analysis method, and a computer-readable recording medium storing the program. In the electrocardiogram analysis apparatus 3, an input processing unit 341 acquires electrocardiogram data transmitted by an electrocardiograph 1 of a patient who has suffered from paroxysmal arrhythmia (S11). The input processing unit 341 inputs a non-attack period portion of the patient's electrocardiogram data to the machine learning unit 33 as teacher electrocardiogram data of a human having paroxysmal arrhythmia. In addition, the input processing unit 341 inputs electrocardiogram data of a human diagnosed by a doctor as not having paroxysmal arrhythmia to the machine learning unit 33 as teacher electrocardiogram data of a human not having paroxysmal arrhythmia.

[0051] The machine learning unit 33 generates a machine learning model that outputs abnormality information regarding whether the input electrocardiogram data is electrocardiogram data of a human who has suffered from paroxysmal arrhythmia by performing machine learning using the input teacher electrocardiogram data (S12). When the machine learning unit 33 uses a machine learning model generated externally, steps S11 to S12 may be omitted.

[0052] The input processing unit 341 acquires electrocardiogram data transmitted by the electrocardiograph 1 of the subject to be analyzed (S13). The input processing unit 341 inputs the acquired electrocardiogram data to the machine learning model possessed by the machine learning unit 33 as electrocardiogram data of the analysis target (S14). When electrocardiogram data is input, the machine learning model of the machine learning unit 33 outputs abnormality information regarding whether the input electrocardiogram data is electrocardiogram data of a human who has suffered from paroxysmal arrhythmia. A result acquisition unit 342 acquires, as an analysis result, the abnormality information output by the machine learning model of the machine learning unit 33 regarding whether the subject to be analyzed has suffered from paroxysmal arrhythmia (S15). An output control unit 343 causes the doctor terminal 2 to display the abnormality information regarding whether the result acquisition unit 342 has acquired it (S16).

[0053] [Effects of the Embodiment] Since paroxysmal arrhythmias occur infrequently, the electrocardiogram waveform of an attack may not be included in the electrocardiogram of a patient measured within a limited period. However, even during the non-attack period when an attack does not occur, the electrocardiogram of a patient with paroxysmal arrhythmias has signs of arrhythmias that are difficult for a doctor to detect. Therefore, the electrocardiogram analysis system S inputs the electrocardiogram data of the subject into a machine learning model that has learned the electrocardiogram data during the non-attack period of a patient with paroxysmal arrhythmias, and outputs abnormality information regarding whether the subject has paroxysmal arrhythmias. Thereby, the electrocardiogram analysis system S can easily identify whether the subject has paroxysmal arrhythmias.

[0054] [Modification Example] In the above-described embodiment, the electrocardiogram analysis device 3 is used to assist a doctor's diagnosis, but it may be used for other purposes. For example, the electrocardiogram analysis device 3 may be used in a health diagnosis result display system that displays the possibility of occurrence of paroxysmal arrhythmias of the subject based on the abnormality information and the necessity of re-examination or detailed examination, an insurance examination support system that includes the abnormality information in the criteria for the insurance application examination of the subject, or an insurance contract document device that includes the abnormality information in the display items of the insurance contract documents of the subject. Thereby, the electrocardiogram analysis device 3 can assist in conducting an insurance application examination or the like based on the health condition of the subject.

[0055] Also, the electrocardiogram analysis device 3 may be used in a clinical trial subject selection system that includes the abnormality information in the display items of information regarding clinical trial candidates. Thereby, the electrocardiogram analysis device 3 can assist in making a judgment such as excluding a candidate who may have paroxysmal arrhythmias from the subjects of a clinical trial regarding the evaluation of an electrocardiograph.

[0056] Further, the electrocardiogram analysis device 3 may be used in a drug administration judgment support system that utilizes abnormal information or a drug administration contraindication judgment support device. Thereby, the electrocardiogram analysis device 3 can support the judgment of whether or not to administer a drug that cannot be administered in the case of arrhythmia to a patient who may have arrhythmia occurring episodically, for example, and can use the abnormal information as a digital biomarker for classifying patients with episodic diseases.

[0057] Also, the electrocardiogram analysis device 3 may be used in an automobile driving control device for safely stopping the vehicle when an arrhythmia occurring episodically during driving of the automobile is detected based on the abnormal information. Further, the electrocardiogram analysis device 3 may be used in an alert system that recommends visiting a medical institution by displaying the occurrence of an arrhythmia occurring episodically on a smart device such as a smartphone based on the abnormal information.

[0058] As described above, the present invention has been described using the embodiments. However, the technical scope of the present invention is not limited to the scope described in the above embodiments, and various modifications and changes are possible within the scope of the gist. For example, all or part of the device can be configured by being functionally or physically distributed and integrated in an arbitrary unit. Also, new embodiments resulting from an arbitrary combination of a plurality of embodiments are also included in the embodiments of the present invention. The effects of the new embodiments resulting from the combination have the effects of the original embodiments combined.

[0059] The processor of the electrocardiogram analysis device 3 serves as the main body of each step (process) included in the electrocardiogram analysis method shown in FIG. 6. That is, the processor of the electrocardiogram analysis device 3 reads out a program for executing the electrocardiogram analysis method shown in FIG. 6 from the storage unit, and executes the program to control each part of the electrocardiogram analysis system S, thereby executing the electrocardiogram analysis method shown in FIG. 6. Some of the steps included in the electrocardiogram analysis method shown in FIG. 6 may be omitted, the order between the steps may be changed, or a plurality of steps may be performed in parallel.

Description of Reference Numerals

[0060] S Electrocardiogram Analysis System 1 Electrocardiograph 2 Physician Terminal 3 Electrocardiogram Analyzer 31 Communication Unit 32 Memory Unit 33 Machine Learning Unit 34 Control Unit 341 Input Processing Unit 342 Result Acquisition Unit 343 Output Control Unit

Claims

1. A machine learning unit having a machine learning model obtained by performing machine learning using first teacher electrocardiogram data during a non-attack period in which it is specified that an abnormal state due to the paroxysmal arrhythmia does not occur, for a patient determined to have paroxysmal arrhythmia, and second teacher electrocardiogram data of a human determined not to have the paroxysmal arrhythmia; An input processing unit that inputs, to the machine learning model, a plurality of electrocardiogram data of an analyzer to be analyzed, the plurality of electrocardiogram data being measured in a plurality of different periods; An output control unit that outputs abnormal information regarding a change in the probability that the analyzer has the paroxysmal arrhythmia in the plurality of different periods output from the machine learning model; An electrocardiogram analysis device comprising:

2. The machine learning unit has the machine learning model obtained by performing machine learning using the first teacher electrocardiogram data during the non-attack period which is at least one predetermined period before and / or after the attack period in which the attack of the paroxysmal arrhythmia is specified. The electrocardiogram analysis device according to Claim 1.

3. The machine learning unit has the machine learning model obtained by performing machine learning using the first teacher electrocardiogram data of the patient determined to have at least one of paroxysmal atrial fibrillation, ventricular tachycardia, supraventricular tachycardia, atrial flutter, and ventricular fibrillation as the paroxysmal arrhythmia. The electrocardiogram analysis device according to Claim 1 or 2.

4. The input processing unit inputs the electrocardiogram data measured at the same sampling rate as the first teacher electrocardiogram data and the second teacher electrocardiogram data to the machine learning model. The electrocardiogram analysis device according to any one of Claims 1 to 3.

5. The input processing unit inputs the electrocardiogram data measured by an electrocardiograph worn by the analyzer during daily life to the machine learning model. The electrocardiogram analysis device according to any one of Claims 1 to 4.

6. The output control unit outputs the abnormal information representing at least one of a value, degree, or score of the probability that the analyzer has the paroxysmal arrhythmia, a value, degree, or score of the probability that the analyzer does not have the paroxysmal arrhythmia, the presence or absence of the paroxysmal arrhythmia in the analyzer, and the presence or absence of signs of the paroxysmal arrhythmia in the analyzer. The electrocardiogram analysis device according to any one of claims 1 to 5.

7. The machine learning unit receives determination information indicating whether the subject has the paroxysmal arrhythmia from the information terminal that outputs the abnormality information, and regenerates the machine learning model by performing machine learning using the determination information and the electrocardiogram data. The electrocardiogram analysis device according to any one of claims 1 to 6.

8. Performed by a computer Obtaining a machine learning model obtained by performing machine learning using first teacher electrocardiogram data during a non-attack period in which it is specified that an abnormal state due to the paroxysmal arrhythmia does not occur in a patient determined to have the paroxysmal arrhythmia, and second teacher electrocardiogram data of a person determined not to have the paroxysmal arrhythmia; Inputting, into the machine learning model, a plurality of electrocardiogram data of a subject to be analyzed, the plurality of electrocardiogram data being measured in a plurality of different periods; Outputting abnormality information regarding a change in the probability that the subject has the paroxysmal arrhythmia in the plurality of different periods output from the machine learning model; An electrocardiogram analysis method comprising:

9. A computer A machine learning unit having a machine learning model obtained by performing machine learning using first teacher electrocardiogram data during a non-attack period in which it is specified that an abnormal state due to the paroxysmal arrhythmia does not occur in a patient determined to have the paroxysmal arrhythmia, and second teacher electrocardiogram data of a person determined not to have the paroxysmal arrhythmia, An input processing unit that inputs, into the machine learning model, a plurality of electrocardiogram data of a subject to be analyzed, the plurality of electrocardiogram data being measured in a plurality of different periods, and An output control unit that outputs abnormality information regarding a change in the probability that the subject has the paroxysmal arrhythmia in the plurality of different periods output from the machine learning model, A program for causing the computer to function as such.

Images