Atrial fibrillation prediction support program, atrial fibrillation prediction support device, atrial fibrillation prediction support method, atrial fibrillation prediction estimation pre-trained model manufacturing device, atrial fibrillation prediction estimation pre-trained model manufacturing method, atrial fibrillation prediction estimation pre-trained model manufacturing program, and recording medium

Abstract

This program provides support for detecting early signs of atrial fibrillation, enabling easy estimation of these signs. [Solution] The atrial fibrillation prediction support program of this disclosure includes a blood pressure measurement information acquisition procedure, a prediction estimation procedure, and an output procedure, The aforementioned blood pressure measurement information acquisition procedure acquires the blood pressure measurement information of the subject, The aforementioned precursor estimation procedure estimates the precursor of atrial fibrillation in the subject based on the blood pressure measurement information, The output procedure outputs the estimated results of the atrial fibrillation precursor.

Description

Technical Field

[0001] The present disclosure relates to an atrial fibrillation prediction detection support program, an atrial fibrillation prediction detection support device, an atrial fibrillation prediction detection support method, a trained model manufacturing device for predicting atrial fibrillation, a trained model manufacturing method for predicting atrial fibrillation, a trained model manufacturing program for predicting atrial fibrillation, and a recording medium.

Background Art

[0002] In recent years, with the progress of the super-aged society, the number of patients with atrial fibrillation has been increasing. In the treatment of atrial fibrillation, it is important to detect atrial fibrillation at an early stage. Therefore, Patent Document 1 describes a method for detecting atrial fibrillation from an electrocardiogram, which includes a preprocessing step of adapting an electrocardiogram signal obtained from the electrocardiogram to a CNN, a first CNN step of extracting an abnormality candidate from the electrocardiogram, and a second CNN step of detecting the atrial fibrillation from the abnormality candidate.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] However, in the initial stage of atrial fibrillation, since the onset occurs irregularly, it is difficult to predict the occurrence of the onset.

[0005] Therefore, an object of the present disclosure is to provide an atrial fibrillation prediction detection support program, an atrial fibrillation prediction detection support device, an atrial fibrillation prediction detection support method, a trained model manufacturing device for predicting atrial fibrillation, a trained model manufacturing method for predicting atrial fibrillation, a trained model manufacturing program for predicting atrial fibrillation, and a recording medium that can easily estimate the prediction of atrial fibrillation. [Means for solving the problem]

[0006] To achieve the aforementioned objectives, the atrial fibrillation prediction support program of this disclosure is: This includes a procedure for acquiring blood pressure measurement information, a procedure for predicting a potential hazard, and an output procedure. The aforementioned blood pressure measurement information acquisition procedure acquires the blood pressure measurement information of the subject, The aforementioned precursor estimation procedure estimates the precursor of atrial fibrillation in the subject based on the blood pressure measurement information, The output procedure outputs the estimated result of the atrial fibrillation precursor. This is a program that assists in detecting atrial fibrillation by having a computer execute each step.

[0007] The atrial fibrillation prediction support device disclosed herein is It includes a blood pressure measurement information acquisition unit, a prediction unit, and an output unit. The blood pressure measurement information acquisition unit acquires the blood pressure measurement information of the subject, The aforementioned precursor estimation unit estimates the precursor of atrial fibrillation in the subject based on the blood pressure measurement information, The output unit outputs the estimated result of the atrial fibrillation precursor.

[0008] The atrial fibrillation prediction support method disclosed herein is: This includes a blood pressure measurement information acquisition process, a prediction estimation process, and an output process. The blood pressure measurement information acquisition step involves acquiring the blood pressure measurement information of the subject, The aforementioned precursor estimation step estimates the precursor of atrial fibrillation in the subject based on the blood pressure measurement information, The output step outputs the estimated result of the atrial fibrillation precursor. This is a method for supporting the detection of atrial fibrillation, in which each step is performed by a computer.

[0009] The recording medium disclosed herein is This includes a procedure for acquiring blood pressure measurement information, a procedure for predicting a potential hazard, and an output procedure. The aforementioned blood pressure measurement information acquisition procedure acquires the blood pressure measurement information of the subject, The aforementioned precursor estimation procedure estimates the precursor of atrial fibrillation in the subject based on the blood pressure measurement information, The output procedure outputs the estimated result of the atrial fibrillation precursor. This is a computer-readable recording medium that stores a program to assist in the detection of atrial fibrillation, which instructs a computer to execute each step.

[0010] The pre-trained model manufacturing apparatus for predicting atrial fibrillation according to this disclosure is It includes a learning information acquisition unit and a trained model generation unit, The aforementioned learning information acquisition unit acquires learning information, The aforementioned learning information includes the subject's blood pressure measurement information and whether or not the subject has developed atrial fibrillation. The pre-trained model generation unit generates a pre-trained model that, when the blood pressure measurement information of a subject is input, outputs an estimated result of the subject's atrial fibrillation precursors by machine learning using the presence or absence of atrial fibrillation as the correct label for the blood pressure measurement information.

[0011] The method for manufacturing a trained model for predicting atrial fibrillation according to this disclosure is: This includes a process for acquiring training information and a process for generating a trained model. The aforementioned learning information acquisition process acquires learning information, The aforementioned learning information includes the subject's blood pressure measurement information and whether or not the subject has developed atrial fibrillation. The pre-trained model generation step generates a pre-trained model that, when the blood pressure measurement information of a subject is input, outputs an estimated result of the subject's atrial fibrillation precursors using machine learning with the presence or absence of atrial fibrillation as the correct label for the blood pressure measurement information.

[0012] The pre-trained model manufacturing program for predicting atrial fibrillation symptoms described herein is This includes procedures for acquiring training information and generating a trained model. The above procedure for acquiring learning information involves acquiring learning information, The learning information includes the blood pressure measurement information of the subject and the presence or absence of the onset of atrial fibrillation in the subject. The learned model generation procedure is a learned model manufacturing program for atrial fibrillation prediction estimation for causing a computer to execute each procedure of generating, as a learned model, a prediction estimation model that outputs an estimation result of a sign of atrial fibrillation in a subject when the blood pressure measurement information of the subject is input by machine learning using the presence or absence of the onset of atrial fibrillation as a correct label for the blood pressure measurement information.

[0013] The recording medium of the present disclosure includes a learning information acquisition procedure and a learned model generation procedure. The learning information acquisition procedure acquires learning information. The learning information includes the blood pressure measurement information of the subject and the presence or absence of the onset of atrial fibrillation in the subject. The learned model generation procedure generates, as a learned model, a prediction estimation model that outputs an estimation result of a sign of atrial fibrillation in a subject when the blood pressure measurement information of the subject is input by machine learning using the presence or absence of the onset of atrial fibrillation as a correct label for the blood pressure measurement information. It is a computer-readable recording medium that records a learned model manufacturing program for atrial fibrillation prediction estimation for causing a computer to execute each procedure.

Advantages of the Invention

[0014] According to the present disclosure, the sign of atrial fibrillation can be easily estimated.

Brief Description of the Drawings

[0015] [Figure 1] FIG. 1 is a block diagram showing a configuration example of an atrial fibrillation prediction detection support device of the present disclosure. [Figure 2] FIG. 2 is a block diagram showing an example of the hardware configuration of the atrial fibrillation prediction detection support device of the present disclosure. [Figure 3] FIG. 3 is a flowchart showing an example of processing in the atrial fibrillation prediction detection support device of the present disclosure. <00001 [Figure 4] Figure 4 is a block diagram showing an example configuration of the trained model manufacturing apparatus for predicting atrial fibrillation according to the present disclosure. [Figure 5] Figure 5 is a block diagram showing an example of the hardware configuration of the trained model manufacturing apparatus for predicting atrial fibrillation according to the present disclosure. [Figure 6] Figure 6 is a flowchart showing an example of processing in the trained model manufacturing apparatus for predicting atrial fibrillation according to this disclosure. [Figure 7] Figure 7 is a graph showing the results of the example. [Figure 8] Figure 8 is a graph showing the results of the example. [Modes for carrying out the invention]

[0016] Next, embodiments of the present disclosure will be described with reference to the drawings. The present disclosure is not limited to the following embodiments. In the following drawings, the same parts are denoted by the same reference numerals. Furthermore, unless otherwise specified, the descriptions of each embodiment can be used interchangeably with those of the others, and unless otherwise specified, the configurations of each embodiment can be combined.

[0017] [Embodiment 1] The atrial fibrillation prediction support program of this disclosure is a program that causes a computer to execute a blood pressure measurement information acquisition procedure, a prediction estimation procedure, and an output procedure. The atrial fibrillation prediction support program of this disclosure can also be described as a program that causes a computer to function as the blood pressure measurement information acquisition procedure, the prediction estimation procedure, and the output procedure. Furthermore, the atrial fibrillation prediction support program of this disclosure can also be described as a program that causes a computer to execute each step of the atrial fibrillation prediction support method described later.

[0018] The aforementioned blood pressure measurement information acquisition procedure acquires the blood pressure measurement information of the subject, The aforementioned precursor estimation procedure estimates the precursor of atrial fibrillation in the subject based on the blood pressure measurement information, The output procedure outputs the estimated results of the atrial fibrillation precursor.

[0019] Each of the aforementioned procedures can be reinterpreted, for example, by substituting "procedure" with "process." The program of this disclosure may also be recorded on a computer-readable recording medium. The recording medium is not particularly limited and includes, for example, random access memory (RAM), read-only memory (ROM), hard disk (HD), flash memory (e.g., SSD (Solid State Drive), USB flash memory, SD / SDHC card, etc.), optical disc (e.g., CD-R / CD-RW, DVD-R / DVD-RW, BD-R / BD-RE, etc.), magneto-optical disk (MO), floppy disk (FD), etc. The atrial fibrillation prediction support program of this disclosure (for example, also referred to as a programming product or atrial fibrillation prediction support program product) may also be delivered, for example, from an external computer. The "delivery" may be, for example, delivered via a communication network or delivered via a wired device. The atrial fibrillation prediction support program of this disclosure may be installed and executed on the delivered device, or it may be executed without being installed.

[0020] Next, the atrial fibrillation prediction support device of this disclosure will be described with reference to Figure 1. Figure 1 is a block diagram showing the configuration of an example of the atrial fibrillation prediction support device 10 of this disclosure. As shown in Figure 1, the atrial fibrillation prediction support device 10 (hereinafter also referred to as "this device 10") includes a blood pressure measurement information acquisition unit 11, a prediction estimation unit 12, and an output unit 13. In addition, although not shown, this device 10 may also include, for example, an input unit, an output unit, a display unit and / or a storage unit. The blood pressure measurement information acquisition unit 11, the prediction estimation unit 12, and the output unit 13 can each execute, for example, the blood pressure measurement information acquisition procedure, the prediction estimation procedure, and the output procedure in the atrial fibrillation prediction support program of this disclosure.

[0021] The device 10 may be, for example, a single device including the aforementioned parts, or it may be a device in which the aforementioned parts can be connected via a communication network. Furthermore, the device 10 can be connected to external devices described later via a communication network. The communication network is not particularly limited and can use any known network, such as a wired or wireless network. Examples of communication networks include the Internet, WWW (World Wide Web), telephone lines, LAN (Local Area Network), SAN (Storage Area Network), DTN (Delay Tolerant Networking), LPWA (Low Power Wide Area), L5G (Local 5G), etc. Examples of wireless communication include Wi-Fi (registered trademark), Bluetooth (registered trademark), Local 5G, LPWA, etc. The wireless communication may be in the form of direct communication between devices (Ad Hoc communication), infrastructure communication, or indirect communication via an access point. The device 10 may, for example, be incorporated into a server as part of a system. Furthermore, the device 10 may be, for example, a personal computer (PC, e.g., desktop or notebook), smartphone, tablet terminal, etc., on which the program disclosed herein is installed. In addition, the device 10 may be in a form such as cloud computing or edge computing, where, for example, at least one of the aforementioned parts is on a server and the other parts are on a terminal. The device 10 may also be, for example, incorporated into a blood pressure monitor.

[0022] Figure 2 illustrates a block diagram of the hardware configuration of the device 10. The device 10 includes, for example, a central processing unit 101, memory 102, bus 103, storage device 104, input device 105, output device 106, communication device (communication unit) 107, etc. Each part of the device 10 is interconnected via the bus 103 through its respective interface (I / F).

[0023] The central processing unit 101 operates in coordination with other components via controllers (system controller, I / O controller, etc.) and is responsible for the overall control of the 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 types of information. Specifically, for example, the central processing unit 101 functions as a blood pressure measurement information acquisition unit 11, a prediction estimation unit 12, and an output unit 13. The device 10 may also include other computing devices such as a CPU, GPU (Graphics Processing Unit), APU (Accelerated Processing Unit), or a combination thereof as computing devices.

[0024] Bus 103 can also be connected to external devices, for example. Examples of such external devices include external storage devices (external databases, etc.), printers, external input devices, external display devices, audio output devices such as speakers, external imaging devices such as cameras, and various sensors such as acceleration sensors, geomagnetic sensors, direction sensors, and blood pressure measuring devices. The device 10 can be connected to an external network (the aforementioned communication network) by a communication device 107 connected to bus 103, for example, and can also be connected to other devices via the external network.

[0025] Memory 102 may be, for example, main memory. When the central processing unit 101 performs processing, memory 102 reads various operational programs, such as the program of this disclosure, stored in the storage device 104 (described later), and the central processing unit 101 receives data from memory 102 and executes the program. The main memory may be, for example, RAM (random access memory). Alternatively, memory 102 may be, for example, ROM (read-only memory).

[0026] The storage device 104 is also called an auxiliary storage device, for example, in relation to the main memory (primary memory). As described above, the storage device 104 stores an operating program including the program of this disclosure. The storage device 104 may be, for example, a combination of a recording medium and a drive for reading and writing to the recording medium. The recording medium is not particularly limited and may be internal or external, for example, an HD (hard disk), CD-ROM, CD-R, CD-RW, MO, DVD, flash memory, memory card, etc. The storage device 104 may be, for example, a hard disk drive (HDD) in which the recording medium and the drive are integrated, or a solid state drive (SSD). If the device 10 includes, for example, a storage unit, the storage device 104 functions as the storage unit. The storage device 104 may store, for example, a predictive model described later.

[0027] In this device 10, the memory 102 and storage device 104 can also store various types of information, such as log information, information obtained from an external database (not shown) or external devices, information generated by this device 10, and information used by this device 10 when executing processing. At least some of this information may be stored, for example, on an external server other than the memory 102 and storage device 104, or it may be stored in a distributed manner across multiple terminals using blockchain technology or the like.

[0028] The device 10 further includes, for example, an input device 105 and an output device 106. The input device 105 may include, for example, a pointing device such as a touch panel, trackpad, or mouse; a keyboard; imaging means such as a camera or scanner; a card reader such as an IC card reader or magnetic card reader; an audio input means such as a microphone; and so on. The output device 106 may include, for example, a display device such as an LED display or liquid crystal display; an audio output device such as a speaker; a printer; and so on. In this disclosure 1, 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 unit, such as a touch panel display.

[0029] Next, an example of processing by the atrial fibrillation prediction support program of this disclosure will be explained based on the flowchart in Figure 3. Processing by the atrial fibrillation prediction support program of this disclosure can be carried out, for example, using the atrial fibrillation prediction support device 10 shown in Figures 1 and 2, as follows. However, processing by the atrial fibrillation prediction support program of this disclosure is not limited to the use of the atrial fibrillation prediction support device 10 shown in Figures 1 and 2.

[0030] First, the blood pressure measurement information acquisition unit 11 acquires the blood pressure measurement information of the subject (S1, blood pressure measurement information acquisition procedure). The subject may be, for example, a person who has already developed atrial fibrillation or a person who has not developed atrial fibrillation. The blood pressure measurement information is not particularly limited as long as it is information obtained by measuring blood pressure. The blood pressure measurement information may be, for example, one type of information or two or more types of information. The blood pressure measurement information may include, for example, information on at least one of blood pressure and pulse rate. The blood pressure information acquisition unit 11 may acquire the blood pressure measurement information from, for example, a blood pressure measuring device (blood pressure monitor), or it may acquire the blood pressure measurement information entered by the user (for example, the subject) into this device 10, or it may acquire the blood pressure measurement information from a recording medium that has recorded the blood pressure measurement information. The blood pressure measurement information acquisition unit 11 may acquire, for example, blood pressure measurement information over multiple days, or blood pressure measurement information over multiple days and multiple time periods. The multiple time periods are not particularly limited and may include, for example, morning, noon, evening, and night. The aforementioned time zone divisions are not particularly limited and can be any time zone. The blood pressure measurement information acquisition unit 11 may, for example, record the acquired blood pressure measurement information in the storage unit.

[0031] The blood pressure measurement information may include, for example, the date and time of measurement and subject information. The subject information is, for example, information about the subject whose blood pressure was measured. The subject information may include, for example, subject identification information and subject attribute information. The subject identification information is not particularly limited as long as it is information that can identify the subject. The subject identification information may include, for example, name, nickname, ID, password, email address, telephone number, address, identification number (for example, hospital patient card number, health insurance card number, My Number (personal information), etc.), IP address, MAC address, device-specific information of the terminal, etc. The subject attribute information is not particularly limited and may include, for example, gender, age, affiliated organization, address, contact information, hobbies, preferences, etc., biometric information (for example, height, weight, etc.).

[0032] The blood pressure measurement information acquisition unit 11 may, for example, acquire patient information linked to the blood pressure measurement information. The patient information may include, for example, a history of myocardial infarction, heart failure, and dialysis; gender; age; vital data such as blood test results and blood pressure; lifestyle habits such as smoking and drinking; information on physical characteristics such as height and weight; medical interview information; echocardiogram information, etc. The medical interview information may include, for example, a set of questions and answers from a medical interview regarding the subjective symptoms of atrial fibrillation. Specific examples of the medical interview include, but are not limited to, questions and answers related to the onset of atrial fibrillation, such as "When did you start experiencing palpitations?" and "When were you first diagnosed with atrial fibrillation?". The echocardiogram information may include, for example, information from the results of an echocardiogram (cardiac ultrasound examination).

[0033] Next, the prediction unit 12 estimates the signs of atrial fibrillation in the subject based on the blood pressure measurement information (S2, prediction estimation procedure). For example, the prediction unit 12 may calculate at least one of the day-to-day variation feature and the day-to-day variation feature of the subject's blood pressure measurement information based on the blood pressure measurement information.

[0034] The aforementioned daily variation feature is, for example, a feature representing the variation in blood pressure measurement information over multiple days. The blood pressure measurement period for a subject used in calculating the daily variation feature is not particularly limited; for example, it may be one day or multiple days of two or more days. In the latter case, the blood pressure measurement period may be, for example, 2, 3, 4, 5, 6, 7, 8, 9, 10, 14, 28 days, etc., but is not limited to these. Specific examples of the aforementioned daily variation feature include, for example, the following features, but is not limited to these. Mean: Mean morning SBP (Mean value of systolic blood pressure (SBP) measured in the morning over a reference period); Mean morning DBP (Mean value of diastolic blood pressure (DBP) measured in the morning over a reference period); Mean morning HR (mean value of heart rate (HR) measured in the morning over a reference period); Mean evening SBP (mean value of systolic blood pressure (SBP) measured in the evening over a reference period); Mean evening DBP (mean value of diastolic blood pressure (DBP) measured in the evening over a reference period); Mean evening HR (mean value of heart rate measured in the evening over a reference period); Max: Maximum morning SBP (the maximum value of systolic blood pressure (SBP) measured in the morning during a reference period); Maximum morning DBP (the maximum value of diastolic blood pressure (DBP) measured in the morning during the reference period); Max morning HR (Maximum value of heart rate (HR) measured in the morning during the reference period); Maximum evening SBP (the maximum systolic blood pressure (SBP) measured in the evening during the reference period); Maximum evening DBP (the highest value of diastolic blood pressure (DBP) measured in the evening during the reference period); Max evening HR (Maximum value of heart rate (HR) measured in the evening during the reference period); Standard deviation (SD): SD of morning SBP (Standard deviation of systolic blood pressure (SBP) measured in the morning over a reference period); SD of morning DBP (Standard deviation of diastolic blood pressure (DBP) measured in the morning over a reference period); SD of morning HR (Standard deviation of heart rate (HR) measured in the morning over a reference period); SD of evening SBP (Standard deviation of systolic blood pressure (SBP) measured in the evening over a reference period); SD of evening DBP (standard deviation of diastolic blood pressure (DBP) measured in the evening over a reference period); SD of evening HR (Standard deviation of heart rate (HR) measured in the evening over a reference period); Coefficient of variation (SD / Mean, CV): CV of morning SBP (Coefficient of variation of morning systolic blood pressure (SBP) over a reference period); CV of morning DBP (Coefficient of variation of morning diastolic blood pressure (DBP) over a reference period); CV of morning HR (Coefficient of variation of heart rate (HR) measured in the morning over a reference period); CV of evening SBP (Coefficient of variation of systolic blood pressure (SBP) measured in the evening over a reference period); CV of evening DBP (Coefficient of variation of diastolic blood pressure (DBP) measured in the evening over a reference period); CV of evening HR (Coefficient of variation of heart rate (HR) measured in the evening over a reference period); Average fluctuation range ARV (sum of absolute difference from the day before / (N-1)): DBP ARV of morning (Mean variation of systolic blood pressure (SBP) measured in the morning over a reference period) SBP ARV of morning (Mean variation of diastolic blood pressure (DBP) measured in the morning over a reference period) HR ARV of morning (Average variation of heart rate (HR) measured in the morning over a reference period) DBP ARV of evening (Mean variation of systolic blood pressure (SBP) measured in the evening over a reference period) SBP ARV of evening (Mean variation of diastolic blood pressure (DBP) measured in the evening over a reference period) HR ARV of evening (Average variation of heart rate (HR) measured in the evening over a reference period) Average fluctuation (time_arv) of data arranged chronologically for morning and evening 1-7 days before the onset of symptoms: SBP time ARV (Mean variation in systolic blood pressure (SBP) over a reference period) DBP time ARV (Mean variation in diastolic blood pressure (DBP) over a reference period) HR time ARV (Average Variability of Heart Rate (HR) over a Reference Period) The average of the morning and evening data from 1-7 days before the onset of symptoms (time_mean): DBP time Mean (Average systolic blood pressure (SBP) over a reference period) SBP time Mean (Average diastolic blood pressure (DBP) over a reference period) HR Time Mean (Average pulse rate (HR) over a reference period) The ratio of the average change in data (morning and evening, time-series) from 1-7 days before symptom onset to the average change in data (morning and evening, time-series) from 8-10 days before symptom onset (arv_ratio) SBP time ARV ratio (Average rate of change in systolic blood pressure (SBP) over a reference period) DBP time ARV ratio (mean rate of change in diastolic blood pressure (DBP) over a reference period) HR time ARV ratio (Average rate of variation of heart rate (HR) over a reference period) The mean ratio of data arranged chronologically for morning and evening 8-10 days before the onset of symptoms. SBP time-mean ratio (average ratio of systolic blood pressure (SBP) over a reference period) DBP time mean ratio (average ratio of diastolic blood pressure (DBP) over a reference period) HR time mean ratio (average ratio of heart rate (HR) measured in the evening over a reference period)

[0035] The aforementioned daily variation feature is, for example, a feature representing the variation in blood pressure measurement information for each time period. The period for measuring the blood pressure of a subject used in calculating the daily variation feature is not particularly limited; for example, it may be one day or multiple days of two or more days. In the latter case, the blood pressure measurement period may be, for example, 2, 3, 4, 5, 6, 7, 8, 9, 10, 14, 28 days, etc., but is not limited to these. Specific examples of the aforementioned daily variation feature include, for example, the following features, but is not limited to these. Mean: Mean morning-evening difference in SBP (Mean value over a reference period of the difference between systolic blood pressure (SBP) measured in the morning and systolic blood pressure (SBP) measured in the evening); Mean morning-evening difference in DBP (Mean difference between diastolic blood pressure (DBP) measured in the morning and diastolic blood pressure (DBP) measured in the evening over a reference period); Mean morning-evening difference in heart rate (HR) over a reference period; Max: Maximum morning-evening difference in SBP (the maximum difference between systolic blood pressure (SBP) measured in the morning and systolic blood pressure (SBP) measured in the evening over a reference period); Maximum morning-evening difference in DBP (the maximum difference between diastolic blood pressure (DBP) measured in the morning and diastolic blood pressure (DBP) measured in the evening over a reference period). Maximum morning-evening difference in heart rate (HR) over a reference period; Standard deviation (SD): SD morning-evening difference in SBP (standard deviation of the difference between systolic blood pressure (SBP) measured in the morning and systolic blood pressure (SBP) measured in the evening over a period of time); SD morning-evening difference in DBP (standard deviation of the difference between diastolic blood pressure (DBP) measured in the morning and diastolic blood pressure (DBP) measured in the evening over a period of time); SD morning-evening difference in HR (standard deviation of the difference between heart rate (HR) measured in the morning and heart rate (HR) measured in the evening over a period of time); Coefficient of variation (SD / Mean, CV): CV (Coefficient of Variation) of the difference in morning-evening systolic blood pressure (SBP) over time; CV (Coefficient of Variation) of the difference in morning-evening diastolic blood pressure (DBP) between morning and evening measurements over time. CV (Coefficient of Variation) of the difference in heart rate (HR) between morning and evening measurements over a period of time. Average fluctuation range ARV (sum of absolute difference from the day before / (N-1)): ARV (Average Variability of Morning-Evening Difference in SBP) ARV (Average Variability of Morning-Evening Difference in DBP) ARV (Average Variability in Morning-Evening Difference in Heart Rate)

[0036] The prediction unit 12 can, for example, input the blood pressure measurement information into a prediction model to estimate the signs of atrial fibrillation in the subject. The prediction model is a model that has been trained by machine learning, using the presence or absence of atrial fibrillation as the ground truth label for the blood pressure measurement information, to output an estimated result of the signs of atrial fibrillation in the subject when the subject's blood pressure measurement information is input.

[0037] Furthermore, the prediction unit 12 may, for example, input at least one of the daily variation feature and the day-to-day variation feature into the prediction model to estimate the signs of atrial fibrillation in the subject. In this case, the prediction model is a model that has been trained by machine learning, using the presence or absence of atrial fibrillation as the ground truth label for at least one of the daily variation feature and the day-to-day variation feature, to output an estimation result of the signs of atrial fibrillation in the subject when at least one of the daily variation feature and the day-to-day variation feature of the subject is input.

[0038] The aforementioned prediction model may be, for example, a model pre-stored in memory 102 and storage device 104, or a model acquired from outside the device 10 via a communication network. In the latter case, it can be acquired, for example, from an external pre-trained model manufacturing device for atrial fibrillation prediction (for example, the pre-trained model manufacturing device 20 for atrial fibrillation prediction described later). The prediction unit 12 can, for example, estimate the signs of an atrial fibrillation attack occurring in the subject as the aforementioned signs.

[0039] The aforementioned predictive model can be generated, for example, by machine learning using a pair of data (training information) that combines the subject's blood pressure measurement information with whether or not the subject has experienced atrial fibrillation. The training information may, for example, be a combination of at least one of the subject's diurnal variation features and diurnal differential variation features, and whether or not the subject has experienced atrial fibrillation. The machine learning can employ, for example, a known machine learning method. Specific examples of statistical models that can be used in the machine learning include, for example, simple linear regression models, Ridge regression, Lasso regression, Elastic Net regression, LightGBM (Light Gradient Boosting Machine), Logistic regression, general additive models, random forest regression, rule-fit regression, gradient boosting trees, extra trees, support vector regression, Gaussian process regression, k-nearest neighbor regression, kernel ridge regression, neural networks, etc. The predictive model may be, for example, a pre-trained model. The pre-trained model may also be a pre-trained model (derived model) that has been retrained using the training data and an already generated pre-trained model. Furthermore, the pre-trained model may be a pre-trained model obtained by transfer learning using a pre-trained model generated using the training data, or a pre-trained model generated by model compression of a pre-trained model generated using the training data.

[0040] Furthermore, the blood pressure measurement information included in the learning information may, for example, be blood pressure measurement information from a predetermined reference period before the subject develops atrial fibrillation. The reference period is not particularly limited and may be 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 14, 28 days prior to the day the subject develops atrial fibrillation (day 0). Furthermore, if the subject is not experiencing atrial fibrillation, the reference period may be, for example, a period of 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 14, or 28 days prior to a predetermined date (day 0). The aforementioned reference period is not particularly limited and may be determined, for example, according to the embodiments described later. Furthermore, there are no particular limitations on how to determine whether or not the subject has developed atrial fibrillation; for example, it can be determined according to the examples described later.

[0041] The aforementioned anomaly estimation model may be a network including, for example, an input layer that takes at least one selected from a group consisting of blood pressure measurement information, diurnal variation features, and diurnal differential variation features as input, an output layer that outputs the estimation result of the anomaly of atrial fibrillation, and at least one intermediate layer provided between the input layer and the output layer. In this case, the anomaly estimation model may be a program module that is part of artificial intelligence software. Examples of the multilayer network include neural networks. Examples of the neural network include convolutional neural networks (CNNs), but are not limited to CNNs, and may also be neural networks other than CNNs, SVMs (Support Vector Machines), Bayesian networks, regression trees, or other learning algorithms, or pre-trained models constructed with such learning algorithms.

[0042] If the patient information is acquired in S1, the prediction unit 12 may, for example, estimate the signs of atrial fibrillation in the subject based on the blood pressure measurement information and the patient information. In this case, the prediction unit 12 can, for example, input the blood pressure measurement information and the patient information into a prediction model to estimate the signs of atrial fibrillation in the subject. The prediction model in this case is the same as described above, except that it is a model that has been trained by machine learning, for example, using the presence or absence of atrial fibrillation in the subject as the ground truth label for a combination of the blood pressure measurement information and the patient information, to output an estimated result of the signs of atrial fibrillation in the subject when the blood pressure measurement information and patient information of the subject are input.

[0043] The output unit 13 then outputs the estimated result of the atrial fibrillation precursor (S3, output step). The output may be, for example, output to the output device 106 of the device 10 (for example, a display), or output to an external device outside the device 10. The output unit 13 may, for example, output the estimated result of the precursor linked to the identification information of the atrial fibrillation patient. The output unit 13 may, for example, output the precursor data of other atrial fibrillation patients who have developed atrial fibrillation (at least one selected from the group consisting of blood pressure measurement data, diurnal variation features, and day-to-day variation features) and the atrial fibrillation precursor data of the subject (at least one selected from the group consisting of blood pressure measurement data, diurnal variation features, and day-to-day variation features) together with the estimated result of the atrial fibrillation precursor. A subject for whom atrial fibrillation precursors have been estimated has not yet developed atrial fibrillation at that time. Therefore, by outputting not only the atrial fibrillation precursor data of the target individual but also the atrial fibrillation precursor data of patients who have already developed atrial fibrillation, it is possible to raise awareness among the target individual regarding the onset of atrial fibrillation and encourage preventive actions such as improving their lifestyle and seeking medical attention.

[0044] The output unit 13 may, for example, change the output format of the estimation results depending on the output destination. Specifically, for example, if the output destination is a medical institution, the output unit 13 may output, in addition to the estimation results of the atrial fibrillation precursors, at least one selected from the group consisting of the subject's blood pressure measurement data, day-to-day variation features, and day-to-day variation features. Also, if the output destination is the subject themselves or a family member of the subject, in addition to the estimation results of the precursors, information regarding the prevention of atrial fibrillation (for example, diet, exercise, sleep, and other lifestyle habits of the subject) may be output. The information regarding the prevention of atrial fibrillation may, for example, be recorded in the storage unit of the device 10. The information regarding the prevention of atrial fibrillation may, for example, be a URL of the information provider. If the output destination is a care facility for the subject, the output unit 13 may, for example, output information to control the terminal to sound an alert, in addition to the estimation results of the precursors. This makes it easier for, for example, the subject's caregiver to take appropriate action, such as dispatching a nurse to the subject. Note that the combinations of output information types and output destinations provided by the output unit 13 are examples only and are not limited to the above combinations; any information may be output to any output destination.

[0045] The atrial fibrillation prediction support method of this disclosure is a method implemented by, for example, replacing each "procedure" in the atrial fibrillation prediction support program of this disclosure with "process". The atrial fibrillation prediction support method of this disclosure can be implemented, for example, using the atrial fibrillation prediction support device 10 of this disclosure shown in Figure 1 or Figure 2. However, the atrial fibrillation prediction support method of this disclosure is not limited to, for example, a method using the atrial fibrillation prediction support device 10. The atrial fibrillation prediction support method of this disclosure can, for example, utilize the descriptions in the atrial fibrillation prediction support program and the atrial fibrillation prediction support device of this disclosure.

[0046] According to this disclosure, it is possible to estimate the onset of atrial fibrillation in a subject based on their blood pressure measurement information. Therefore, according to this disclosure, it is possible to estimate the onset of atrial fibrillation using daily blood pressure measurements with a blood pressure monitor without requiring an electrocardiogram. As mentioned above, symptoms of atrial fibrillation are not always present, so this disclosure is useful for the early detection of atrial fibrillation. In addition, a characteristic of the early stages of atrial fibrillation is that attacks occur irregularly. Therefore, patients who have developed atrial fibrillation are constantly anxious about "when the next attack will occur." According to this disclosure, it is possible to notify the subject of the onset of atrial fibrillation based on their blood pressure measurement information, so that the subject can take preventive measures before an atrial fibrillation attack occurs. Therefore, according to this disclosure, it is also possible to reduce anxiety about the risk of developing atrial fibrillation in patients who have developed atrial fibrillation, and in people at high risk of developing atrial fibrillation (for example, patients with hypertension).

[0047] [Embodiment 2] Embodiment 2 is an example of a pre-trained model manufacturing apparatus for predicting atrial fibrillation according to the present disclosure.

[0048] The trained model manufacturing apparatus for predicting atrial fibrillation according to this disclosure will be described with reference to Figure 4. Figure 4 is a block diagram showing an example configuration of the trained model manufacturing apparatus 20 for predicting atrial fibrillation according to this disclosure. As shown in Figure 4, the trained model manufacturing apparatus 20 for predicting atrial fibrillation (hereinafter also referred to as "the apparatus 20") includes a training information acquisition unit 21 and a trained model generation unit 22. The apparatus 20 may also include, for example, an input unit, an output unit, a display unit and / or a storage unit, although these are not shown.

[0049] The pre-trained model manufacturing device 20 for predicting atrial fibrillation may be, for example, a single device including the aforementioned parts, or a device in which each of the aforementioned parts can be connected via a communication network. Furthermore, the pre-trained model manufacturing device 20 for predicting atrial fibrillation may be connected to an external device described later via a communication network. The communication network is not particularly limited and a known network can be used, for example, it may be wired or wireless. Examples of communication networks include the Internet, WWW (World Wide Web), telephone lines, LAN (Local Area Network), SAN (Storage Area Network), DTN (Delay Tolerant Networking), LPWA (Low Power Wide Area), L5G (Local 5G), etc. Examples of wireless communication include Wi-Fi (registered trademark), Bluetooth (registered trademark), Local 5G, LPWA, etc. The wireless communication may be in the form of direct communication between each device (Ad Hoc communication), infrastructure communication, indirect communication via an access point, etc. The pre-trained model manufacturing device 20 for predicting atrial fibrillation may, for example, be incorporated into a server as a system. Alternatively, the pre-trained model manufacturing device 20 for predicting atrial fibrillation may be, for example, a personal computer (PC, e.g., desktop or notebook), smartphone, tablet terminal, etc., on which the program of this disclosure is installed. Furthermore, the pre-trained model manufacturing device 20 for predicting atrial fibrillation may be in the form of cloud computing or edge computing, for example, in which at least one of the above components is on a server and the other components are on a terminal.

[0050] Figure 5 illustrates a block diagram of the hardware configuration of the Atrial Fibrillation Precursor Estimation Pre-trained Model Manufacturing Device 20. As shown in Figure 5, the Atrial Fibrillation Precursor Estimation Pre-trained Model Manufacturing Device 20 includes, for example, a central processing unit 201, memory 202, bus 203, storage device 204, input device 205, output device 206, communication device 207, etc. The description of each component of the Atrial Fibrillation Precursor Estimation Pre-trained Model Manufacturing Device 20 can be based on the description of each component of the Atrial Fibrillation Precursor Detection Support Device 10. Each part of the Atrial Fibrillation Precursor Estimation Pre-trained Model Manufacturing Device 20 is connected via the bus 203 by its respective interface (I / F). In the Atrial Fibrillation Precursor Estimation Pre-trained Model Manufacturing Device 20, the central processing unit 201 functions as a learning information acquisition unit 21 and a pre-trained model generation unit 22.

[0051] Next, an example of a method for manufacturing the trained model of this disclosure will be described based on the flowchart in Figure 6. The method for manufacturing the trained model of this disclosure is carried out as follows, for example, using the trained model manufacturing apparatus 20 for predicting atrial fibrillation shown in Figures 4 and 5. However, the method for manufacturing the trained model of this disclosure is not limited to the use of the trained model manufacturing apparatus 20 for predicting atrial fibrillation shown in Figures 4 and 5.

[0052] First, the learning information acquisition unit 21 acquires learning information (S21, learning information acquisition step). The learning information includes the subject's blood pressure measurement information and whether or not the subject has developed atrial fibrillation. The learning information may also be, for example, a set of at least one of the subject's diurnal variation feature and diurnal differential variation feature and whether or not the subject has developed atrial fibrillation. Furthermore, the learning information may also include, for example, at least one selected from the group consisting of the attribute information of the atrial fibrillation patient, medical interview information, and echocardiogram information. The information included in the learning information is the same as in Embodiment 1, and its explanation can be used with reference. The learning information acquisition unit 21 may, for example, acquire information that has been previously stored in the memory 202 and the storage device 204, or it may acquire information that has been stored in an external database via a communication network.

[0053] Next, the trained model generation unit 21 generates a trained model for predictive atrial fibrillation estimation, which outputs an estimated result of predictive atrial fibrillation in a subject when the subject's blood pressure measurement information is input, using machine learning with the presence or absence of atrial fibrillation as the correct label for the blood pressure measurement information (S21, training process). The trained model generation unit 21 can generate the predictive atrial fibrillation estimation model by machine learning using, for example, the subject's blood pressure measurement information and the presence or absence of atrial fibrillation in the subject as training data.

[0054] The trained model generation unit 21 may generate a trained model for predictive atrial fibrillation estimation, which outputs an estimated result of predictive atrial fibrillation in a subject when at least one of the subject's daily variation features and daily variation features is input. This model is generated by machine learning using the presence or absence of atrial fibrillation as the ground truth label for at least one of the daily variation features and daily variation features. For example, the trained model generation unit 21 can generate the predictive atrial fibrillation estimation model by machine learning using a pair of at least one of the subject's daily variation features and daily variation features, and the presence or absence of atrial fibrillation in the subject, as training data.

[0055] The aforementioned machine learning can employ, for example, a known machine learning method. Specific examples of statistical models that can be used in the machine learning include, for example, simple linear regression models, Ridge regression, Lasso regression, Elastic Net regression, LightGBM (Light Gradient Boosting Machine), Logistic regression, general additive models, random forest regression, rule-fit regression, gradient boosting trees, extra trees, support vector regression, Gaussian process regression, k-nearest neighbor regression, kernel ridge regression, neural networks, etc. The predictive estimation model may, for example, be a pre-generated, trained model. Furthermore, the trained model may be a trained model (derived model) retrained using the training data and an already generated trained model. Additionally, the trained model may be a trained model obtained by transfer learning using a trained model generated with the training data, or a trained model generated by model compression of a trained model generated with the training data.

[0056] The trained model generation unit 21 may generate a network that includes, for example, an input layer that takes at least one selected from the group consisting of blood pressure measurement information, diurnal variation features, and diurnal differential variation features as input, an output layer that outputs the estimation result of the atrial fibrillation precursor, and at least one intermediate layer provided between the input layer and the output layer. In this case, the precursor estimation model may be a program module that is part of artificial intelligence software. Examples of the multilayer network include neural networks. Examples of the neural network include convolutional neural networks (CNNs), but are not limited to CNNs, and trained models constructed with other neural networks, SVMs (Support Vector Machines), Bayesian networks, regression trees, and other learning algorithms may also be used.

[0057] If the training information includes the patient information, the trained model generation unit 21 may generate a prediction estimation model that has been trained to output an estimated result of the prediction of atrial fibrillation in the subject when the subject's blood pressure measurement information, daily variation features, and daily variation features, along with patient information, is input, by machine learning using, for example, the presence or absence of atrial fibrillation in the subject as the ground truth label for a combination of the blood pressure measurement information, daily variation features, and daily variation features, along with patient information. In this case, the prediction estimation model is the same as described above, except that it is a model that has been trained to output an estimated result of the prediction of atrial fibrillation in the subject when the subject's blood pressure measurement information and patient information are input, by machine learning using, for example, the presence or absence of atrial fibrillation in the subject as the ground truth label for a combination of the electrocardiogram information and patient information.

[0058] The trained model generated by this disclosure can be used, for example, in the atrial fibrillation prediction support device described in Embodiment 1. This makes it possible to estimate the signs of atrial fibrillation based on the blood pressure measurement information of the subject.

[0059] [Embodiment 3] The pre-trained model manufacturing program for predicting atrial fibrillation according to this disclosure is a program that causes a computer to execute each step of the pre-trained model manufacturing method for predicting atrial fibrillation described above. Specifically, the pre-trained model manufacturing program for predicting atrial fibrillation according to this disclosure is a program that causes a computer to execute a learning information acquisition unit and a pre-trained model generation unit.

[0060] The above procedure for acquiring learning information involves acquiring learning information, The aforementioned learning information includes the subject's blood pressure measurement information and whether or not the subject has developed atrial fibrillation. The aforementioned pre-trained model generation procedure generates a pre-trained model that, when the blood pressure measurement information of a subject is input, outputs an estimated result of the subject's atrial fibrillation precursors, using machine learning with the presence or absence of atrial fibrillation as the ground truth label for the blood pressure measurement information.

[0061] Furthermore, the trained model manufacturing program for predictive maintenance described herein can also be described as a program that causes a computer to function as a procedure for acquiring training information and a procedure for generating a trained model.

[0062] The pre-trained model manufacturing program for predicting atrial fibrillation according to this disclosure can be based on the descriptions in the pre-trained model manufacturing apparatus and pre-trained model manufacturing method for predicting atrial fibrillation according to this disclosure. In each of the aforementioned steps, for example, “step” can be read as “process.” The program of this disclosure may also be recorded on a computer-readable recording medium, for example. The recording medium is not particularly limited and includes, for example, random access memory (RAM), read-only memory (ROM), hard disk (HD), flash memory (e.g., SSD (Solid State Drive), USB flash memory, SD / SDHC card, etc.), optical disc (e.g., CD-R / CD-RW, DVD-R / DVD-RW, BD-R / BD-RE, etc.), magneto-optical disk (MO), floppy disk (FD), etc. The pre-trained model manufacturing program for predicting atrial fibrillation according to this disclosure (for example, also referred to as the programming product or the pre-trained model manufacturing program product for predicting atrial fibrillation) may also be delivered, for example, from an external computer. The aforementioned "distribution" may be, for example, distribution via a communication network or distribution via a wired device. The trained model manufacturing program for predictive maintenance of the present disclosure may be installed and executed on the distributed device, or it may be executed without being installed. [Examples]

[0063] This disclosure confirms that the predictive model for atrial fibrillation can predict the onset of symptoms.

[0064] (1) Participants The subjects of the experiment were 4,078 individuals aged 60 or younger who were recruited from across Japan between April 25, 2022, and June 30, 2023, and who met the following selection criteria and did not meet the following exclusion criteria. Selection Criteria (i) Patients with a history of hypertension who are taking antihypertensive medication. (ii) Age 60 or older (iii) Patients who have given their consent to participate in the experiment ·Exclusion criteria (i) Patients currently participating in or scheduled to participate in an intervention trial (ii) Patients who have already been diagnosed with atrial fibrillation (self-reported) (iii) Patients currently using anticoagulants (self-reported) (iv) Patients with an implanted pacemaker or defibrillator (v) Resident outside Japan (vi) Patients who are deemed unsuitable to participate in this experiment by their attending physician

[0065] (2-1) Experimental Method An upper arm blood pressure monitor with an electrocardiogram (HCR-7800T, Omron Healthcare Co., Ltd.) was sent to each participant's home. For three months, each participant used the monitor in the morning and evening to self-measure their electrocardiogram data, blood pressure data, and pulse rate data. Based on the collected electrocardiogram data, the relationship between the presence or absence of atrial fibrillation in each participant and the blood pressure measurement data (feature data extracted from blood pressure and pulse rate data) was analyzed.

[0066] (2-2) Per Protocol Set (PPS) All registered cases (4,078) from each experimental participant were treated as the Full Analysis Set (FAS). The FAS included 258 cases (6.3%) in which blood pressure was never measured, and 363 cases (8.9%) in which participation in the study was discontinued before the 3-month follow-up period. Participants in the FAS who did not discontinue participation were classified as the Per Protocol Set (PPS) if they had at least 0.6 measurements (number of days from the first measurement to day 90 or the last measurement, whichever was shorter). Cases that discontinued participation were classified as the Per Protocol Set (PPS) if they had at least 0.6 measurements (number of days from the first measurement to the date of discontinuation). As a result, 3,309 cases were classified as the Per Protocol Set (PPS).

[0067] (3-1) Definition of the date of onset of atrial fibrillation For each experimental subject, if at least two of the three event judges determined that the electrocardiogram data measured by the aforementioned home-use electrocardiograph in which Possible AF was detected represented the onset of atrial fibrillation, then the subject was defined as having developed atrial fibrillation. Furthermore, for each experimental subject, the earliest date on which the onset of atrial fibrillation was determined was defined as the date of new onset of atrial fibrillation for that subject. New onset of atrial fibrillation was observed in 220 cases (5.4% of FAS) among the experimental subjects. The time to the onset of new onset of atrial fibrillation among the experimental subjects ranged from 3 to 109 days (median: 28 days).

[0068] (3-2) Definition of the reference period for detecting early signs of atrial fibrillation The period from one day to seven days prior to the onset of atrial fibrillation was used as the reference period for detecting early signs of atrial fibrillation. For subjects who had not yet experienced atrial fibrillation, the reference date was set to 30 days after the start of the experiment, and the reference period was set from one day to seven days prior to the aforementioned reference date. Blood pressure measurement data (blood pressure data and pulse rate data) for each subject during the reference period was then extracted.

[0069] (3-3) Data to be analyzed In the PPS (Pregnancy-Promoting System) study, cases with missing morning and evening blood pressure measurement data for three or more days during the aforementioned 7-day reference period were excluded from the analysis. The analysis included 3099 cases, of which 193 developed atrial fibrillation and 2906 did not.

[0070] (4) Feature extraction The following features were extracted from the data to be analyzed. Mean: Mean morning SBP (Mean value of systolic blood pressure (SBP) measured in the morning over a reference period); Mean morning DBP (Mean value of diastolic blood pressure (DBP) measured in the morning over a reference period); Mean morning HR (mean value of heart rate (HR) measured in the morning over a reference period); Mean evening SBP (mean value of systolic blood pressure (SBP) measured in the evening over a reference period); Mean evening DBP (mean value of diastolic blood pressure (DBP) measured in the evening over a reference period); Mean evening HR (mean value of heart rate measured in the evening over a reference period); Mean morning-evening difference in SBP (Mean value over a reference period of the difference between systolic blood pressure (SBP) measured in the morning and systolic blood pressure (SBP) measured in the evening); Mean morning-evening difference in DBP (Mean difference between diastolic blood pressure (DBP) measured in the morning and diastolic blood pressure (DBP) measured in the evening over a reference period); Mean morning-evening difference in heart rate (HR) over a reference period; Max: Maximum morning SBP (the maximum value of systolic blood pressure (SBP) measured in the morning during a reference period); Maximum morning DBP (the maximum value of diastolic blood pressure (DBP) measured in the morning during the reference period); Max morning HR (Maximum value of heart rate (HR) measured in the morning during the reference period); Maximum evening SBP (the maximum systolic blood pressure (SBP) measured in the evening during the reference period); Maximum evening DBP (the highest value of diastolic blood pressure (DBP) measured in the evening during the reference period); Max evening HR (Maximum value of heart rate (HR) measured in the evening during the reference period); Maximum morning-evening difference in SBP (the maximum difference between systolic blood pressure (SBP) measured in the morning and systolic blood pressure (SBP) measured in the evening over a reference period); Maximum morning-evening difference in DBP (the maximum difference between diastolic blood pressure (DBP) measured in the morning and diastolic blood pressure (DBP) measured in the evening over a reference period). Maximum morning-evening difference in heart rate (HR) over a reference period; Standard deviation (SD): SD of morning SBP (Standard deviation of systolic blood pressure (SBP) measured in the morning over a reference period); SD of morning DBP (Standard deviation of morning diastolic blood pressure (DBP) over a reference period); SD of morning HR (Standard deviation of heart rate (HR) measured in the morning over a reference period); SD of evening SBP (Standard deviation of systolic blood pressure (SBP) measured in the evening over a reference period); SD of evening DBP (standard deviation of diastolic blood pressure (DBP) measured in the evening over a reference period); SD of evening HR (Standard deviation of heart rate (HR) measured in the evening over a reference period); SD morning-evening difference in SBP (standard deviation of the difference between systolic blood pressure (SBP) measured in the morning and systolic blood pressure (SBP) measured in the evening over a reference period); SD morning-evening difference in DBP (standard deviation of the difference between diastolic blood pressure (DBP) measured in the morning and diastolic blood pressure (DBP) measured in the evening over a reference period); SD morning-evening difference in HR (standard deviation of the difference between heart rate (HR) measured in the morning and heart rate (HR) measured in the evening over a reference period); Coefficient of variation (SD / Mean, CV): CV of morning SBP (Coefficient of variation of morning systolic blood pressure (SBP) over a reference period); CV of morning DBP (Coefficient of variation of morning diastolic blood pressure (DBP) over a reference period); CV of morning HR (Coefficient of variation of heart rate (HR) measured in the morning over a reference period); CV of evening SBP (Coefficient of variation of systolic blood pressure (SBP) measured in the evening over a reference period); CV of evening DBP (Coefficient of variation of diastolic blood pressure (DBP) measured in the evening over a reference period); CV of evening HR (Coefficient of variation of heart rate (HR) measured in the evening over a reference period); CV (Coefficient of Variation) of the difference between morning-evening systolic blood pressure (SBP) and evening systolic blood pressure (SBP) over a reference period. CV (Coefficient of Variation) of the difference in morning-evening diastolic blood pressure (DBP) over a reference period; CV (Coefficient of Variation) of the difference in heart rate (HR) between morning and evening measurements over a reference period. Average fluctuation range ARV (sum of absolute difference from the day before / (N-1)): DBP ARV of morning (Mean variation of systolic blood pressure (SBP) measured in the morning over a reference period) SBP ARV of morning (Mean variation of diastolic blood pressure (DBP) measured in the morning over a reference period) HR ARV of morning (Average variation of heart rate (HR) measured in the morning over a reference period) DBP ARV of evening (Mean variation of systolic blood pressure (SBP) measured in the evening over a reference period) SBP ARV of evening (Mean variation of diastolic blood pressure (DBP) measured in the evening over a reference period) HR ARV of evening (Average variation of heart rate (HR) measured in the evening over a reference period) ARV (Average Variability of Morning-Evening Difference in SBP) ARV (Average Variability of Morning-Evening Difference in DBP) ARV (Average Variability in Morning-Evening Difference in Heart Rate) Mean: 8-10 days before symptom onset: Mean (8-10) morning SBP (mean value of systolic blood pressure (SBP) measured in the morning over a reference period); Mean (8-10) morning DBP (mean value of diastolic blood pressure (DBP) measured in the morning over the reference period); Mean(8-10) morning HR (mean value of heart rate (HR) measured in the morning over a reference period); Mean (8-10) evening SBP (mean value of systolic blood pressure (SBP) measured in the evening over the reference period); Mean (8-10) evening DBP (mean value of diastolic blood pressure (DBP) measured in the evening over the reference period); Mean (8-10) evening HR (mean value of heart rate (HR) measured in the evening over a reference period); Average fluctuation (time_arv) of data arranged chronologically for morning and evening 1-7 days before the onset of symptoms: SBP time ARV (Mean variation in systolic blood pressure (SBP) over a reference period) DBP time ARV (Mean variation in diastolic blood pressure (DBP) over a reference period) HR time ARV (Average Variability of Heart Rate (HR) over a Reference Period) The average of the morning and evening data from 1-7 days before the onset of symptoms (time_mean): DBP time Mean (Average systolic blood pressure (SBP) over a reference period) SBP time Mean (Average diastolic blood pressure (DBP) over a reference period) HR Time Mean (Average pulse rate (HR) over a reference period) The ratio of the average change in data (morning and evening, time-series) from 1-7 days before symptom onset to the average change in data (morning and evening, time-series) from 8-10 days before symptom onset (arv_ratio) SBP time ARV ratio (Average rate of change in systolic blood pressure (SBP) over a reference period) DBP time ARV ratio (mean rate of change in diastolic blood pressure (DBP) over a reference period) HR time ARV ratio (Average rate of variation of heart rate (HR) over a reference period) The mean ratio of data arranged chronologically for morning and evening 8-10 days before the onset of symptoms. SBP time-mean ratio (average ratio of systolic blood pressure (SBP) over a reference period) DBP time mean ratio (average ratio of diastolic blood pressure (DBP) over a reference period) HR time mean ratio (average ratio of heart rate (HR) measured in the evening over a reference period)

[0071] (5) Development of a predictive model Of the features extracted from the aforementioned data to be analyzed, 80% were randomly divided and used as training data, and the remaining 20% ​​as test data. Then, using the training data as explanatory variables, machine learning (XGBoost (eXtreme Gradient Boosting)) was performed on the data of experimental subjects who developed atrial fibrillation, setting the dependent variable to 1, and the dependent variable to 0, for each training dataset, to create a machine learning model (predictive prediction model). In the machine learning process, the training data was divided into five sets (fold0, fold1, fold2, fold3, fold4) and cross-validation was performed. Figure 7 shows the results of validating the created predictive prediction model on the test data.

[0072] As shown in Figure 7, in the prediction model, the AUC (Area Under the Curve) of fold0 was 0.78, the AUC of fold1 was 0.83, the AUC of fold2 was 0.78, the AUC of fold3 was 0.79, and the AUC of fold4 was 0.80. The average AUC for each fold was 0.80, and the result of validation with test data was 0.79. Therefore, it was found that the prediction model of this disclosure can estimate the signs of atrial fibrillation with an accuracy of approximately 80% based on the blood pressure measurement information.

[0073] Furthermore, we used SHAP (SHapley Additive exPlanations) values ​​to search for features that contribute to the onset of atrial fibrillation from each feature. The results are shown in Figure 8. Figure 8 is a graph showing SHAP (SHapley Additive exPlanations) values ​​on the horizontal axis and the type of feature on the vertical axis. As shown in Figure 8, the following features are suggested to contribute to the predictive model of this disclosure. However, this disclosure is not limited to the following suggestions. HR_time_arv (Average variation (time_arv) of pulse rate (HR) during the reference period, based on the average variation of morning and evening data arranged in time series 1-7 days before symptom onset) Age (age of the experiment participants) HR_me_arv(ARV morning-evening difference in HR (Average variation in the difference between heart rate (HR) measured in the morning and heart rate (HR) measured in the evening over a reference period)) HR_time_arv_ratio(HR time ARV ratio (average fluctuation ratio of heart rate (HR) over the reference period) is calculated as the ratio of the average fluctuation of time-series data of morning and evening data from 1-7 days before symptom onset to the average fluctuation of time-series data of morning and evening data from 8-10 days before symptom onset (arv_ratio)) DBP_time_mean_ratio(The DBP time Mean ratio (the average ratio of diastolic blood pressure (DBP) over the reference period) is calculated based on the mean ratio (mean_ratio) of data arranged in time series from morning to evening 8-10 days before the onset of symptoms.) Height (height of the experiment subject) DBP_m_8-10 (Mean morning DBP (Mean value of diastolic blood pressure (DBP) measured in the morning during the reference period) 8-10 days prior to symptom onset) DBP_me_arv(ARV morning-evening difference in DBP (mean variation over the reference period of the difference between diastolic blood pressure (DBP) measured in the morning and diastolic blood pressure (DBP) measured in the evening)) SBP_e_max(Max evening SBP (Maximum value of systolic blood pressure (SBP) measured in the evening during the reference period)) SBP_time_arv_ratio (The SBP time ARV ratio (average fluctuation ratio of systolic blood pressure (SBP) over the reference period) is calculated as the ratio (arv_ratio) of the average fluctuation of morning and evening data arranged in time series 1-7 days before symptom onset to the average fluctuation of morning and evening data arranged in time series 8-10 days before symptom onset.)

[0074] Although the present disclosure has been described above with reference to embodiments, the present disclosure is not limited to the embodiments described above. Various modifications to the structure and details of the present disclosure are possible, which can be understood by those skilled in the art within the scope of the present disclosure.

[0075] <Note> Some or all of the above embodiments may be described as follows, but are not limited to the following: (Note 1) This includes a procedure for acquiring blood pressure measurement information, a procedure for predicting a potential hazard, and an output procedure. The aforementioned blood pressure measurement information acquisition procedure acquires the blood pressure measurement information of the subject, The aforementioned precursor estimation procedure estimates the precursor of atrial fibrillation in the subject based on the blood pressure measurement information, The output procedure outputs the estimated result of the atrial fibrillation precursor. A program that assists in detecting atrial fibrillation by having a computer execute each step. (Note 2) The aforementioned blood pressure measurement information acquisition procedure acquires blood pressure measurement information over multiple days. The aforementioned predictive indicator estimation procedure is: Based on the blood pressure measurement information, the daily variation characteristic of the subject's blood pressure measurement information is calculated. An atrial fibrillation prediction support program as described in Appendix 1, which estimates the presence of atrial fibrillation in the subject based on the aforementioned diurnal variation characteristics. (Note 3) The blood pressure measurement information acquisition procedure acquires blood pressure measurement information over multiple time periods. The aforementioned predictive indicator estimation procedure is: Based on the blood pressure measurement information, the daily variation characteristics of the subject's blood pressure measurement information are calculated. An atrial fibrillation prediction support program as described in Appendix 1 or 2, which estimates the presence of atrial fibrillation in the subject based on the aforementioned diurnal variation features. (Note 4) The aforementioned precursor estimation procedure is an atrial fibrillation precursor detection support program as described in any of Appendix 1 to 3, which inputs the blood pressure measurement information into a precursor estimation model to estimate the precursor of atrial fibrillation in the subject. (Note 5) It includes a blood pressure measurement information acquisition unit, a prediction unit, and an output unit. The blood pressure measurement information acquisition unit acquires the blood pressure measurement information of the subject, The aforementioned precursor estimation unit estimates the precursor of atrial fibrillation in the subject based on the blood pressure measurement information, The output unit is an atrial fibrillation precursor detection support device that outputs the estimated result of the atrial fibrillation precursor. (Note 6) The blood pressure measurement information acquisition unit acquires blood pressure measurement information over multiple days. The aforementioned predictive estimation unit, Based on the blood pressure measurement information, the daily variation characteristic of the subject's blood pressure measurement information is calculated. An atrial fibrillation prediction support device as described in Appendix 5, which estimates the signs of atrial fibrillation in the subject based on the aforementioned diurnal variation characteristics. (Note 7) The blood pressure measurement information acquisition unit acquires blood pressure measurement information for multiple time periods. The aforementioned predictive estimation unit, Based on the blood pressure measurement information, the daily variation characteristics of the subject's blood pressure measurement information are calculated. An atrial fibrillation prediction support device according to Appendix 5 or 6, which estimates the signs of atrial fibrillation in the subject based on the aforementioned diurnal variation characteristics. (Note 8) The atrial fibrillation precursor detection support device according to any one of appendices 5 to 7, wherein the precursor estimation unit inputs the blood pressure measurement information into the precursor estimation model to estimate the precursor of atrial fibrillation in the subject. (Note 9) This includes a blood pressure measurement information acquisition process, a prediction estimation process, and an output process. The blood pressure measurement information acquisition step involves acquiring the blood pressure measurement information of the subject, The aforementioned precursor estimation step estimates the precursor of atrial fibrillation in the subject based on the blood pressure measurement information, The output step outputs the estimated result of the atrial fibrillation precursor. A method for supporting the detection of atrial fibrillation, in which each step is performed by a computer. (Note 10) The blood pressure measurement information acquisition process acquires blood pressure measurement information over multiple days. The aforementioned predictive estimation process is: Based on the blood pressure measurement information, the daily variation characteristic of the subject's blood pressure measurement information is calculated. An atrial fibrillation prediction support program as described in Appendix 9, which estimates the signs of atrial fibrillation in the subject based on the aforementioned diurnal variation characteristics. (Note 11) The blood pressure measurement information acquisition step acquires blood pressure measurement information for multiple time periods. The aforementioned predictive estimation process is: Based on the blood pressure measurement information, the daily variation characteristics of the subject's blood pressure measurement information are calculated. An atrial fibrillation prediction support program as described in Appendix 9 or 10, which estimates the presence of atrial fibrillation in the subject based on the aforementioned diurnal variation characteristics. (Note 12) The atrial fibrillation precursor detection support program according to any one of appendices 9 to 11, wherein the precursor estimation step involves inputting the blood pressure measurement information into a precursor estimation model to estimate the precursor of atrial fibrillation in the subject. (Note 13) This includes a procedure for acquiring blood pressure measurement information, a procedure for predicting a potential hazard, and an output procedure. The aforementioned blood pressure measurement information acquisition procedure acquires the blood pressure measurement information of the subject, The aforementioned precursor estimation procedure estimates the precursor of atrial fibrillation in the subject based on the blood pressure measurement information, The output procedure outputs the estimated result of the atrial fibrillation precursor. A computer-readable recording medium containing a program that assists in the detection of atrial fibrillation, which instructs a computer to perform each step of the procedure. (Note 14) The aforementioned blood pressure measurement information acquisition procedure acquires blood pressure measurement information over multiple days. The aforementioned predictive indicator estimation procedure is: Based on the blood pressure measurement information, the daily variation characteristic of the subject's blood pressure measurement information is calculated. A recording medium as described in Appendix 13, which estimates the signs of atrial fibrillation in the subject based on the aforementioned daily variation characteristics. (Note 15) The blood pressure measurement information acquisition procedure acquires blood pressure measurement information over multiple time periods. The aforementioned predictive indicator estimation procedure is: Based on the blood pressure measurement information, the daily variation characteristics of the subject's blood pressure measurement information are calculated. A recording medium according to Appendix 13 or 14, which estimates the signs of atrial fibrillation in the subject based on the aforementioned diurnal variation characteristics. (Note 16) The aforementioned precursor estimation procedure involves inputting the blood pressure measurement information into a precursor estimation model to estimate the precursor of atrial fibrillation in the subject, as described in any of appendices 13 to 15. (Note 17) It includes a learning information acquisition unit and a trained model generation unit, The aforementioned learning information acquisition unit acquires learning information, The aforementioned learning information includes the subject's blood pressure measurement information and whether or not the subject has developed atrial fibrillation. The pre-trained model generation unit generates a pre-trained model for predicting atrial fibrillation, which outputs an estimated result of predicting atrial fibrillation in a subject when the subject's blood pressure measurement information is input, using machine learning with the presence or absence of atrial fibrillation as the correct label for the blood pressure measurement information. (Note 18) The learning information includes at least one of the daily variation features and the daily difference features of the blood pressure measurement information. The pre-trained model generation unit generates a pre-trained model for predictive atrial fibrillation estimation, which outputs an estimated result of predictive atrial fibrillation in a subject when blood pressure measurement information of the subject is input, using machine learning with the presence or absence of the onset of atrial fibrillation as the correct label for at least one of the day-to-day variation feature and the day-to-day difference variation feature. This pre-trained model generation unit generates a pre-trained model for predictive atrial fibrillation estimation, as described in Appendix 17. (Note 19) The pre-trained model manufacturing apparatus for predictive behavior manufacturing according to Appendix 17 or 18, wherein the blood pressure measurement information is blood pressure measurement information from a predetermined reference period before the subject develops atrial fibrillation. (Note 20) This includes a process for acquiring training information and a process for generating a trained model. The aforementioned learning information acquisition process acquires learning information, The aforementioned learning information includes the subject's blood pressure measurement information and whether or not the subject has developed atrial fibrillation. The pre-trained model generation step is a method for manufacturing a pre-trained model for estimating the signs of atrial fibrillation, wherein the pre-trained model generates a pre-trained model that outputs an estimated result of the signs of atrial fibrillation in a subject when the subject's blood pressure measurement information is input, using machine learning with the presence or absence of the onset of atrial fibrillation as the correct label for the blood pressure measurement information. (Note 21) The learning information includes at least one of the daily variation features and the daily difference features of the blood pressure measurement information. The method for manufacturing a pre-trained model for predictive atrial fibrillation, as described in Appendix 20, wherein the pre-trained model generation step generates a pre-trained model that outputs an estimated result of predictive atrial fibrillation in a subject when blood pressure measurement information of the subject is input, by machine learning using the presence or absence of the onset of atrial fibrillation as the ground truth label for at least one of the day-to-day variation feature and the day-to-day difference variation feature. (Note 22) A method for manufacturing a trained model for predicting atrial fibrillation, as described in Appendix 20 or 21, wherein the blood pressure measurement information is blood pressure measurement information from a predetermined reference period before the subject develops atrial fibrillation. (Note 23) This includes procedures for acquiring training information and generating a trained model. The above procedure for acquiring learning information involves acquiring learning information, The aforementioned learning information includes the subject's blood pressure measurement information and whether or not the subject has developed atrial fibrillation. The aforementioned pre-trained model generation procedure is a program for manufacturing a pre-trained model for atrial fibrillation prediction, which generates a prediction prediction model that outputs an estimated result of the prediction of atrial fibrillation in a subject when the subject's blood pressure measurement information is input, using machine learning with the presence or absence of atrial fibrillation as the correct label for the blood pressure measurement information. (Note 24) The learning information includes at least one of the daily variation features and the daily difference features of the blood pressure measurement information. The pre-trained model generation procedure is a pre-trained model for manufacturing a pre-trained model for predictive atrial fibrillation, as described in Appendix 23. This pre-trained model generates a predictive atrial fibrillation prediction model that outputs an estimated result of predictive atrial fibrillation in a subject when blood pressure measurement information of the subject is input, using machine learning with the presence or absence of the onset of atrial fibrillation as the ground truth label for at least one of the day-to-day variation feature and the day-to-day difference variation feature. (Note 25) A method for manufacturing a trained model for predictive behavior as described in Appendix 23 or 24, wherein the blood pressure measurement information is blood pressure measurement information from a predetermined reference period before the subject develops atrial fibrillation. (Note 26) This includes procedures for acquiring training information and generating a trained model. The above procedure for acquiring learning information involves acquiring learning information, The aforementioned learning information includes the subject's blood pressure measurement information and whether or not the subject has developed atrial fibrillation. The aforementioned pre-trained model generation procedure generates a pre-trained model that, when the blood pressure measurement information of a subject is input, outputs an estimated result of the subject's atrial fibrillation precursor, using machine learning with the presence or absence of atrial fibrillation as the ground truth label for the blood pressure measurement information. A computer-readable recording medium containing a trained model manufacturing program for predicting atrial fibrillation, which is used to have a computer execute each step. (Note 27) The learning information includes at least one of the daily variation features and the daily difference features of the blood pressure measurement information. The recording medium described in Appendix 26 generates a pre-trained model, which outputs an estimated result of the premonitory signs of atrial fibrillation in a subject when blood pressure measurement information of the subject is input, by machine learning using the presence or absence of the onset of atrial fibrillation as the ground truth label for at least one of the day-to-day variation feature and the day-to-day difference variation feature. (Note 28) The recording medium according to Appendix 26 or 27, wherein the blood pressure measurement information is blood pressure measurement information from a predetermined reference period before the subject develops atrial fibrillation. [Industrial applicability]

[0076] According to this disclosure, it is possible to estimate the signs of atrial fibrillation in a subject based on blood pressure measurement information. For this reason, this disclosure is useful, for example, in the medical field. [Explanation of Symbols]

[0077] 10. Atrial fibrillation prediction support device 11. Blood pressure measurement information acquisition unit 12 Prediction Estimation Unit 13 Output section 101 Central Processing Unit 102 memory 103 Bus 104 Storage device 105 Input device 106 Output device 107 Communication devices 20. Machine for generating trained models for predicting atrial fibrillation. 21 Learning Information Acquisition Unit 22 Pre-trained model generation unit 201 Central Processing Unit 202 memory Bus 203 204 Storage device 205 Input device 206 Output device 207 Communication devices

Claims

1. This includes a procedure for acquiring blood pressure measurement information, a procedure for predicting a potential hazard, and an output procedure. The aforementioned blood pressure measurement information acquisition procedure acquires the blood pressure measurement information of the subject, The aforementioned precursor estimation procedure estimates the precursor of atrial fibrillation in the subject based on the blood pressure measurement information, The output procedure outputs the estimated result of the atrial fibrillation precursor. A program that assists in detecting atrial fibrillation by having a computer execute each step.

2. The aforementioned blood pressure measurement information acquisition procedure acquires blood pressure measurement information over multiple days. The aforementioned predictive indicator estimation procedure is: Based on the blood pressure measurement information, the daily variation characteristic of the subject's blood pressure measurement information is calculated. An atrial fibrillation prediction support program according to claim 1, which estimates the signs of atrial fibrillation in the subject based on the aforementioned diurnal variation features.

3. The blood pressure measurement information acquisition procedure acquires blood pressure measurement information over multiple time periods. The aforementioned predictive indicator estimation procedure is: Based on the blood pressure measurement information, the daily variation characteristics of the subject's blood pressure measurement information are calculated. An atrial fibrillation prediction support program according to claim 1 or 2, which estimates the presence of atrial fibrillation in the subject based on the aforementioned diurnal variation features.

4. The atrial fibrillation precursor detection support program according to claim 1 or 2, wherein the precursor estimation procedure involves inputting the blood pressure measurement information into a precursor estimation model to estimate the precursor of atrial fibrillation in the subject.

5. It includes a blood pressure measurement information acquisition unit, a prediction unit, and an output unit. The blood pressure measurement information acquisition unit acquires the blood pressure measurement information of the subject, The aforementioned precursor estimation unit estimates the precursor of atrial fibrillation in the subject based on the blood pressure measurement information, The output unit is an atrial fibrillation precursor detection support device that outputs the estimated result of the atrial fibrillation precursor.

6. This includes a blood pressure measurement information acquisition process, a prediction estimation process, and an output process. The blood pressure measurement information acquisition step involves acquiring the blood pressure measurement information of the subject, The aforementioned precursor estimation step estimates the precursor of atrial fibrillation in the subject based on the blood pressure measurement information, The output step outputs the estimated result of the atrial fibrillation precursor. A method for supporting the detection of atrial fibrillation, in which each step is performed by a computer.

7. This includes a procedure for acquiring blood pressure measurement information, a procedure for predicting a potential hazard, and an output procedure. The aforementioned blood pressure measurement information acquisition procedure acquires the blood pressure measurement information of the subject, The aforementioned precursor estimation procedure estimates the precursor of atrial fibrillation in the subject based on the blood pressure measurement information, The output procedure outputs the estimated result of the atrial fibrillation precursor. A computer-readable recording medium containing a program that assists in the detection of atrial fibrillation, which instructs a computer to perform each step of the procedure.

8. It includes a learning information acquisition unit and a trained model generation unit, The aforementioned learning information acquisition unit acquires learning information, The aforementioned learning information includes the subject's blood pressure measurement information and whether or not the subject has developed atrial fibrillation. The pre-trained model generation unit generates a pre-trained model for predicting atrial fibrillation, which outputs an estimated result of predicting atrial fibrillation in a subject when the subject's blood pressure measurement information is input, using machine learning with the presence or absence of atrial fibrillation as the correct label for the blood pressure measurement information.

9. The learning information includes at least one of the daily variation features and the daily difference features of the blood pressure measurement information. The pre-trained model generation unit generates a pre-trained model for predicting atrial fibrillation, which outputs an estimated result of predicting atrial fibrillation in a subject when blood pressure measurement information of the subject is input, by machine learning using the presence or absence of the onset of atrial fibrillation as the correct label for at least one of the day-to-day variation feature and the day-to-day difference variation feature.

10. The device for manufacturing a trained model for predicting atrial fibrillation according to claim 8 or 9, wherein the blood pressure measurement information is blood pressure measurement information from a predetermined reference period before the subject develops atrial fibrillation.

11. This includes a process for acquiring training information and a process for generating a trained model. The aforementioned learning information acquisition process acquires learning information, The aforementioned learning information includes the subject's blood pressure measurement information and whether or not the subject has developed atrial fibrillation. The pre-trained model generation step is a method for manufacturing a pre-trained model for estimating the signs of atrial fibrillation, wherein the pre-trained model generates a pre-trained model that outputs an estimated result of the signs of atrial fibrillation in a subject when the subject's blood pressure measurement information is input, using machine learning with the presence or absence of the onset of atrial fibrillation as the correct label for the blood pressure measurement information.

12. This includes procedures for acquiring training information and generating a trained model. The above procedure for acquiring learning information involves acquiring learning information, The aforementioned learning information includes the subject's blood pressure measurement information and whether or not the subject has developed atrial fibrillation. The aforementioned pre-trained model generation procedure is a program for manufacturing a pre-trained model for atrial fibrillation prediction, which generates a prediction prediction model that outputs an estimated result of the prediction of atrial fibrillation in a subject when the subject's blood pressure measurement information is input, using machine learning with the presence or absence of atrial fibrillation as the correct label for the blood pressure measurement information.

13. This includes procedures for acquiring training information and generating a trained model. The above procedure for acquiring learning information involves acquiring learning information, The aforementioned learning information includes the subject's blood pressure measurement information and whether or not the subject has developed atrial fibrillation. The aforementioned pre-trained model generation procedure generates a pre-trained model that, when the blood pressure measurement information of a subject is input, outputs an estimated result of the subject's atrial fibrillation precursor, using machine learning with the presence or absence of atrial fibrillation as the ground truth label for the blood pressure measurement information. A computer-readable recording medium containing a trained model manufacturing program for predicting atrial fibrillation, which is used to have a computer execute each step.

Images