Information processing device

Abstract

In the present invention, an acquisition unit acquires, by inputting electrocardiogram data of a subject to a first trained model trained to output an onset risk of a specific disease of a person and the certainty degree of the onset risk with respect to an input of the electrocardiogram data of the person, the onset risk and the certainty degree outputted by the first trained model. When the certainty degree is equal to or greater than a prescribed value, a main determination unit determines the onset risk of the subject on the basis of the onset risk and the certainty degree outputted by the first trained model. When the certainty degree is less than the prescribed value, a sub-determination unit acquires the cardiac age of the subject, and determines the onset risk of the subject on the basis of the acquired cardiac age. An output unit outputs a determination result by the main determination unit when determination is performed by the main determination unit, and outputs a determination result by the sub-determination unit when determination is performed by the sub-determination unit.

Description

[Technical Field]

[0001] The present invention relates to an information processing apparatus, an information processing method, and a recording medium for predicting the onset risk of a specific disease. [Background Art]

[0002] Predicting the onset risk of a specific disease has been performed by analyzing measured electrocardiogram of a subject.

[0003] For example, Patent Document 1 discloses an apparatus which uses electrocardiogram data for training obtained during a non-paroxysmal period where paroxysmal arrhythmia does not occur in a patient with paroxysmal arrhythmia to obtain a machine-learned model, inputs electrocardiogram data of a subject to the model, and outputs abnormality information output from the model regarding whether the subject has paroxysmal arrhythmia.

[0004] Further, Non-Patent Document 1 discloses an apparatus that uses a DNN (Deep Neural Network) machine-learned using resting electrocardiogram data for training, gender and age to predict new onset atrial fibrillation in a subject without atrial fibrillation. [Prior Art Documents] [Patent Documents]

[0005] [Patent Document 1] Japanese Unexamined Patent Publication No. 2022-37153 [Non-Patent Documents]

[0006] [Non-Patent Document 1] Sushravya Raghunath et al. "Deep Neural Networks Can Predict New-Onset Atrial Fibrillation From the 12-Lead ECG and Help Identify Those at Risk of Atrial Fibrillation-Related Stroke". Circulation. 2021;143:1287-1298. DOI: 10.1161 / CIRCULATIONAHA.120.047829. [Overview of the project] [Problems that the invention aims to solve]

[0007] When predicting the risk of developing a specific disease based on an electrocardiogram (ECG) obtained from a subject, information about the subject's physical condition is considered in addition to the ECG, as described in Non-Patent Document 1. However, it has been difficult to improve accuracy with information that does not match the subject's actual physical condition.

[0008] The object of the present invention is to provide an information processing device that solves the above-mentioned problems. [Means for solving the problem]

[0009] An information processing device according to one embodiment of the present invention is An acquisition unit acquires the risk of developing a specific disease and the confidence level of that risk by inputting the electrocardiogram data of a subject into a first trained model that has been trained to output the risk of developing a specific disease and the confidence level of that risk in response to the input of the person's electrocardiogram data, thereby acquiring the risk of developing a disease and the confidence level output by the first trained model. If the confidence level is above a predetermined value, the main determination unit determines the subject's risk of developing the disease based on the risk of developing the disease output by the first trained model and the confidence level. If the confidence level is less than a predetermined value, the sub-determination unit obtains the subject's cardiac age and determines the subject's risk of developing the disease based on the obtained cardiac age. An output unit that outputs the determination result of the main determination unit when a determination is made by the main determination unit, and outputs the determination result of the sub-determination unit when a determination is made by the sub-determination unit, It is configured to include the following:

[0010] Information processing methods according to other embodiments of the present invention are: By inputting the electrocardiogram data of a subject into a first trained model that has been trained to output the risk of developing a specific disease and the confidence level of said risk in response to the input of the person's electrocardiogram data, the risk of developing the disease and the confidence level output by the first trained model are obtained. If the confidence level is above a predetermined value, the subject's risk of developing the disease is determined based on the disease risk output by the first trained model and the confidence level. If the confidence level is less than a predetermined value, the subject's cardiac age is obtained, and the subject's risk of developing the disease is determined based on the obtained cardiac age. If the confidence level is equal to or greater than a predetermined value, the system outputs the disease risk output by the first trained model and the judgment result based on the confidence level; if the confidence level is less than a predetermined value, the system outputs the judgment result based on the cardiac age. It is structured in this way.

[0011] Furthermore, a computer-readable recording medium according to another embodiment of the present invention is: On the computer, The process involves inputting the electrocardiogram data of a subject into a first trained model that has been trained to output the risk of developing a specific disease and the confidence level of that risk in response to the input of the person's electrocardiogram data, thereby obtaining the risk of developing the disease and the confidence level output by the first trained model. If the confidence level is above a predetermined value, the process of determining the subject's risk of developing the disease is performed based on the risk of developing the disease output by the first trained model and the confidence level. If the confidence level is less than a predetermined value, the process involves obtaining the subject's cardiac age and determining the subject's risk of developing the disease based on the obtained cardiac age. a process of outputting an onset risk output by said first trained model and a determination result based on said certainty degree when said certainty degree is equal to or higher than a predetermined value, and outputting a determination result based on said heart age when said certainty degree is less than said predetermined value; The computer-readable medium is configured to store a program for causing a computer to perform the above processing.

Effects of the Invention

[0012] The present invention, having the configuration as described above, can improve the prediction accuracy of the onset risk of a specific disease in a subject based on electrocardiogram data.

Brief Description of Drawings

[0013] [Figure 1] It is a block diagram of the information processing apparatus according to the first embodiment of the present invention. [Figure 2] It is a flowchart showing an example of the operation in the operation phase of the information processing apparatus according to the first embodiment of the present invention. [Figure 3] It is a block diagram of the information processing apparatus according to the second embodiment of the present invention.

Mode for Carrying Out the Invention

[0014] [First Embodiment] Figure 1 is a block diagram of the information processing apparatus 10 according to the first embodiment of the present invention. The information processing apparatus 10 has a function of predicting and outputting the onset risk of a specific disease in a subject based on electrocardiogram data acquired from the subject.

[0015] Electrocardiogram data is data indicating measurement results relating to an electrocardiogram of a subject. In the present embodiment, the electrocardiogram data is assumed to be 12-lead electrocardiogram data obtained by an examination in which a total of 10 electrodes are attached to 6 positions on the chest and both wrists and both ankles to record electrical activities and changes of the heart. However, the electrocardiogram data may be electrocardiogram data other than 12-lead electrocardiogram data. Further, the electrocardiogram data may be raw data of the subject's electrocardiogram output by an electrocardiogram measurement device, or may be an electrocardiogram file converted into a predetermined format (e.g., an electrocardiogram image or a PDF file).

[0016] Further, in the present embodiment, it is assumed that the specific disease is atrial fibrillation. However, the specific disease is not limited to atrial fibrillation, and may be other cardiovascular diseases such as ventricular tachycardia, supraventricular tachycardia, atrial flutter, and ventricular fibrillation. Further, in the present embodiment, the onset risk is defined as whether or not the specific disease newly develops within a certain period (e.g., within one year) in a subject who does not have the specific disease. Hereinafter, having an onset risk is referred to as "positive", and not having an onset risk is referred to as "negative". However, the onset risk is not limited to two categories such as positive and negative, and may be obtained by classifying the probability or degree of onset into three or more categories.

[0017] As main functional units, the information processing apparatus 10 includes a communication interface unit (hereinafter referred to as a communication I / F unit) 11, an operation input unit 12, a screen display unit 13, a storage unit 14, and an arithmetic processing unit 15.

[0018] The communication I / F unit 11 is composed of a dedicated data communication circuit, and has a function of performing data communication with various devices such as an electrocardiogram measurement apparatus (not shown) and a medical staff terminal (not shown) connected via a wired or wireless connection. The operation input unit 12 is composed of an operation input device such as a keyboard or a mouse, and has a function of detecting an operator's operation and outputting the detected operation to the arithmetic processing unit 15. The screen display unit 13 is composed of a screen display device such as an LCD (Liquid Crystal Display) or a PDP (Plasma Display Panel), and has a function of screen-displaying various information such as the predicted onset risk in accordance with an instruction from the arithmetic processing unit 15.

[0019] The memory unit 14 consists of a storage device such as a hard disk or memory, and has the function of storing processing information and programs 141 necessary for various processes in the arithmetic processing unit 15. The programs 141 are programs that realize various processing processes when read and executed by the arithmetic processing unit 15, and are pre-read from external devices (not shown) or storage media (not shown) via data input / output functions such as the communication I / F unit 11 and stored in the memory unit 14. The main processing information stored in the memory unit 14 includes a specific disease learned model 142 and a cardiac age learned model 143.

[0020] The specific disease-trained model 142 is a model trained to output the risk of a person developing a specific disease and the confidence level of that risk in response to the input of the person's electrocardiogram data. The confidence level is a probability that represents the likelihood of the model's judgment result, with a higher value indicating greater likelihood. In this embodiment, the confidence level is set to be between 0 and 1. The specific disease-trained model 142 may output the confidence level for either the presence or absence of the risk of developing the disease. Alternatively, the specific disease-trained model 142 may output the confidence level for the presence of the risk of developing the disease and the confidence level for the absence of the risk of developing the disease, with the sum of the two confidence levels being 1. The specific disease-trained model 142 may be created on a computer different from the information processing device 10 and stored in the storage unit 14. Alternatively, the specific disease-trained model 142 may be created on the information processing device 10 and stored in the storage unit 14.

[0021] The cardiac age trained model 143 is a model trained to output a person's cardiac age in response to an input of electrocardiogram data. Cardiac age is an indicator of the degree of cardiac aging, and by comparing it with chronological age (actual age), it is possible to know how much the heart has aged. The cardiac age trained model 143 may be created on a computer different from the information processing device 10 and stored in the memory unit 14. Alternatively, the cardiac age trained model 143 may be created in the information processing device 10 and stored in the memory unit 14.

[0022] For the above-mentioned pre-trained models for specific diseases 142 and cardiac age 143, for example, a DNN may be used. However, models such as support vector machines, decision trees, random forests, and logistic regression may also be used for these pre-trained models.

[0023] The arithmetic processing unit 15 has a microprocessor such as a CPU (Central Processing Unit) and its peripheral circuits, and has the function of realizing various processing functions by having the above hardware and program 141 cooperate to read and execute the program 141 from the storage unit 14. The main processing functions realized by the arithmetic processing unit 15 are the acquisition unit 151, the main determination unit 152, the sub-determination unit 153, and the output unit 154.

[0024] The acquisition unit 151 is configured to acquire the subject's electrocardiogram data. The acquisition unit 151 may acquire the subject's electrocardiogram data from, for example, an electrocardiogram measuring device (not shown) via the communication interface unit 11. Alternatively, the acquisition unit 151 may acquire the subject's electrocardiogram data via the communication interface unit 11 from an electronic medical record (not shown) that records basic physical data such as the subject's age, sex, height, and weight, measurement data such as heart rate, body temperature, blood pressure, and electrocardiogram data, and medical condition data such as state of consciousness, current or past illnesses (medical history), circumstances at the time of examination, and circumstances at the time of testing. Alternatively, if the subject's electrocardiogram data is stored in the storage unit 14 beforehand, the acquisition unit 151 may acquire the subject's electrocardiogram data from the storage unit 14.

[0025] Furthermore, the acquisition unit 151 is configured to input the acquired electrocardiogram data into a specific disease-learned model 142 and to acquire the risk of developing the specific disease and the confidence level of that risk, which are output from the learning model. The acquisition unit 151 may also acquire the risk of developing the disease and the confidence level calculated based on the subject's electrocardiogram data using a computer different from the information processing device 10.

[0026] The main determination unit 152 is configured to determine whether the subject's risk of developing a specific disease is positive or negative, based on the disease risk and confidence level obtained by the acquisition unit 151, if the confidence level obtained by the acquisition unit 151 is above a threshold. The main determination unit 152 determines the disease risk with the higher confidence level output by the specific disease trained model 142 as the subject's disease risk.

[0027] As mentioned above, for example, the sum of the confidence levels for determining whether there is a risk of developing the disease ("present" or "absent") is 1. The result with the higher confidence level is output as the determination. For example, if the confidence level for "present" is 0.9 and the confidence level for "absent" is 0.1, the determination of the risk of developing the disease will be "present". If the confidence level for "present" is 0.1 and the confidence level for "absent" is 0.9, the determination of the risk of developing the disease will be "absent". The maximum confidence level for "present" is 1, and the closer to 1, the higher the confidence level of the determination. Therefore, a positive result determined when the confidence level for "present" is above a threshold close to 1 can be said to be a clear positive or a strongly positive. Similarly, a negative result determined when the confidence level for "absent" is above a threshold close to 1 can be said to be a clear negative or a strongly negative. The threshold can be set as appropriate, for example, to a value of 0.5 or higher, depending on the usage scenario. Therefore, for example, if the threshold is set to 0.6, a positive result between 0.5 and 0.6 can be said to be a positive with low confidence or a weak positive. A negative result between 0.5 and 0.6 can be considered a low-confidence negative or weakly negative. In such cases, it is judged as "questionable."

[0028] The main determination unit 152 is configured to transmit the determination result to the output unit 154.

[0029] Furthermore, the information processing device 10 may include a specific disease model learning unit 155. The specific disease model learning unit 155 has the function of creating a specific disease trained model 142. For example, the specific disease model learning unit 155 generates a specific disease trained model 142 by machine learning using a machine learning algorithm such as DNN, using electrocardiogram data obtained from various people, similar to Patent Document 1 and Non-Patent Document 1, and the presence or absence of the specific disease in those people as training data.

[0030] The secondary determination unit 153 is configured to acquire the subject's cardiac age if the confidence level acquired by the acquisition unit 151 is below a threshold, and to determine the subject's risk of developing the disease as either positive or negative based on the acquired cardiac age. Alternatively, the acquisition unit 151 may acquire the subject's cardiac age. In this case, the acquisition unit 151 acquires the cardiac age upon receiving instructions from the secondary determination unit 153.

[0031] The secondary determination unit 153 may use the cardiac age trained model 143 to obtain the subject's cardiac age. Specifically, the secondary determination unit 153 inputs the electrocardiogram data acquired by the acquisition unit 151 into the cardiac age trained model 143 and obtains the predicted cardiac age output from the trained model.

[0032] Furthermore, the sub-determination unit 153 uses, for example, one of the following two determination methods.

[0033] <Judgment method 1> The secondary determination unit 153 makes a determination based on the result of comparing cardiac age with a threshold. The threshold may be, for example, 65 years old. This is based on the finding that the incidence of atrial fibrillation is high in elderly people aged 65 and over. However, the threshold is not limited to 65 years old.

[0034] The sub-determination unit 153 determines, for example, that the cardiac age is positive if it is above a threshold, and negative if it is below a threshold. The reason for this is that a cardiac age below a threshold suggests that the degree of cardiac aging is not as advanced as that of a person at the threshold age, while a cardiac age above a threshold suggests that the degree of cardiac aging is more advanced than that of a person at the threshold age.

[0035] Furthermore, the sub-determination unit 153 may perform a determination using the disease risk acquired by the acquisition unit 151. That is, the sub-determination unit 153 may be configured to determine the subject's disease risk as either positive or negative based on the acquired cardiac age and the disease risk output by the specific disease-learned model 142. For example, the sub-determination unit 153 determines a subject to be negative only if the disease risk output by the specific disease-learned model 142 is negative and the cardiac age is below a threshold; otherwise, it determines a subject to be positive. By classifying only subjects who are considered to have a reliably low risk of developing the disease as negative, it is possible to prevent overlooking subjects who have a positive risk of developing the disease.

[0036] <Judgment method 2> The sub-determination unit 153 makes a determination based on the result of comparing the cardiac age with the subject's chronological age. The sub-determination unit 153 may input the subject's chronological age from the operator via the operation input unit 12, or it may obtain it from an electronic medical record (not shown) that records the subject's age, etc., via the communication I / F unit 11.

[0037] The secondary determination unit 153 determines, for example, that the heart age is positive if it is equal to or greater than the chronological age, and negative if it is less than the chronological age. The reason for this is that a heart age less than the chronological age suggests that the degree of cardiac aging is not as advanced as in people of the same chronological age, while a heart age equal to or greater than the chronological age suggests that the degree of cardiac aging is more advanced than in people of the same chronological age.

[0038] Furthermore, the sub-determination unit 153 may perform a determination using the disease risk acquired by the acquisition unit 151. That is, the sub-determination unit 153 may be configured to determine the subject's disease risk as either positive or negative based on the acquired cardiac age and the disease risk output by the specific disease-learned model 142. For example, the sub-determination unit 153 determines a subject to be negative only if the disease risk output by the specific disease-learned model 142 is negative and the cardiac age is less than the chronological age; otherwise, it determines a subject to be positive. Similar to determination method 1, by classifying only subjects who are considered to have a reliably low risk of developing the disease as negative, it is possible to prevent overlooking subjects with a positive risk of developing the disease.

[0039] The sub-determination unit 153 is configured to transmit the result, which is determined to be either positive or negative, to the output unit 154.

[0040] Furthermore, the information processing device 10 may include a cardiac age model learning unit 156. The cardiac age model learning unit 156 has the function of creating a cardiac age trained model 143. For example, the cardiac age model learning unit 156 generates a cardiac age trained model 143 by machine learning using a machine learning algorithm such as a DNN, using electrocardiogram data of various people and the cardiac age of various people determined by medical professionals based on the results of cardiac ultrasound examinations of those people as training data.

[0041] The output unit 154 receives the determination results from the main determination unit 152 and the sub-determination unit 153. If the main determination unit 152 determines whether the result is positive or negative, the output unit 154 selects the determination result from the main determination unit 152. If the sub-determination unit 153 determines whether the result is positive or negative, the output unit 154 selects the determination result from the sub-determination unit 153. The output unit 154 is also configured to display the selected determination result on the screen display unit 13 and / or transmit it to an external device via the communication interface unit 11.

[0042] Next, the operation of the information processing device 10 will be described. The operation of the information processing device 10 can be broadly divided into a learning phase and an operation phase.

[0043] During the learning phase, the information processing device 10 creates a specific disease-trained model 142 and a cardiac age-trained model 143. The specific disease-trained model 142 is created by the specific disease model learning unit 155. The specific disease model learning unit 155 generates the specific disease-trained model 142 using machine learning algorithms such as DNN, with electrocardiogram data obtained from various individuals and the presence or absence of specific diseases in those individuals as training data. However, if the specific disease-trained model 142 has already been created on a computer other than the information processing device 10, the process of creating the specific disease-trained model 142 by the specific disease model learning unit 155 is omitted. The cardiac age-trained model 143 is created by the cardiac age model learning unit 156. The cardiac age model learning unit 156 generates the cardiac age-trained model 143 using machine learning algorithms such as DNN, with electrocardiogram data from various individuals and the cardiac age of those individuals determined by medical professionals based on the results of cardiac ultrasound examinations of those individuals as training data. However, if the cardiac age-learned model 143 has already been created on a computer different from the information processing device 10, the process of creating the cardiac age-learned model 143 by the cardiac age model learning unit 156 is omitted.

[0044] During the operational phase, the information processing device 10 uses the specific disease-learned model 142 and the cardiac age-learned model 143 to predict and output the risk of developing specific diseases in the subject based on electrocardiogram data acquired from the subject. The operational phase processing will be explained below with reference to Figure 2. Figure 2 is a flowchart showing an example of the operation of the information processing device 10 during the operational phase.

[0045] First, the acquisition unit 151 acquires the subject's electrocardiogram data from an electrocardiogram measurement device (not shown) or the subject's electronic medical record (not shown) (step S1). Next, the acquisition unit 151 inputs the acquired electrocardiogram data into the specific disease-learned model 142 and acquires the subject's risk of developing the specific disease and the confidence level of that risk, which are output from the learning model (step S2). Next, the main determination unit 152 determines the subject's risk of developing the specific disease based on the risk of developing the disease and the confidence level of that risk, if the confidence level acquired by the acquisition unit 151 is above a threshold (YES in step S3) (step S4). Then, the output unit 154 displays the determination result of the main determination unit 152 on the screen display unit 13 and / or transmits it to an external device via the communication I / F unit 11 (step S8). Finally, the information processing device 10 completes the process shown in Figure 2.

[0046] On the other hand, if the confidence level obtained by the acquisition unit 151 is below the threshold (NO in step S3), the sub-determination unit 153 acquires the subject's electrocardiogram data in the same manner as the main determination unit 152 (step S5). Next, the sub-determination unit 153 inputs the acquired electrocardiogram data into the cardiac age learned model 143 and obtains the predicted cardiac age output from the learned model (step S6). Next, the sub-determination unit 153 uses the aforementioned determination method 1 or determination method 2 to determine the subject's risk of developing the disease based on their cardiac age (step S7). Then, the output unit 154 displays the determination result of the sub-determination unit 153 on the screen display unit 13 and / or transmits it to an external device via the communication I / F unit 11 (step S8). Finally, the process shown in Figure 4 is completed.

[0047] As explained above, the information processing device 10 predicts the risk of developing a specific disease by considering not only electrocardiogram data but also cardiac age, which is information representing the subject's actual physical condition. Therefore, it can improve the accuracy of predicting the risk of developing a specific disease in the subject compared to prediction based on electrocardiogram data alone.

[0048] Furthermore, if the confidence level of the subject's risk of developing a specific disease, output from the specific disease-trained model 142, is above a threshold, the information processing device 10 determines the subject's risk of developing a specific disease based on the risk of development and the confidence level of that risk, output from the specific disease-trained model 142. This is because it takes into account that the prediction accuracy is high when the confidence level of the risk of development output from the specific disease-trained model 142 is above a threshold. As a result, it is possible to prevent unnecessary determination based on cardiac age.

[0049] Furthermore, the information processing device 10 obtains the subject's cardiac age based on the subject's electrocardiogram data. Therefore, compared to performing an ultrasound examination on the subject and obtaining the subject's cardiac age determined by a medical professional based on the results, the required examination equipment and the burden on the subject can be significantly reduced.

[0050] Next, a modified example of this embodiment will be described.

[0051] The sub-determination unit 153 may obtain the subject's heart age in a manner that does not utilize the heart age learned model 143. For example, if the subject's electronic medical record contains information about their heart age, the sub-determination unit 153 may obtain the subject's heart age from the electronic medical record. Alternatively, the sub-determination unit 153 may display several questions about the subject's physical condition on the screen display unit 13 and calculate the subject's heart age by performing predetermined calculations based on the answers to these questions. For example, the questions could include, but are not limited to, chronological age, sex, smoking status, height, weight, presence or absence of diabetes, rheumatoid arthritis, chronic kidney disease, or atrial fibrillation, presence or absence of cardiovascular disease in blood relatives, cholesterol levels, and blood pressure.

[0052] If the test result output by the output unit 154 is positive, the service may be provided to guide the subject to optional tests or detailed examinations (such as a 24-hour Holter electrocardiogram).

[0053] [Second Embodiment] Next, an information processing device 20 according to a second embodiment of the present invention will be described. Figure 3 is a block diagram of the information processing device 20.

[0054] Referring to Figure 3, the information processing device 20 includes an acquisition unit 21, a main determination unit 22, a sub-determination unit 23, and an output unit 24.

[0055] The acquisition unit 21 is configured to acquire the risk of developing a specific disease and the confidence level of that risk by inputting the subject's electrocardiogram data into a first trained model that has been trained to output the risk of developing a specific disease and the confidence level of that risk in response to the input of the person's electrocardiogram data. The acquisition unit 21 can be configured, for example, in the same way as the acquisition unit 151 in Figure 1, but is not limited thereto. The main determination unit 22 is configured to determine the subject's risk of developing the disease based on the risk of developing the disease output by the first trained model and the confidence level obtained by the acquisition unit 21, if the confidence level obtained by the acquisition unit 21 is equal to or greater than a predetermined value. The main determination unit 22 can be configured in the same way as, for example, the main determination unit 152 in Figure 1, but is not limited thereto.

[0056] The sub-determination unit 23 is configured to acquire the subject's cardiac age if the confidence level acquired by the acquisition unit 21 is less than a predetermined value, and to determine the subject's risk of developing the disease based on the acquired cardiac age. The sub-determination unit 23 can be configured similarly to, for example, the sub-determination unit 153 in Figure 1, but is not limited thereto.

[0057] The output unit 24 is configured to output the determination result made by the main determination unit 22 when the determination is made by the main determination unit 22, and to output the determination result made by the sub-determination unit 23 when the determination is made by the sub-determination unit 23. The output unit 24 can be configured in the same way as, for example, the output unit 154 in Figure 1, but is not limited to that.

[0058] The information processing device 20 having the configuration described above operates as follows. First, the acquisition unit 21 inputs the subject's electrocardiogram data into a first trained model, which has been trained to output the risk of developing a specific disease and the confidence level of this risk in response to the input of the person's electrocardiogram data, thereby acquiring the risk of developing the disease and its confidence level output by the first trained model. Next, the main determination unit 22 determines the subject's risk of developing the disease based on the risk of developing the disease and its confidence level output by the first trained model, if the confidence level acquired by the acquisition unit 21 is greater than or equal to a predetermined value. On the other hand, the sub-determination unit 23 acquires the subject's cardiac age and determines the subject's risk of developing the disease based on the acquired cardiac age, if the confidence level acquired by the acquisition unit 21 is less than a predetermined value. Finally, the output unit 24 outputs the determination result by the main determination unit 22 if the determination is made by the main determination unit 22, and outputs the determination result by the sub-determination unit 23 if the determination is made by the sub-determination unit 23.

[0059] The information processing device 20, configured and operating as described above, can improve the accuracy of predicting the risk of developing a specific disease in a subject compared to predictions based solely on electrocardiogram data. This is because, in addition to electrocardiogram data, it considers cardiac age, which represents the subject's actual physical condition, to predict the risk of developing a specific disease.

[0060] Although the present invention has been described above with reference to the embodiments described above, the present invention is not limited to the embodiments described above. Various modifications to the configuration and details of the present invention can be made within the scope of the present invention as can be understood by those skilled in the art.

[0061] For example, instead of the CPU mentioned above, an information processing device can use a GPU (Graphic Processing Unit), DSP (Digital Signal Processor), MPU (Micro Processing Unit), FPU (Floating Number Processing Unit), PPU (Physics Processing Unit), TPU (Tensor Processing Unit), quantum processor, microcontroller, or a combination of these. [Industrial applicability]

[0062] This invention can be used to support diagnosis through electrocardiogram analysis, aiming for the early detection of cardiovascular diseases, including silent atrial fibrillation. [Explanation of Symbols]

[0063] 10, 20 Information Processing Devices 11. Communication I / F Section 12 Operation Input Section 13 Screen display section 14 Storage section 15. Arithmetic Processing Unit 21, 151 Acquisition Department 22, 152 Main judgment section 23, 153 Subjudgment Department 24, 154 Output section 141 Programs 142 Specific Disease Trained Models 143 Heart Age Learned Model 155 Specific Disease Model Learning Department 156 Cardiac Age Model Learning Section

Claims

1. An acquisition unit acquires the risk of developing a specific disease and the confidence level of that risk by inputting the electrocardiogram data of a subject into a first trained model that has been trained to output the risk of developing a specific disease and the confidence level of that risk in response to the input of the person's electrocardiogram data, thereby acquiring the risk of developing a disease and the confidence level output by the first trained model. If the confidence level is above a predetermined value, the main determination unit determines the subject's risk of developing the disease based on the risk of developing the disease output by the first trained model and the confidence level. If the confidence level is less than a predetermined value, the sub-determination unit obtains the subject's cardiac age and determines the subject's risk of developing the disease based on the obtained cardiac age. An output unit that outputs the determination result of the main determination unit when a determination is made by the main determination unit, and outputs the determination result of the sub-determination unit when a determination is made by the sub-determination unit, Equipped with, The sub-determination unit is an information processing device that, in acquiring the cardiac age, inputs the subject's electrocardiogram data into a second trained model that has been trained to output the person's cardiac age in response to the input of the person's electrocardiogram data, thereby acquiring the cardiac age output by the second trained model.

2. The second trained model was trained using machine learning with electrocardiogram data from various individuals and cardiac age determined by medical professionals based on the results of cardiac ultrasound examinations of those individuals. The information processing apparatus according to claim 1.

3. A computer, By inputting the electrocardiogram data of a subject into a first trained model that has been trained to output the risk of developing a specific disease and the confidence level of said risk in response to the input of the person's electrocardiogram data, the risk of developing the disease and the confidence level output by the first trained model are obtained. If the confidence level is above a predetermined value, the risk of developing the disease for the subject is determined based on the disease risk output by the first trained model and the confidence level. If the confidence level is less than a predetermined value, the subject's cardiac age is obtained, and the subject's risk of developing the disease is determined based on the obtained cardiac age. If the confidence level is greater than or equal to a predetermined value, the onset risk output by the first trained model and the judgment result based on the confidence level are output; if the confidence level is less than a predetermined value, the judgment result based on the cardiac age is output. In obtaining the cardiac age, the electrocardiogram data of the subject is input to a second trained model that has been trained to output the cardiac age of a person in response to the input of the person's electrocardiogram data, thereby obtaining the cardiac age output by the second trained model. Information processing methods.

4. On the computer, The process involves inputting the electrocardiogram data of a subject into a first trained model that has been trained to output the risk of developing a specific disease and the confidence level of that risk in response to the input of the person's electrocardiogram data, thereby obtaining the risk of developing the disease and the confidence level output by the first trained model. If the confidence level is above a predetermined value, the process of determining the subject's risk of developing the disease is performed based on the risk of developing the disease output by the first trained model and the confidence level. If the confidence level is less than a predetermined value, the process involves obtaining the subject's cardiac age and determining the subject's risk of developing the disease based on the obtained cardiac age. If the confidence level is greater than or equal to a predetermined value, the process outputs the disease risk output by the first trained model and the judgment result based on the confidence level; if the confidence level is less than a predetermined value, the process outputs the judgment result based on the cardiac age. Have them do it, The program for obtaining the cardiac age involves inputting the subject's electrocardiogram data into a second trained model that has been trained to output the person's cardiac age in response to the person's electrocardiogram data, thereby obtaining the cardiac age output by the second trained model.

Images