Atrial fibrillation duration estimation support device, learned model production device for atrial fibrillation duration estimation, atrial fibrillation duration estimation support method, learned model production method for atrial fibrillation duration estimation, atrial fibrillation duration estimation support program, learned model production program for atrial fibrillation duration estimation, and recording medium

Abstract

To provide an atrial fibrillation duration estimation support device capable of estimating a duration of a sufferer from an atrial fibrillation.SOLUTION: The atrial fibrillation duration estimation support device comprises an electrocardiogram information acquisition part, a duration estimation part, and an output part. The electrocardiogram information acquisition part acquires electrocardiogram information of a sufferer from an atrial fibrillation. The duration estimation part estimates a duration of the atrial fibrillation of the sufferer from the atrial fibrillation based on the electrocardiogram information. The output part outputs the duration.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present disclosure relates to an atrial fibrillation duration estimation support device, a trained model manufacturing device for estimating the duration of atrial fibrillation, a method for estimating the duration of atrial fibrillation, a trained model manufacturing method for estimating the duration of atrial fibrillation, a program for estimating the duration of atrial fibrillation, a trained model manufacturing program for estimating the duration of atrial fibrillation, and a recording medium. [Background technology]

[0002] In recent years, the number of patients with atrial fibrillation has been increasing with the progress of the super-aging society. Catheter ablation surgery is known as a method for treating atrial fibrillation. For example, Patent Document 1 describes a catheter ablation system that can be used to treat atrial fibrillation and the like. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] JP 2016-147018 A Summary of the Invention [Problem to be solved by the invention]

[0004] In the treatment of atrial fibrillation, it is known that the period (duration) from when a patient develops atrial fibrillation is related to the recurrence rate of atrial fibrillation after catheter ablation surgery. However, in clinical practice, the duration of atrial fibrillation is determined by interviewing the patient himself, and the patient's memory may be vague. In addition, some atrial fibrillation patients are asymptomatic and do not experience any subjective symptoms, making it difficult to know the exact duration of atrial fibrillation.

[0005] Therefore, an object of the present disclosure is to provide an atrial fibrillation duration estimation support device capable of estimating the duration of atrial fibrillation in a patient. [Means for solving the problem]

[0006] In order to achieve the above object, the present disclosure provides an atrial fibrillation duration estimation assistance device, The device includes an electrocardiogram information acquisition unit, a duration estimation unit, and an output unit, The electrocardiogram information acquisition unit acquires electrocardiogram information of a patient having atrial fibrillation, the duration estimation unit estimates a duration of atrial fibrillation of the atrial fibrillation patient based on the electrocardiogram information; The output unit outputs the duration.

[0007] The trained model generating device for estimating the duration of atrial fibrillation according to the present disclosure comprises: A learning information acquisition unit and a trained model generation unit are included, The learning information acquisition unit acquires learning information, the learning information includes electrocardiogram information of an atrial fibrillation patient and a duration of atrial fibrillation of the atrial fibrillation patient; The trained model generation unit generates a duration estimation model as a trained model, which outputs the duration of atrial fibrillation of a subject when electrocardiogram information of the subject is input, through machine learning using the duration as a correct label for the electrocardiogram information.

[0008] The method for assisting in estimating duration of atrial fibrillation according to the present disclosure includes: The method includes an electrocardiogram information acquiring step, a duration estimating step, and an output step, The electrocardiogram information acquiring step acquires electrocardiogram information of a patient having atrial fibrillation, the duration estimation step estimates a duration of atrial fibrillation of the atrial fibrillation patient based on the electrocardiogram information; The output step outputs the duration, Each of the steps is a computer-implemented method.

[0009] The method for generating a trained model for estimating the duration of atrial fibrillation disclosed herein includes the steps of: The method includes a learning information acquisition step and a trained model generation step, The learning information acquiring step acquires learning information, the learning information includes electrocardiogram information of an atrial fibrillation patient and a duration of atrial fibrillation of the atrial fibrillation patient; The trained model generation process uses machine learning with the duration as a correct label for the electrocardiogram information to generate a duration estimation model as a trained model that outputs the duration of atrial fibrillation of the subject when the subject's electrocardiogram information is input.

[0010] The duration estimation support program of the present disclosure is The method includes an electrocardiogram information acquisition step, a duration estimation step, and an output step, The electrocardiogram information acquisition step acquires electrocardiogram information of a patient with atrial fibrillation, the duration estimation step includes estimating a duration of atrial fibrillation of the atrial fibrillation patient based on the electrocardiogram information; The output step outputs the duration; The atrial fibrillation duration estimation support program causes a computer to execute each of the above procedures.

[0011] The present disclosure relates to a program for creating a trained model for estimating the duration of atrial fibrillation, The method includes a learning information acquisition procedure and a trained model generation procedure, The learning information acquisition step acquires learning information, the learning information includes electrocardiogram information of an atrial fibrillation patient and a duration of atrial fibrillation of the atrial fibrillation patient; the trained model generation step generates, as a trained model, a duration estimation model that outputs the duration of atrial fibrillation of a subject when electrocardiogram information of the subject is input, by machine learning using the duration as a correct answer label for the electrocardiogram information; This is a program for creating a trained model for estimating the duration of atrial fibrillation, which causes a computer to execute each of the above procedures.

[0012] The recording medium of the present disclosure includes: The method includes an electrocardiogram information acquisition step, a duration estimation step, and an output step, The electrocardiogram information acquisition step acquires electrocardiogram information of a patient with atrial fibrillation, the duration estimation step includes estimating a duration of atrial fibrillation of the atrial fibrillation patient based on the electrocardiogram information; The output step outputs the duration; The present invention is a computer-readable recording medium having recorded thereon an atrial fibrillation duration estimation assistance program for causing a computer to execute each of the above-mentioned procedures.

[0013] The recording medium of the present disclosure includes: The method includes a learning information acquisition procedure and a trained model generation procedure, The learning information acquisition step acquires learning information, the learning information includes electrocardiogram information of an atrial fibrillation patient and a duration of atrial fibrillation of the atrial fibrillation patient; the trained model generation step generates, as a trained model, a duration estimation model that outputs the duration of atrial fibrillation of a subject when electrocardiogram information of the subject is input, by machine learning using the duration as a correct answer label for the electrocardiogram information; This is a computer-readable recording medium that records a program for creating a trained model for estimating the duration of atrial fibrillation, causing a computer to execute each of the above procedures. Effect of the Invention

[0014] According to the present disclosure, the duration of atrial fibrillation in patients can be estimated. [Brief description of the drawings]

[0015] [Figure 1] FIG. 1 is a block diagram showing a configuration of an example of an atrial fibrillation duration estimation support device according to the present disclosure. [Diagram 2] FIG. 2 is a block diagram showing an example of a hardware configuration of the atrial fibrillation duration estimation support device of the present disclosure. [Diagram 3] FIG. 3 is a flowchart showing an example of processing in the atrial fibrillation duration estimation assistance device of the present disclosure. [Figure 4]FIG. 4 is a block diagram showing a configuration of an example of an atrial fibrillation duration estimation support device according to the present disclosure. [Diagram 5] FIG. 5 is a flowchart showing an example of processing in the atrial fibrillation duration estimation assistance device of the present disclosure. [Figure 6] FIG. 6 is a block diagram showing a configuration of an example of a trained model creation device for estimating the duration of atrial fibrillation disclosed herein. [Figure 7] FIG. 7 is a block diagram showing an example of the hardware configuration of the trained model creation device for estimating the duration of atrial fibrillation disclosed herein. [Figure 8] FIG. 8 is a flowchart showing an example of processing in the trained model creation device for estimating the duration of atrial fibrillation disclosed herein. [Figure 9] FIG. 9 is a graph showing the results of the examples. [Figure 10] FIG. 10 is a graph showing the results of the example. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0016] Next, an embodiment of the present disclosure will be described with reference to the drawings. The present disclosure is not limited to the following embodiment. In each of the following drawings, the same parts are given the same reference numerals. In addition, the description of each embodiment can be mutually incorporated unless otherwise specified, and the configurations of each embodiment can be combined unless otherwise specified.

[0017] [Embodiment 1] The atrial fibrillation duration estimation support device of this embodiment will be described with reference to Fig. 1. Fig. 1 is a block diagram showing the configuration of an example of an atrial fibrillation duration estimation support device 10 of this embodiment. As shown in Fig. 1, the atrial fibrillation duration estimation support device 10 (hereinafter also referred to as "the device 10") includes an electrocardiogram information acquisition unit 11, a duration estimation unit 12, and an output unit 13. In addition, the device 10 may include, for example, an input unit, an output unit, a display unit, and / or a storage unit, although these are not shown.

[0018] The device 10 may be, for example, a single device including each of the above-mentioned units, or a device to which each of the above-mentioned units can be connected via a communication line network. The device 10 may be connected to an external device described later via a communication line network. The communication line network is not particularly limited, and a known network may be used, for example, wired or wireless. Examples of the communication line network include an Internet line, a World Wide Web (WWW), a telephone line, a Local Area Network (LAN), a Storage Area Network (SAN), a Delay Tolerant Networking (DTN), a Low Power Wide Area (LPWA), and a Local 5G (L5G). Examples of wireless communication include Wi-Fi (registered trademark), Bluetooth (registered trademark), a Local 5G, and a LPWA. Examples of the wireless communication may include a form in which each device directly communicates with each other (Ad Hoc communication), an infrastructure (infrastructure communication), and an indirect communication via an access point. The device 10 may be incorporated into a server as a system, for example. The device 10 may be, for example, a personal computer (PC, for example, desktop type or notebook type) on which the program of the present disclosure is installed, a smartphone, a tablet terminal, etc. Furthermore, the device 10 may be in the form of cloud computing or edge computing, for example, in which at least one of the units is on a server and the other units are on a terminal.

[0019] 2 illustrates a block diagram of the hardware configuration of the present device 10. The present device 10 includes, for example, a central processing unit 101, a memory 102, a bus 103, a storage device 104, an input device 105, an output device 106, and a communication device (communication unit) 107. The components of the present device 10 are connected to each other via the bus 103 and their respective interfaces (I / F).

[0020] The central processing unit 101 cooperates with other components through a controller (such as a system controller or an I / O controller) and controls the entire device 10. In the device 10, the central processing unit 101 executes, for example, the program disclosed herein and other programs, and also reads and writes various information. Specifically, for example, the central processing unit 101 functions as an electrocardiogram information acquisition unit 11, a duration estimation unit 12, and an output unit 13. The device 10 may include, as a computing device, other computing devices such as a CPU, a GPU (Graphics Processing Unit), an APU (Accelerated Processing Unit), or a combination of these.

[0021] The bus 103 can also be connected to, for example, an external device. Examples of the external device include an external storage device (external database, etc.), an electrocardiograph, a printer, an external input device, an external display device, an audio output device such as a speaker, an external imaging device such as a camera, and various sensors such as an acceleration sensor, a geomagnetic sensor, and a direction sensor. The present device 10 can be connected to an external network (the communication line network) by, for example, a communication device 107 connected to the bus 103, and can also be connected to other devices via the external network.

[0022] An example of the memory 102 is a main memory (primary storage device). When the central processing unit 101 performs processing, the memory 102 reads various operation programs, such as the program of the present disclosure, stored in the storage device 104 described below, and the central processing unit 101 receives data from the memory 102 and executes the program. The main memory is, for example, a RAM (random access memory). The memory 102 may also be, for example, a ROM (read only memory).

[0023] The storage device 104 is also called an auxiliary storage device, for example, in contrast to the main memory (primary storage device). As described above, the storage device 104 stores an operation program including the program of the present disclosure. The storage device 104 may be, for example, a combination of a recording medium and a drive that reads and writes from and to the recording medium. The recording medium is not particularly limited, and may be, for example, an internal or external type, such as a HD (hard disk), a CD-ROM, a CD-R, a CD-RW, an MO, a DVD, a flash memory, a memory card, and the like. The storage device 104 may be, for example, a hard disk drive (HDD) in which a recording medium and a drive are integrated, and a solid state drive (SSD). When the present device 10 includes, for example, the storage unit, the storage device 104 functions as the storage unit. The storage device 104 may store, for example, a duration estimation model described later.

[0024] In the present device 10, the memory 102 and the storage device 104 can also store various information such as log information, information acquired from an external database (not shown) or an external device, information generated by the present device 10, and information used when the present device 10 executes processing. Note that at least a portion of the information may be stored, for example, in an external server other than the memory 102 and the storage device 104, or may be stored in a distributed manner in multiple terminals using block chain technology or the like.

[0025] The device 10 further includes, for example, an input device 105 and an output device 106. The input device 105 may be, for example, a pointing device such as a touch panel, a track pad, or a mouse; a keyboard; an imaging means such as a camera or a scanner; a card reader such as an IC card reader or a magnetic card reader; and an audio input means such as a microphone. The output device 106 may be, for example, a display device such as an LED display or a liquid crystal display; an audio output device such as a speaker; a printer; or the like. In the first embodiment, the input device 105 and the output device 106 are configured separately, but the input device 105 and the output device 106 may be configured as an integrated device, such as a touch panel display.

[0026] Next, an example of the method for supporting estimation of atrial fibrillation duration according to this embodiment will be described with reference to the flowchart of Fig. 3. The method for supporting estimation of atrial fibrillation duration according to this embodiment can be implemented as follows, for example, using the atrial fibrillation duration estimation support device 10 shown in Figs. 1 and 2. Note that the method for supporting estimation of atrial fibrillation duration according to this embodiment is not limited to use of the atrial fibrillation duration estimation support device 10 shown in Figs. 1 and 2.

[0027] First, the electrocardiogram information acquiring unit 11 acquires electrocardiogram information of the atrial fibrillation patient (S1, electrocardiogram information acquiring step). The electrocardiogram information is, for example, information of a graph image recording the electrocardiogram of the atrial fibrillation patient. The recording method of the electrocardiogram is not particularly limited, and may be, for example, a 12-lead electrocardiogram test, an intraesophageal electrocardiogram test, an intracardiac electrocardiogram test, or an electrocardiogram monitor. In addition, the electrocardiogram examination method of the atrial fibrillation patient is not particularly limited, and may be, for example, a resting electrocardiogram, a stress electrocardiogram, a Holter electrocardiogram, or a fetal electrocardiogram. The electrocardiogram information acquiring unit 11 may acquire the electrocardiogram information from, for example, various devices (electrocardiographs) for electrocardiogram measurement, or may acquire the electrocardiogram information from a recording medium on which the electrocardiogram information is recorded.

[0028] The electrocardiogram information acquiring unit 11 may acquire other information by linking it to the electrocardiogram information, for example. Examples of the other information include identification information of the atrial fibrillation patient; patient information of the atrial fibrillation patient; and the like. The identification information is not particularly limited as long as it is information that can identify the atrial fibrillation patient, and examples of the identification information include name, address, telephone number, email address, and identification number (for example, hospital appointment card number, health insurance card number, My Number (personal number), etc.). The patient information includes, for example, at least one selected from the group consisting of attribute information, interview information, and echocardiographic information of the atrial fibrillation patient. The attribute information is not particularly limited, and examples of the attribute information include medical history such as myocardial infarction history, heart failure, and dialysis; gender; age; vital data such as blood test results and blood pressure; lifestyle habits such as smoking and drinking; and information on physical characteristics such as height and weight; and the like. The interview information is, for example, information of a set of questions and answers in an interview regarding the subjective symptoms of atrial fibrillation. Specific examples of the interview include questions and answers related to the onset of atrial fibrillation, such as "When did you start having palpitations?" and "When were you first diagnosed with atrial fibrillation?", but are not limited to these. The echocardiographic information is, for example, information on the results of an echocardiographic examination (cardiac ultrasound examination).

[0029] Next, the duration estimation unit 12 estimates the duration of atrial fibrillation of the atrial fibrillation patient based on the electrocardiogram information (S2, duration estimation step). The duration estimation unit 12 can, for example, input the electrocardiogram information into a duration estimation model to estimate the duration of atrial fibrillation of the atrial fibrillation patient. The duration estimation model is a model trained to output the duration of atrial fibrillation of the subject when electrocardiogram information of the subject, who is an atrial fibrillation patient, is input by machine learning using the duration of atrial fibrillation of the atrial fibrillation patient as a correct answer label for the electrocardiogram information. The duration estimation model may be stored in advance in the memory 102 and the storage device 104, for example, or may be acquired from an external atrial fibrillation duration estimation trained model manufacturing device (for example, atrial fibrillation duration estimation trained model manufacturing device 20 described later) via a communication line network. The duration estimation unit 12 may estimate, for example, the number of days from the onset of atrial fibrillation in the atrial fibrillation patient, or may estimate the elapsed time from the onset of atrial fibrillation, as the duration. In the latter case, the duration estimation unit 12 may estimate, for example, a classification based on the duration of atrial fibrillation in the atrial fibrillation patient. The classification may be, for example, paroxysmal atrial fibrillation whose duration is within 7 days from the onset; persistent atrial fibrillation whose duration is more than 7 days and less than 1 year from the onset; long-term persistent atrial fibrillation whose duration is more than 1 year from the onset; and the like. The duration estimation unit 12 may store the estimated duration in the storage unit of the device 10, for example.

[0030] The duration estimation model can be generated by machine learning using, for example, a pair of electrocardiogram information of an atrial fibrillation patient and the duration of atrial fibrillation of the atrial fibrillation patient as teacher data. For example, a known machine learning method can be adopted for the machine learning. As a specific example, a statistical model adopted for the machine learning can be, for example, a simple linear regression model, Ridge regression, Lasso regression, Elastic Net regression, LightGBM (Light Gradient Boosting Machine), Logistic regression, general additive model, random forest regression, rule fit regression, gradient boosting decision tree, extra tree, support vector regression, Gaussian process regression, k-nearest neighbor regression, kernel ridge regression, neural network, etc. Note that the duration estimation model may be, for example, a trained model generated in advance. In addition, the trained model may be a trained model (derived model) retrained using the teacher data and a trained model already generated. Furthermore, the trained model may be a trained model obtained by transfer learning using a trained model generated using the training data, or may be a trained model generated by model compression of a trained model generated using the training data.

[0031] The duration estimation model may be a network including, for example, an input layer for inputting electrocardiogram information, an output layer for outputting the duration, and at least one intermediate layer provided between the input layer and the output layer. In this case, the duration estimation model may be a program module that is a part of artificial intelligence software. An example of the multi-layered network is a neural network. An example of the neural network is a convolution neural network (CNN), but is not limited to CNN, and may be a trained model constructed by other learning algorithms such as a neural network other than CNN, a support vector machine (SVM), a Bayesian network, or a regression tree.

[0032] When the patient information is acquired in S1, the duration estimation unit 12 may estimate the duration of atrial fibrillation of the atrial fibrillation patient based on, for example, the electrocardiogram information and the patient information. In this case, the duration estimation unit 12 can, for example, input the electrocardiogram information and the patient information into a duration estimation model to estimate the duration of atrial fibrillation of the atrial fibrillation patient. The duration estimation model in this case is the same as that described above, except that it is a model trained to output the duration of atrial fibrillation of the subject when electrocardiogram information and patient information of the subject, who is an atrial fibrillation patient, are input, for example, by machine learning using the duration of atrial fibrillation of the atrial fibrillation patient as a correct answer label for a combination of the electrocardiogram information and the patient information.

[0033] Then, the output unit 13 outputs the duration (S3, output step). The output may be, for example, output to the output device 106 (for example, a display) of the device 10, or may be output to an external device outside the device 10. The output unit 13 may output the duration in association with identification information of the atrial fibrillation patient, for example.

[0034] In a patient with atrial fibrillation, when a doctor estimated the duration of atrial fibrillation based on the patient's interview, the predictive ability (AUC: Area Under Curve) was 0.65. In addition, when a doctor estimated the duration based on the results of an echocardiogram in addition to the interview, the predictive ability (AUC) was 0.72. In contrast, the atrial fibrillation duration estimation support device of the present disclosure was able to estimate the duration of atrial fibrillation with high accuracy, with a predictive ability (AUC) of 0.79. Therefore, according to the present disclosure, it is possible to objectively and accurately estimate the duration of atrial fibrillation in a patient based on the patient's electrocardiogram information. This makes it possible to optimize treatment, for example, by proposing catheter ablation surgery to a patient whose duration is short and for whom catheter ablation surgery is expected to be highly effective, and by proposing treatment other than catheter ablation surgery to a patient whose duration is long and for whom catheter ablation surgery is not expected to be very effective (high possibility of recurrence). Therefore, according to the present disclosure, it is possible to expect the effect of suppressing medical expenses by optimizing treatment.

[0035] [Embodiment 2] The second embodiment is another example of an atrial fibrillation duration estimation assistance device according to the present disclosure.

[0036] FIG. 4 is a block diagram showing an example of the configuration of the duration estimation support device 10A. As shown in FIG. 4, the duration estimation support device 10A includes a recurrence prediction unit 14 in addition to the configuration of the duration estimation support device 10 of the first embodiment. The hardware configuration of the duration estimation support device 10A is the same as that of the duration estimation support device 10 of FIG. 2, except that the central processing unit 101 has the configuration of the duration estimation support device 10A of FIG. 4 instead of the configuration of the duration estimation support device 10 of FIG. 1. The process of the recurrence prediction unit 14 will be described below. The process of the recurrence prediction unit 14 may be inserted at any position in the flowchart of FIG. 3 described in the first embodiment, or may be a separate independent process as shown in FIG. 5.

[0037] The recurrence prediction unit 14 predicts the recurrence rate of the atrial fibrillation patient based on, for example, at least one of the electrocardiogram information and the duration (S11, recurrence prediction step). The recurrence rate is, for example, the possibility of recurrence of atrial fibrillation after a radical treatment for atrial fibrillation, such as a catheter ablation operation. The recurrence prediction unit 14 may, for example, predict the recurrence rate qualitatively or quantitatively. The qualitative prediction is, for example, a prediction of whether or not there is a possibility of recurrence, and the quantitative prediction is, for example, a prediction that expresses the possibility of recurrence in a numerical value, such as a probability of recurrence of n% (n is a positive number). The recurrence prediction unit 14 may, for example, predict the recurrence rate based on the electrocardiogram information, may predict the recurrence rate based on the duration, or may predict the recurrence rate based on the electrocardiogram information and the duration. The recurrence prediction unit 14 can predict the recurrence rate by using, for example, a machine learning model. The machine-learned model is a model trained to output the recurrence rate of atrial fibrillation patients when at least one of the electrocardiogram information and the duration is input, by machine learning using the recurrence rate of the atrial fibrillation patients as a correct answer label for at least one of the electrocardiogram information and the duration. The machine-learned model may be stored in advance in the memory 102 and the storage device 104, or may be acquired from an external device via a communication network.

[0038] The output unit 13 outputs, for example, the recurrence rate (S12, output step (recurrence rate output step)). The output may be, for example, output to the output device 106 (for example, a display) of the device 10, or may be output to an external device outside the device 10. The output unit 13 may output, for example, the recurrence rate in association with identification information of the atrial fibrillation patient.

[0039] According to the atrial fibrillation duration estimation support device 10A of this embodiment, for example, the recurrence prediction unit 14 can predict the recurrence rate of the atrial fibrillation patient based on at least one of the electrocardiogram information and the duration. This makes it possible to optimize treatment, for example, by proposing catheter ablation surgery to patients with a low recurrence rate who can be expected to benefit from catheter ablation surgery, and conversely, by proposing treatment other than catheter ablation surgery to patients with a high recurrence rate who can be expected to benefit from catheter ablation surgery. Therefore, according to the present disclosure, it is expected that the optimization of treatment will have an effect of reducing medical expenses.

[0040] [Embodiment 3] Embodiment 3 is an example of a trained model creation device for estimating the duration of atrial fibrillation according to the present disclosure.

[0041] The trained model producing device for estimating the duration of atrial fibrillation of this embodiment will be described with reference to Fig. 6. Fig. 6 is a block diagram showing an example of the configuration of the trained model producing device for estimating the duration of atrial fibrillation 20 of this embodiment. As shown in Fig. 6, the trained model producing device for estimating the duration of atrial fibrillation 20 (hereinafter also referred to as "this device 20") includes a learning information acquiring unit 21 and a trained model generating unit 22. In addition, although not shown, this device 20 may include, for example, an input unit, an output unit, a display unit and / or a memory unit.

[0042] The trained model manufacturing apparatus 20 for estimating the duration of atrial fibrillation may be, for example, a single apparatus including each of the above-mentioned units, or each of the above-mentioned units may be an apparatus that can be connected via a communication network. The trained model manufacturing apparatus 20 for estimating the duration of atrial fibrillation may be connected to an external apparatus described later via a communication network. The communication network is not particularly limited, and a known network may be used, for example, wired or wireless. Examples of the communication network include the Internet line, WWW (World Wide Web), telephone line, LAN (Local Area Network), SAN (Storage Area Network), DTN (Delay Tolerant Networking), LPWA (Low Power Wide Area), L5G (Local 5G), etc. Examples of the wireless communication include Wi-Fi (registered trademark), Bluetooth (registered trademark), local 5G, LPWA, etc. Examples of the wireless communication may include a form in which each device directly communicates with each other (Ad Hoc communication), infrastructure (infrastructure communication), indirect communication via an access point, etc. The trained model producing apparatus 20 for estimating the duration of atrial fibrillation may be incorporated in a server as a system, for example. The trained model producing apparatus 20 for estimating the duration of atrial fibrillation may be, for example, a personal computer (PC, for example, desktop type or notebook type) on which the program of the present disclosure is installed, a smartphone, a tablet terminal, or the like. Furthermore, the trained model producing apparatus 20 for estimating the duration of atrial fibrillation may be in the form of cloud computing or edge computing, for example, in which at least one of the above-mentioned units is on a server and the other units are on a terminal.

[0043] FIG. 7 illustrates a block diagram of a hardware configuration of the trained model manufacturing device 20 for estimating the duration of atrial fibrillation. As shown in FIG. 7, the trained model manufacturing device 20 for estimating the duration of atrial fibrillation includes, for example, a central processing unit 201, a memory 202, a bus 203, a storage device 204, an input device 205, an output device 206, a communication device 207, and the like. The description of each component of the trained model manufacturing device 20 for estimating the duration of atrial fibrillation can be used with reference to the description of each component of the atrial fibrillation duration estimation support device 10. Each unit of the trained model manufacturing device 20 for estimating the duration of atrial fibrillation is connected via a bus 203 by each interface (I / F). In the trained model manufacturing device 20 for estimating the duration of atrial fibrillation, the central processing unit 201 functions as a learning information acquisition unit 21 and a trained model generation unit 22.

[0044] Next, an example of a method for manufacturing a trained model of this embodiment will be described based on the flowchart of Fig. 8. The method for manufacturing a trained model of this embodiment is implemented as follows, for example, using the trained model manufacturing device 20 for estimating the duration of atrial fibrillation of Fig. 6 and Fig. 7. Note that the method for manufacturing a trained model of this embodiment is not limited to the use of the trained model manufacturing device 20 for estimating the duration of atrial fibrillation of Fig. 6 and Fig. 7.

[0045] First, the learning information acquisition unit 21 acquires learning information (S21, learning information acquisition step). The learning information includes, for example, electrocardiogram information of an atrial fibrillation patient and the duration of atrial fibrillation of the atrial fibrillation patient. The learning information may also include, for example, at least one selected from the group consisting of attribute information, interview information, and echocardiographic information of the atrial fibrillation patient. Each piece of information included in the learning information is similar to that in the first and second embodiments, and the description thereof can be used. The learning information acquisition unit 21 may acquire, for example, information previously stored in the memory 202 and the storage device 204, or may acquire information stored in an external database via a communication network.

[0046] Next, the trained model generating unit 22 generates a duration estimation model that outputs the duration of atrial fibrillation of the subject as a trained model when the subject's electrocardiogram information is inputted by machine learning using the duration as a correct answer label for the electrocardiogram information (S22, learning step). The trained model generating unit 22 can generate the duration estimation model by machine learning using, for example, a pair of electrocardiogram information of an atrial fibrillation patient and the duration of atrial fibrillation of the atrial fibrillation patient as teacher data. For example, a known machine learning method can be adopted for the machine learning. As a specific example, a statistical model adopted for the machine learning can be, for example, a simple linear regression model, Ridge regression, Lasso regression, Elastic Net regression, LightGBM (Light Gradient Boosting Machine), Logistic regression, general additive model, random forest regression, rule fit regression, gradient boosting decision tree, extra tree, support vector regression, Gaussian process regression, k-nearest neighbor regression, kernel ridge regression, neural network, etc. The duration estimation model may be, for example, a trained model generated in advance. The trained model may also be a trained model (derived model) retrained using the teacher data and a trained model that has already been generated. Furthermore, the trained model may be a trained model obtained by transfer learning using a trained model generated using the teacher data, or a trained model generated by model compression of a trained model generated using the teacher data.

[0047] The trained model generating unit 22 may generate, for example, a network including an input layer for inputting electrocardiogram information, an output layer for outputting the duration, and at least one intermediate layer provided between the input layer and the output layer as the duration estimation model. In this case, the duration estimation model may be a program module that is a part of artificial intelligence software. An example of the multi-layered network is a neural network. An example of the neural network is a convolution neural network (CNN), but is not limited to CNN, and may be a trained model constructed by other learning algorithms such as a neural network other than CNN, a support vector machine (SVM), a Bayesian network, or a regression tree.

[0048] When the learning information includes the patient information, the trained model generation unit 22 may generate a duration estimation model trained to output the duration of atrial fibrillation of a subject when electrocardiogram information and patient information of a subject who has atrial fibrillation are input, for example, by machine learning using the duration of atrial fibrillation of the atrial fibrillation patient as a correct answer label for the combination of the electrocardiogram information and the patient information. The duration estimation model in this case is the same as that described above, except that it is a model trained to output the duration of atrial fibrillation of the subject when electrocardiogram information and patient information of a subject who has atrial fibrillation are input, for example, by machine learning using the duration of atrial fibrillation of the atrial fibrillation patient as a correct answer label for the combination of the electrocardiogram information and the patient information.

[0049] The trained model generated by this embodiment is used, for example, in the atrial fibrillation duration estimation support device described in the above-described embodiments 1 and 2. This makes it possible to estimate the duration of atrial fibrillation based on electrocardiogram information of a patient with atrial fibrillation.

[0050] [Embodiment 4] The atrial fibrillation duration estimation support program of this embodiment is a program for causing a computer to execute each step of the above-mentioned atrial fibrillation duration estimation support method. Specifically, the atrial fibrillation duration estimation support program of this embodiment is a program for causing a computer to execute an electrocardiogram information acquisition procedure, a duration estimation procedure, and an output procedure.

[0051] The method includes an electrocardiogram information acquisition step, a duration estimation step, and an output step, The electrocardiogram information acquisition step acquires electrocardiogram information of a patient with atrial fibrillation, the duration estimation step includes estimating a duration of atrial fibrillation of the atrial fibrillation patient based on the electrocardiogram information; The output step outputs the duration.

[0052] The duration estimation assistance program of the present embodiment can also be said to be a program that causes a computer to function as an electrocardiogram information acquisition procedure, a duration estimation procedure, and an output procedure.

[0053] The duration estimation support program of this embodiment can be implemented by using the description of the atrial fibrillation duration estimation support device and the atrial fibrillation duration estimation support method of the present disclosure. In each of the steps, for example, "step" can be read as "processing". In addition, the program of this embodiment may be recorded in a computer-readable recording medium. The recording medium is not particularly limited, and examples thereof include a random access memory (RAM), a read-only memory (ROM), a hard disk (HD), a flash memory (for example, a solid state drive (SSD), a USB flash memory, an SD / SDHC card, etc.), an optical disk (for example, a CD-R / CD-RW, a DVD-R / DVD-RW, a BD-R / BD-RE, etc.), a magneto-optical disk (MO), a floppy disk (FD), etc. In addition, the duration estimation support program of this embodiment (for example, also referred to as a programming product or a duration estimation support program product) may be distributed from an external computer, for example. The "distribution" may be, for example, distribution via a communication network or distribution via a device connected by a wire. The duration estimation assistance program of the present embodiment may be installed and executed in the device to which it is distributed, or may be executed without being installed.

[0054] [Embodiment 5] The trained model creation program for estimating the duration of atrial fibrillation of this embodiment is a program for causing a computer to execute each step of the trained model creation method for estimating the duration of atrial fibrillation described above. Specifically, the trained model creation program for estimating the duration of atrial fibrillation of this embodiment is a program for causing a computer to execute a training information acquisition unit and a trained model generation unit.

[0055] The learning information acquisition step acquires learning information, the learning information includes electrocardiogram information of an atrial fibrillation patient and a duration of atrial fibrillation of the atrial fibrillation patient; The trained model generation procedure uses machine learning with the duration as a correct label for the electrocardiogram information to generate a duration estimation model as a trained model that outputs the duration of atrial fibrillation of the subject when the subject's electrocardiogram information is input.

[0056] In addition, the program for creating a trained model for duration estimation in this embodiment can also be said to be a program that causes a computer to function as a learning information acquisition procedure and a trained model generation procedure.

[0057] The program for producing a trained model for estimating duration of the present embodiment may be described in the device for producing a trained model for estimating duration of atrial fibrillation and the method for producing a trained model for estimating duration of atrial fibrillation of the present disclosure. In each of the steps, for example, "step" can be read as "processing". The program of the present embodiment may be recorded in a computer-readable recording medium. The recording medium is not particularly limited, and examples thereof include random access memory (RAM), read-only memory (ROM), hard disk (HD), flash memory (for example, SSD (Solid State Drive), USB flash memory, SD / SDHC card, etc.), optical disk (for example, CD-R / CD-RW, DVD-R / DVD-RW, BD-R / BD-RE, etc.), magneto-optical disk (MO), floppy (registered trademark) disk (FD), etc. In addition, the program for producing a trained model for estimating duration of the present embodiment (for example, also referred to as a programming product or a program product for producing a trained model for estimating duration) may be in a form distributed from an external computer, for example. The "distribution" may be, for example, distribution via a communication network or distribution via a device connected by a wire. The program for creating a trained model for duration estimation of the present embodiment may be installed and executed in the device to which it is distributed, or may be executed without being installed. EXAMPLES

[0058] Next, examples of the present disclosure will be described. However, the present disclosure is not limited to the following examples. Commercially available reagents and equipment were used according to their protocols unless otherwise indicated.

[0059] [Example 1] It was confirmed that the duration estimation support device disclosed herein can estimate the duration of atrial fibrillation in patients.

[0060] (1) Patients with atrial fibrillation First, 272 patients aged 20 to 90 years old who were admitted to the Japanese Red Cross Kyoto Second Hospital for their first catheter ablation (CA) between January 1, 2015 and December 31, 2023, and whose electrocardiogram, patient background, echocardiogram, and test data were available within the following three months, were included in the study. We excluded 14 patients who had pacemaker rhythm recorded on their electrocardiogram, 30 patients whose QRS waves were difficult to exclude or whose extracted F waves were too short for characteristic analysis, and 39 patients whose duration of atrial fibrillation was difficult to determine, resulting in a total of 189 patients being included in the study.

[0061] Next, each patient was classified as having PeAF (atrial fibrillation duration less than 1 year) or LsPeAF (long standing persistent atrial fibrillation, atrial fibrillation duration 1 year or more) according to the following criteria. Patients who did not meet any of the following criteria were excluded as cases of unknown duration. PeAF (atrial fibrillation lasting less than 1 year) Patients who had a continuous record of atrial fibrillation after a sinus rhythm recorded on an electrocardiogram within one year, excluding electrocardiograms after defibrillation, or who had a continuous record of atrial fibrillation after clinical symptoms or arrhythmias clearly related to atrial fibrillation within one year. LsPeAF (atrial fibrillation lasting more than 1 year) Patients with electrocardiogram records of atrial fibrillation confirmed for more than one year consecutively, or patients who have had clinical symptoms or arrhythmias clearly related to atrial fibrillation for more than one year prior and have had electrocardiogram records of atrial fibrillation confirmed for more than one year thereafter.

[0062] Tables 1 to 3 below show information on the test patients. Table 1 below shows the age, sex, height, weight, body mass index (BMI), and body surface area (BSA) of the test patients (all patients, PeAF patients, and LsPeAF patients). Table 2 below shows the medical history of the test patients (all patients, PeAF patients, and LsPeAF patients), and shows the number (n) and percentage (%) of heart failure, stroke, hypertension, diabetes, dyslipidemia, chronic kidney disease (CKD), and chronic pulmonary disease (CPD) to the total. Table 3 below shows echocardiographic data of the test patients, including left atrial diameter (LAD), ejection fraction, and mitral regurgitation (MR).

[0063] [Table 1]

[0064] [Table 2]

[0065] [Table 3]

[0066] (2) Electrocardiogram data analysis For 189 test patients, 10 seconds of electrocardiogram data was analyzed using a 12-lead electrocardiogram recording device (FCP-8800 or FCP-8700, both manufactured by Fukuda Denshi Co., Ltd.), and the heart rate (HR), QRS width, S1 amplitude, RV5 amplitude, and R+S (=S1 amplitude+RV5 amplitude) of each patient were obtained. Table 4 below shows the electrocardiogram data of the test patients (all patients, PeAF patients, and LsPeAF patients). Table 4 is a table showing the electrocardiogram data of the test patients, and shows the heart rate (Heart Rate), QRS width (QRS width), SV1 amplitude (SV1 amplitude), RV5 amplitude (RV5 amplitude), and SV1 amplitude and RV5 amplitude (R+S: SV1 amplitude+RV5 amplitude).

[0067] [Table 4]

[0068] Next, F waves were extracted from 10 seconds of ECG data. Since F waves are in the 4-9 Hz band, the cutoff frequency was set to 0.8-40 Hz. For QRST cancellation, QRST intervals were detected, and R wave times were peak-detected using the Pan-Tompkins algorithm. Q wave times were calculated by subtracting 37 ms, which is a typical ventricular activation time, from the R wave times. T wave times were calculated by adding 200 ms to the R wave times. For each QRST interval, f waves were extracted using principal component analysis. Since f waves are in the 4-9 Hz band, the cutoff frequency was set to 3-20 Hz. An example of an extracted F wave is shown in Figure 9.

[0069] Next, the characteristics of the extracted F-waves were analyzed. The root mean square (RMS) of the amplitude was defined as the square root of the mean square, which is the arithmetic mean of the squares of the signal amplitudes in the time domain. Sample entropy (SampEn) was used to evaluate the irregularity of the f-waves. SampEn is an index of entropy and is a method designed to reduce the bias of approximate entropy and obtain more robust statistics. The definition of sample entropy is shown in Equation 1 below. The threshold was 3.5 and the sample size was 3. After fast Fourier transformation, the dominant frequency (DF) and organization index (OI) were obtained. DF is one of the most widely used indices for frequency analysis of F-waves, and the frequency with the highest power value in the frequency distribution was defined as DF. OI was used as an index of organization of F-waves. OI was defined as the ratio of the area under the highest peak and its harmonics (not including the fifth harmonic peak) to the remainder of the spectrum in the 3-15 Hz band.

[0070]

number

[0071] Tables 5 to 8 show the F wave characteristics of the study patients (all patients, PeAF patients, and LsPeAF patients).

[0072] [Table 5]

[0073] [Table 6]

[0074] [Table 7]

[0075] [Table 8]

[0076] (3) Building a trained model (duration estimation model) Of the 189 test patients, electrocardiogram information from 145 patients was used as the training data group, and electrocardiogram information from the remaining 44 patients was used as the test data group to build a model to predict PeAF or LsPeAF. Specifically, a binary classification was performed using patients with LsPeAF=1 and patients with PeAF=0 as the objective variable, and the model was trained using a Gradient Boosting Decision Tree (GBDT). GBDT implemented XGBoost (eXtreme Gradient Boost). The hyperparameters of XGBoost were as follows: Learning rate: 0.3703 Max depth: 2 Min child weight: 2 subsample: 0.9100 colsample_bytree: 0.1043 reg lambda: 0.0056 reg alpha: 0.0041 gamma: 0.4426

[0077] A stratified 5-fold cross-validation was used to evaluate the model, so that the balance of classes was maintained in each division. Specifically, the dataset was divided into 5 parts, and each part was used as validation data, and the remaining part was used as training data, and this was repeated 5 times. For the electrocardiogram information used for training, variables with p-values ​​less than 0.05 in univariate analysis were extracted from various information on the test subjects shown in Tables 1 to 8, and information selected by the stepwise method was used. In addition, features with high correlation coefficients were excluded to eliminate multicollinearity. AUC was calculated using an ROC curve, and the cutoff value for continuous variables was calculated based on the maximum Youden index. Python version 3.10.12 (Python Software Foundation, Wilmington, DE, US) was used for statistical analysis.

[0078] (4) Evaluation of predictive accuracy of trained model (duration estimation model) As an evaluation method, the accuracy of the model was evaluated using a test data group. Five different random seeds were set, the average ROC curve-AUC of the model for each setting was calculated, and the optimal cutoff point was selected using the Youden index. The average value of the ROC curve-AUC was 0.82. In addition, the SHARP value was used to interpret the model, and the importance of each feature was evaluated. In the duration estimation model disclosed herein, heart rate (HR) was treated as the most important feature. The importance of other features was in the order of RMS, SampEn, amplitude, left ventricular ejection fraction, sex, orthostatic intolerance, obesity level, QRS wave amplitude, QRS width, left atrial diameter (LAD), and age.

[0079] The prediction results of the test data by the duration estimation model are shown in Table 9. As shown in Table 9, the duration estimation model of the present disclosure had a prediction accuracy (Accuracy) of 0.82, a sensitivity (Sensitivity) of 0.72, and a specificity (Specificity) of 0.89 for the test data, and was able to estimate the duration of atrial fibrillation patients with high accuracy.

[0080] [Table 9]

[0081] (5) Comparison of the prediction accuracy of the trained model (duration estimation model) with that of doctors Ten cardiologists (Group A) were asked to diagnose the electrocardiogram data of 44 test patients as test data, and to answer whether each patient had atrial fibrillation that had lasted for at least one year based on the patient's information (Phase 1). After that, each doctor was presented with the predicted duration by the duration estimation model of the present disclosure and the SHAP value calculated by the contribution of each variable (feature amount) to the predicted result of the model, and then asked to answer again (Phase 2). Furthermore, the remaining 10 patients (Group B) were shown the diagnosis results of the doctors in Group A, and after recognizing the doctors' diagnostic ability for atrial fibrillation duration in advance, a two-phase test was conducted in which they were asked to answer Phase 1 and Phase 2 in the same way as Group A. The results are shown in FIG. 10. FIG. 10 is a graph showing the diagnostic accuracy by doctors in Group A and Group B. In the graphs for Group A and Group B, the black circles connected by solid lines show the results in Phase 1 and Phase 2 by the same doctor. Moreover, the prediction accuracy by the duration estimation model (ML model) of the present disclosure is shown in the graph of Fig. 10. As shown in Fig. 10, it was found that the duration estimation model of the present disclosure can estimate the duration of an atrial fibrillation patient with higher accuracy than the accuracy of diagnosis by a doctor.

[0082] The correct answer rate of Group A was 63.9 ± 9.6% in Phase 1, but improved to 71.6 ± 9.3% in Phase 2 (p = 0.01). The correct answer rate of Group B was 59.8 ± 5.3% in Phase 1, but improved to 68.2 ± 5.9% in Phase 2 (p < 0.01). In Phase 2, there was no difference in the correct answer rate between Groups A and B (p = 0.48). The average percentage of doctors who gave answers different from the prediction of the duration estimation model of the present disclosure in Phase 2 (answers in which the prediction of the duration estimation model of the present disclosure was correct and the doctor gave an incorrect answer) was 17.3 ± 10.3% in Group A and 20.9 ± 5.0% in Group B. Therefore, there was no significant difference in the increase in the correct answer rate between Groups A and B (p = 0.85). In other words, it has been found that the duration estimation model disclosed herein can significantly improve the diagnostic accuracy of cardiologists.

[0083] Although the present disclosure has been described above with reference to the embodiments, the present disclosure is not limited to the above-mentioned embodiments. Various modifications that can be understood by a person skilled in the art can be made to the configuration and details of the present disclosure within the scope of the present disclosure.

[0084] This application claims priority based on Japanese Patent Application No. 2023-183282, filed on October 25, 2023, the disclosure of which is incorporated herein in its entirety.

[0085] <Additional Notes> Some or all of the above embodiments may be described as follows, but are not limited to the following supplementary notes. (Appendix 1) The device includes an electrocardiogram information acquisition unit, a duration estimation unit, and an output unit, The electrocardiogram information acquisition unit acquires electrocardiogram information of a patient having atrial fibrillation, the duration estimation unit estimates a duration of atrial fibrillation of the atrial fibrillation patient based on the electrocardiogram information; The output unit outputs the duration. (Appendix 2) 2. The duration estimation support device according to claim 1, wherein the duration estimation unit inputs the electrocardiogram information to a duration estimation model to estimate a duration of atrial fibrillation of the atrial fibrillation patient. (Appendix 3) the electrocardiogram information acquisition unit acquires patient information of the atrial fibrillation patient in association with the electrocardiogram information; The patient information includes at least one selected from the group consisting of attribute information of the atrial fibrillation patient, interview information, and echocardiographic information; 3. The duration estimation assistance device according to claim 1, wherein the duration estimation unit estimates a duration of atrial fibrillation of the atrial fibrillation patient based on the electrocardiogram information and the patient information. (Appendix 4) A recurrence prediction unit is included, The recurrence prediction unit predicts a recurrence rate of the atrial fibrillation patient based on at least one of the electrocardiogram information and the duration, 4. The duration estimation assistance device according to claim 1, wherein the output unit outputs the recurrence rate. (Appendix 5) A learning information acquisition unit and a trained model generation unit are included, The learning information acquisition unit acquires learning information, the learning information includes electrocardiogram information of an atrial fibrillation patient and a duration of atrial fibrillation of the atrial fibrillation patient; The trained model generation unit is a trained model manufacturing device for estimating the duration of atrial fibrillation, which generates a duration estimation model that outputs the duration of atrial fibrillation of a subject when electrocardiogram information of the subject is input, as a trained model through machine learning using the duration as a correct label for the electrocardiogram information. (Appendix 6) the learning information includes patient information of the atrial fibrillation patient, The patient information includes at least one selected from the group consisting of attribute information of the atrial fibrillation patient, interview information, and echocardiographic information; The trained model generation unit of the device for producing a trained model for duration estimation described in Appendix 5 generates, as a trained model, a duration estimation model that outputs the duration of atrial fibrillation of a subject when the subject's electrocardiogram information and patient information are input, by machine learning using the duration as a correct label for a combination of the electrocardiogram information and the patient information. (Appendix 7) The method includes an electrocardiogram information acquiring step, a duration estimating step, and an output step, The electrocardiogram information acquiring step acquires electrocardiogram information of a patient having atrial fibrillation, the duration estimation step estimates a duration of atrial fibrillation of the atrial fibrillation patient based on the electrocardiogram information; The output step outputs the duration, A method for assisting in estimation of duration of atrial fibrillation, wherein each of the steps is executed by a computer. (Appendix 8) 8. The duration estimation support method according to claim 7, wherein the duration estimation step inputs the electrocardiogram information to a duration estimation model to estimate the duration of atrial fibrillation of the atrial fibrillation patient. (Appendix 9) the electrocardiogram information acquiring step acquires patient information of the atrial fibrillation patient in association with the electrocardiogram information; The patient information includes at least one selected from the group consisting of attribute information of the atrial fibrillation patient, interview information, and echocardiographic information; 9. The duration estimation support method according to claim 7, wherein the duration estimation step estimates a duration of atrial fibrillation of the atrial fibrillation patient based on the electrocardiogram information and the patient information. (Appendix 10) A recurrence prediction step is included, the recurrence prediction step predicts a recurrence rate of the atrial fibrillation patient based on at least one of the electrocardiogram information and the duration; The duration estimation support method according to any one of appendices 7 to 9, wherein the output step outputs the recurrence rate. (Appendix 11) The method includes a learning information acquisition step and a trained model generation step, The learning information acquiring step acquires learning information, the learning information includes electrocardiogram information of an atrial fibrillation patient and a duration of atrial fibrillation of the atrial fibrillation patient; The trained model generation process is a method for producing a trained model for estimating the duration of atrial fibrillation, which generates a duration estimation model as a trained model by machine learning using the duration as a correct label for the electrocardiogram information, when the electrocardiogram information of a subject is input. (Appendix 12) the learning information includes patient information of the atrial fibrillation patient, The patient information includes at least one selected from the group consisting of attribute information of the atrial fibrillation patient, interview information, and echocardiographic information; The trained model generation process of the method for producing a trained model for duration estimation described in Appendix 11 generates, as a trained model, a duration estimation model that outputs the duration of atrial fibrillation of a subject when the subject's electrocardiogram information and patient information are input, by machine learning using the duration as a correct answer label for a combination of the electrocardiogram information and the patient information. (Appendix 13) The method includes an electrocardiogram information acquisition step, a duration estimation step, and an output step, The electrocardiogram information acquisition step acquires electrocardiogram information of a patient with atrial fibrillation, the duration estimation step includes estimating a duration of atrial fibrillation of the atrial fibrillation patient based on the electrocardiogram information; The output step outputs the duration; An atrial fibrillation duration estimation support program for causing a computer to execute each of the above procedures. (Appendix 14) The duration estimation support program according to claim 13, wherein the duration estimation step inputs the electrocardiogram information into a duration estimation model to estimate the duration of atrial fibrillation of the atrial fibrillation patient. (Appendix 15) the electrocardiogram information acquisition step acquires patient information of the atrial fibrillation patient in association with the electrocardiogram information; The patient information includes at least one selected from the group consisting of attribute information of the atrial fibrillation patient, interview information, and echocardiographic information; 15. The duration estimation support program according to claim 13, wherein the duration estimation step estimates a duration of atrial fibrillation of the atrial fibrillation patient based on the electrocardiogram information and the patient information. (Appendix 16) Includes a recurrence prediction procedure; the recurrence prediction step predicts a recurrence rate of the atrial fibrillation patient based on at least one of the electrocardiogram information and the duration; 16. The program for estimating disease duration according to any one of appendices 13 to 15, wherein the output step outputs the recurrence rate. (Appendix 17) The method includes a learning information acquisition procedure and a trained model generation procedure, The learning information acquisition step acquires learning information, the learning information includes electrocardiogram information of an atrial fibrillation patient and a duration of atrial fibrillation of the atrial fibrillation patient; the trained model generation step generates, as a trained model, a duration estimation model that outputs the duration of atrial fibrillation of a subject when electrocardiogram information of the subject is input, by machine learning using the duration as a correct answer label for the electrocardiogram information; A program for creating a trained model for estimating the duration of atrial fibrillation, for causing a computer to execute each of the above steps. (Appendix 18) the learning information includes patient information of the atrial fibrillation patient, The patient information includes at least one selected from the group consisting of attribute information of the atrial fibrillation patient, interview information, and echocardiographic information; The trained model generation procedure is a program for creating a trained model for duration estimation described in Appendix 17, which generates a duration estimation model as a trained model by machine learning using the duration as a correct label for a combination of the electrocardiogram information and the patient information, when the electrocardiogram information and patient information of a subject are input. (Appendix 19) The method includes an electrocardiogram information acquisition step, a duration estimation step, and an output step, The electrocardiogram information acquisition step acquires electrocardiogram information of a patient with atrial fibrillation, the duration estimation step includes estimating a duration of atrial fibrillation of the atrial fibrillation patient based on the electrocardiogram information; The output step outputs the duration; A computer-readable recording medium having recorded thereon an atrial fibrillation duration estimation assistance program for causing a computer to execute each of the above procedures. (Appendix 20) 20. The recording medium of claim 19, wherein the duration estimation step inputs the electrocardiogram information into a duration estimation model to estimate a duration of atrial fibrillation of the atrial fibrillation patient. (Appendix 21) the electrocardiogram information acquisition step acquires patient information of the atrial fibrillation patient in association with the electrocardiogram information; The patient information includes at least one selected from the group consisting of attribute information of the atrial fibrillation patient, interview information, and echocardiographic information; 21. The recording medium according to claim 19, wherein the duration estimation step estimates a duration of atrial fibrillation of the atrial fibrillation patient based on the electrocardiogram information and the patient information. (Appendix 22) Includes a recurrence prediction procedure; the recurrence prediction step predicts a recurrence rate of the atrial fibrillation patient based on at least one of the electrocardiogram information and the duration; 22. The recording medium according to any one of appendices 19 to 21, wherein the output step outputs the recurrence rate. (Appendix 23) The method includes a learning information acquisition procedure and a trained model generation procedure, The learning information acquisition step acquires learning information, the learning information includes electrocardiogram information of an atrial fibrillation patient and a duration of atrial fibrillation of the atrial fibrillation patient; the trained model generation step generates, as a trained model, a duration estimation model that outputs the duration of atrial fibrillation of a subject when electrocardiogram information of the subject is input, by machine learning using the duration as a correct answer label for the electrocardiogram information; A computer-readable recording medium recording a program for creating a trained model for estimating the duration of atrial fibrillation, for causing a computer to execute each of the above procedures. (Appendix 24) the learning information includes patient information of the atrial fibrillation patient, The patient information includes at least one selected from the group consisting of attribute information of the atrial fibrillation patient, interview information, and echocardiographic information; The recording medium described in Appendix 23, wherein the trained model generation procedure generates, as a trained model, a duration estimation model that outputs the duration of atrial fibrillation of a subject when the subject's electrocardiogram information and patient information are input, by machine learning using the duration as a correct answer label for a combination of the electrocardiogram information and the patient information. [Industrial Applicability]

[0086] According to the present disclosure, it is possible to estimate the duration of atrial fibrillation in a patient suffering from atrial fibrillation based on electrocardiogram information. Therefore, the present disclosure is useful, for example, in the medical field. [Explanation of symbols]

[0087] 10, 10A Atrial fibrillation duration estimation support device 11 Electrocardiogram information acquisition unit 12 Duration Estimation Unit 13 Output section 14 Recurrence prediction section 101 Central Processing Unit 102 Memory 103 Bus 104 Storage device 105 Input Device 106 Output Device 107 Communication Devices 20. Device for creating trained models for estimating the duration of atrial fibrillation 21 Learning information acquisition unit 22 Trained model generation unit 201 Central Processing Unit 202 Memory 203 Bus 204 Storage device 205 Input Device 206 Output Device 207 Communication Devices

Claims

1. The device includes an electrocardiogram information acquisition unit, a duration estimation unit, and an output unit, The electrocardiogram information acquisition unit acquires electrocardiogram information of a patient having atrial fibrillation, the duration estimation unit estimates a duration of atrial fibrillation of the atrial fibrillation patient based on the electrocardiogram information; The output unit outputs the duration.

2. 2. The duration estimation support device according to claim 1, wherein the duration estimation unit inputs the electrocardiogram information to a duration estimation model to estimate the duration of atrial fibrillation of the atrial fibrillation patient.

3. the electrocardiogram information acquisition unit acquires patient information of the atrial fibrillation patient in association with the electrocardiogram information; The patient information includes at least one selected from the group consisting of attribute information of the atrial fibrillation patient, interview information, and echocardiographic information; 3. The duration estimation assistance device according to claim 1, wherein the duration estimation unit estimates a duration of atrial fibrillation of the atrial fibrillation patient based on the electrocardiogram information and the patient information.

4. A recurrence prediction unit is included, The recurrence prediction unit predicts a recurrence rate of the atrial fibrillation patient based on at least one of the electrocardiogram information and the duration, The duration estimation assistance device according to claim 1 , wherein the output unit outputs the recurrence rate.

5. A learning information acquisition unit and a trained model generation unit are included, The learning information acquisition unit acquires learning information, the learning information includes electrocardiogram information of an atrial fibrillation patient and a duration of atrial fibrillation of the atrial fibrillation patient; The trained model generation unit is a trained model manufacturing device for estimating the duration of atrial fibrillation, which generates a duration estimation model that outputs the duration of atrial fibrillation of a subject when electrocardiogram information of the subject is input, as a trained model through machine learning using the duration as a correct label for the electrocardiogram information.

6. the learning information includes patient information of the atrial fibrillation patient, The patient information includes at least one selected from the group consisting of attribute information of the atrial fibrillation patient, interview information, and echocardiographic information; The trained model generation unit of the device for producing a trained model for duration estimation as described in claim 5 generates a duration estimation model as a trained model that outputs the duration of atrial fibrillation of a subject when the subject's electrocardiogram information and patient information are input, by machine learning using the duration as a correct answer label for a combination of the electrocardiogram information and the patient information.

7. The method includes an electrocardiogram information acquiring step, a duration estimating step, and an output step, The electrocardiogram information acquiring step acquires electrocardiogram information of a patient having atrial fibrillation, the duration estimation step estimates a duration of atrial fibrillation of the atrial fibrillation patient based on the electrocardiogram information; The output step outputs the duration, A method for assisting in estimation of duration of atrial fibrillation, wherein each of the steps is executed by a computer.

8. The method includes a learning information acquisition step and a trained model generation step, The learning information acquiring step acquires learning information, the learning information includes electrocardiogram information of an atrial fibrillation patient and a duration of atrial fibrillation of the atrial fibrillation patient; The trained model generation process is a method for producing a trained model for estimating the duration of atrial fibrillation, which generates a duration estimation model as a trained model by machine learning using the duration as a correct label for the electrocardiogram information, when the electrocardiogram information of a subject is input.

9. The method includes an electrocardiogram information acquisition step, a duration estimation step, and an output step, The electrocardiogram information acquisition step acquires electrocardiogram information of a patient with atrial fibrillation, the duration estimation step includes estimating a duration of atrial fibrillation of the atrial fibrillation patient based on the electrocardiogram information; The output step outputs the duration; An atrial fibrillation duration estimation support program for causing a computer to execute each of the above procedures.

10. The method includes a learning information acquisition procedure and a trained model generation procedure, The learning information acquisition step acquires learning information, the learning information includes electrocardiogram information of an atrial fibrillation patient and a duration of atrial fibrillation of the atrial fibrillation patient; the trained model generation step generates, as a trained model, a duration estimation model that outputs the duration of atrial fibrillation of a subject when electrocardiogram information of the subject is input, by machine learning using the duration as a correct answer label for the electrocardiogram information; A program for creating a trained model for estimating the duration of atrial fibrillation, for causing a computer to execute each of the above steps.

11. The method includes an electrocardiogram information acquisition step, a duration estimation step, and an output step, The electrocardiogram information acquisition step acquires electrocardiogram information of a patient with atrial fibrillation, the duration estimation step includes estimating a duration of atrial fibrillation of the atrial fibrillation patient based on the electrocardiogram information; The output step outputs the duration; A computer-readable recording medium having recorded thereon an atrial fibrillation duration estimation assistance program for causing a computer to execute each of the above procedures.

12. The method includes a learning information acquisition procedure and a trained model generation procedure, The learning information acquisition step acquires learning information, the learning information includes electrocardiogram information of an atrial fibrillation patient and a duration of atrial fibrillation of the atrial fibrillation patient; the trained model generation step generates, as a trained model, a duration estimation model that outputs the duration of atrial fibrillation of a subject when electrocardiogram information of the subject is input, by machine learning using the duration as a correct answer label for the electrocardiogram information; A computer-readable recording medium recording a program for creating a trained model for estimating the duration of atrial fibrillation, for causing a computer to execute each of the above procedures.

Images