Computer program, information processing method, information processing device, information processing system, and method for generating trained model

Abstract

Provided are a computer program, an information processing method, an information processing device, an information processing system, and a method for generating a trained model for evaluating the exacerbation of heart failure more accurately than before.　This computer program causes a computer to execute processing comprising: acquiring first biological information related to hemodynamics of a patient at a first time point and second biological information related to a compensatory mechanism associated with heart failure of the patient at a second time point different from the first time point; and deriving an index related to the progression level of exacerbation of heart failure of the patient at an evaluation time point later than the first time point and the second time point on the basis of the first biological information and the second biological information acquired.

Description

Computer program, information processing method, information processing device, information processing system, and method for generating trained model

[0001] The present invention relates to a computer program, an information processing method, an information processing device, an information processing system, and a method for generating a trained model for evaluating the worsening of heart failure.

[0002] Methods for assessing the worsening of heart failure have been explored. For example, there is a method that uses specific hormones contained in the blood, such as BNP (brain natriuretic peptide), as a biomarker for the worsening of heart failure. Furthermore, a technique for assessing the worsening of heart failure using biological information representing the patient's physiological state has been proposed. Patent Document 1 discloses a technique for determining the amount of water contained in the body and assessing the worsening of heart failure based on the determined amount of water.

[0003] Special table 2019-509153 publication

[0004] Using a patient's biological information makes it easier to routinely evaluate the progression of heart failure. However, biological information fluctuates due to various factors, such as the patient's movements. This can lead to variations in the results of evaluating the progression of heart failure.

[0005] An object of the present invention is to provide a computer program, an information processing method, an information processing device, an information processing system, and a method for generating a trained model for evaluating the worsening of heart failure with greater accuracy than conventional methods.

[0006] (1) A computer program according to one embodiment of the present invention is characterized in that it causes a computer to acquire first biological information relating to a patient's hemodynamics at a first time point and second biological information relating to compensatory mechanisms associated with the patient's heart failure at a second time point different from the first time point, and derive an index relating to the degree of progression of the patient's heart failure at an evaluation time point that is later than the first time point and the second time point based on the acquired first biological information and second biological information.

[0007] (2) In the computer program of (1), the second point in time is preferably a point in time earlier than the first point in time.

[0008] (3) In the computer program of (1) or (2), it is preferable that the first biological information includes pulse rate variability, and the second biological information includes a body water content.

[0009] (4) In any one of the computer programs (1) to (3), it is preferable that the computer executes a process of intermittently or continuously acquiring the first biometric information and the second biometric information and intermittently or continuously deriving the index.

[0010] (5) It is preferable that any one of the computer programs (1) to (4) causes a computer to execute a process of calculating a representative value of the index within a predetermined period of time from the derived multiple indexes.

[0011] (6) It is preferable that the computer program of any one of (1) to (5) above causes a computer to execute a process of outputting a graph showing a change in the index over time.

[0012] (7) It is preferable that any one of the computer programs (1) to (6) causes a computer to execute a process of outputting the index and the first biometric information and the second biometric information used to derive the index.

[0013] (8) It is preferable that any one of the computer programs (1) to (7) causes a computer to execute a process of deriving the index based on the first biometric information, the second biometric information, and personal information of the patient.

[0014] (9) In the computer program according to any one of (1) to (8), the index is preferably an estimated value of blood concentration of BNP (Brain Natriuretic Peptide) or NT-proBNP (N-terminal proBNP).

[0015] (10) It is preferable that any one of the computer programs of (1) to (9) causes a computer to execute a process of deriving the index by inputting the acquired first biometric information and the second biometric information into a trained model that outputs the index when the first biometric information and the second biometric information are input, and acquiring the index output by the trained model.

[0016] (11) In any one of the computer programs (1) to (10), it is preferable that the first biological information includes pulse variability, pulse rate, blood pressure, and respiratory rate, and the first time point is a different time point with respect to at least two of the pulse variability, pulse rate, blood pressure, and respiratory rate.

[0017] (12) It is preferable that any one of the computer programs (1) to (11) acquires activity amount information representing the patient's activity amount at a predetermined time point relative to the first time point, the second time point, and the evaluation time point, and causes a computer to execute a process of deriving the index based on the first biometric information, the second biometric information, and the activity amount information.

[0018] (13) It is preferable that any one of the computer programs (1) to (12) causes a computer to execute a process of outputting a graph showing the change over time of the index, a graph showing the change over time of the first biometric information, and a graph showing the change over time of the second biometric information, with the times shown by each graph aligned to be the same.

[0019] (14) It is preferable that any one of the computer programs (1) to (12) causes a computer to execute a process of calculating a representative value of the derived multiple indicators within a predetermined period of time, calculating a representative value of the first biometric information and the second biometric information within the period from the acquired multiple first biometric information and the second biometric information, and outputting a graph showing the time change of the representative value of the indicators, a graph showing the time change of the representative value of the first biometric information, and a graph showing the time change of the representative value of the second biometric information, with the times shown by each graph aligned to the same time.

[0020] (15) An information processing method according to one embodiment of the present invention is characterized in that it acquires first biological information relating to a patient's hemodynamics at a first time point and second biological information relating to compensatory mechanisms associated with the patient's heart failure at a second time point different from the first time point, and derives an index relating to the degree of progression of the patient's heart failure at an evaluation time point that is later than the first time point and the second time point based on the acquired first biological information and second biological information.

[0021] (16) An information processing device according to one embodiment of the present invention includes a calculation unit, which acquires first biological information relating to the hemodynamics of a patient at a first time point and second biological information relating to compensatory mechanisms associated with the patient's heart failure at a second time point different from the first time point, and derives an index relating to the degree of progression of the patient's heart failure at an evaluation time point that is later than the first time point and the second time point based on the acquired first biological information and second biological information.

[0022] (17) An information processing system according to one embodiment of the present invention comprises a wearable device and an information processing device, wherein the wearable device has a first sensor that measures data necessary to generate first biometric information regarding the hemodynamics of a patient wearing the wearable device, and a second sensor that measures data necessary to generate second biometric information regarding compensatory mechanisms associated with heart failure, and the information processing device has a calculation unit that generates the first biometric information at a first time point based on data measured by the first sensor, generates the second biometric information at a second time point different from the first time point based on data measured by the second sensor, and derives an index regarding the progression of the patient's heart failure at an evaluation time point that is later than the first time point and the second time point based on the generated first biometric information and second biometric information.

[0023] (18) In the information processing system of (17), it is preferable that the wearable device measures data intermittently or continuously using the first sensor and the second sensor, and the calculation unit generates the first biometric information and the second biometric information intermittently or continuously.

[0024] (19) A method for generating a trained model according to one embodiment of the present invention is characterized in that it acquires training data including first biological information relating to a patient's hemodynamics at a first time point, second biological information relating to a compensatory mechanism associated with the patient's heart failure at a second time point different from the first time point, and a correct value of an index relating to the progression of the patient's heart failure at an evaluation time point that is later than the first time point and the second time point, and generates a trained model that outputs the index when the first biological information and the second biological information are input by learning based on the training data.

[0025] (20) In the method for generating a trained model of (19), it is preferable that the correct value of the index included in the training data is a value related to a component in blood, and the index output by the trained model is an estimated value of the value related to the component in blood.

[0026] In one aspect of the present invention, an index related to the progression of worsening heart failure is derived based on first biological information related to hemodynamics at a first time point and second biological information related to compensatory mechanisms associated with heart failure at a second time point. The derived index is an index related to the worsening of heart failure at an evaluation time point that is later than the first and second time points. When a patient's heart failure worsens, hemodynamics changes, and the first biological information changes. When a patient's heart failure worsens, compensatory mechanisms are activated, and the second biological information changes. Because the index is derived based on both two types of biological information, the first biological information and the second biological information, variability in the index is reduced, and a more reliable index can be obtained.

[0027] According to the present invention, by using a more reliable index, it becomes possible to more accurately evaluate the progression of a patient's heart failure.

[0028] 1 is a schematic diagram showing an example of the configuration of an information processing system for evaluating the worsening of heart failure in a patient. FIG. 1 is a block diagram showing an example of the internal configuration of a wearable device. FIG. 1 is a block diagram showing an example of the internal configuration of an information processing device. FIG. 2 is a conceptual diagram showing an example of the functions of a trained model. FIG. 2 is a block diagram showing an example of the internal configuration of a terminal device. FIG. 3 is a flowchart showing an example of the processing procedure for generating first biological information and second biological information performed by an information processing system. FIG. 4 is a flowchart showing an example of the processing procedure for deriving an index related to the worsening of heart failure performed by an information processing device. FIG. 5 is a conceptual diagram showing an example of the functions of a trained model using values ​​of multiple types of biological information. FIG. 6 is a diagram showing a first example of a graph representing the time changes of first biological information, second biological information, and estimated BNP values. FIG. 7 is a diagram showing a second example of a graph representing the time changes of first biological information, second biological information, and estimated BNP values. FIG. 8 is a diagram showing a third example of a graph representing the time changes of first biological information, second biological information, and estimated BNP values. FIG. 9 is a block diagram showing an example of the internal functional configuration of a learning device. FIG. 10 is a flowchart showing an example of the processing procedure executed by the learning device. FIG. 11 is a conceptual diagram showing an example of the functions of a trained model using activity amount information.

[0029] The present invention will be described in detail below with reference to the drawings illustrating embodiments thereof. FIG. 1 is a schematic diagram showing an example configuration of an information processing system 100 for evaluating the worsening of heart failure in a patient 5. The information processing system 100 acquires biological information representing the physiological state of the patient 5 and executes an information processing method for deriving an index related to the worsening of heart failure. The information processing system 100 includes a wearable device 2, an information processing device 1, and a terminal device 3. The wearable device 2 is worn on the body of the patient 5. For example, the wearable device 2 is worn on the arm of the patient 5. The wearable device 2 has a sensor that measures data related to the body of the patient 5. The wearable device 2 transmits measurement data to the information processing device 1 via a communication network 4 such as the Internet.

[0030] The information processing device 1 receives data transmitted from the wearable device 2 and generates biometric information representing the physiological state of the patient 5 based on the received data. The information processing device 1 derives an index related to the worsening of heart failure of the patient 5 based on the biometric information. The information processing device 1 communicates with the terminal device 3 via a communication network 4. The information processing device 1 outputs the index related to the worsening of heart failure by displaying it on the terminal device 3. The index displayed by the terminal device 3 is checked by a medical professional 6 such as a doctor, who evaluates the worsening of heart failure of the patient 5. The information processing system 100 may include multiple wearable devices 2 or multiple terminal devices 3.

[0031] 2 is a block diagram showing an example of the internal configuration of the wearable device 2. The wearable device 2 includes a control unit 21, a first sensor 22, a second sensor 23, and a communication unit 24. The first sensor 22 is a sensor that measures data related to hemodynamics of the body of the patient 5. Hemodynamics refers to the state of blood flow in the body. The data related to hemodynamics is data necessary to generate first biological information related to hemodynamics. The first biological information includes pulse rate, pulse variability, blood pressure, respiratory rate, etc. If the patient 5's heart failure worsens, the hemodynamics changes and the first biological information changes. Therefore, it is possible to evaluate the worsening of heart failure based on the first biological information.

[0032] The first sensor 22 is, for example, a PPG (Photoplethysmography) sensor. A PPG sensor irradiates the skin with light and measures changes in the intensity of light reflected from blood vessels beneath the skin. The intensity of light reflected from the blood vessels changes depending on the blood flow rate. That is, the first sensor 22 measures the time change in blood flow rate, specifically, the pulse wave, as data related to hemodynamics. A PPG sensor is preferable as the first sensor 22 because it can continuously acquire data related to hemodynamics while worn on the arm. However, the first sensor 22 may also be a sensor other than a PPG sensor that can acquire data related to hemodynamics, such as an electrocardiogram sensor or a blood pressure sensor.

[0033] The second sensor 23 is a sensor that measures data related to compensatory mechanisms associated with heart failure in the body of the patient 5. A compensatory mechanism is when a function of one part of the body is reduced and another part compensates for that function. An example of a compensatory mechanism associated with heart failure is an increase in the amount of water in the body due to the kidneys in response to a decrease in blood flow caused by heart failure. The data related to the compensatory mechanisms associated with heart failure is data necessary for generating second biological information related to the compensatory mechanisms associated with heart failure. The second biological information is, for example, fat-free mass, or information related to the amount of water in the body, such as extracellular fluid volume, extracellular water change rate, or extracellular water ratio. If the heart failure of the patient 5 worsens, the compensatory mechanisms are activated and the second biological information changes. Therefore, it is possible to evaluate the worsening of heart failure based on the biological information related to the compensatory mechanisms.

[0034] The second sensor 23 is, for example, a bioimpedance sensor. The bioimpedance sensor passes a small current through the body and measures the body's impedance (bioimpedance). When the patient's 5 heart failure worsens, a compensatory mechanism that increases the amount of water in the body is activated, causing a change in bioimpedance. Therefore, bioimpedance is data related to the compensatory mechanism associated with heart failure. In other words, the second sensor 23 measures bioimpedance as data related to the compensatory mechanism associated with heart failure. The second sensor 23 may also be a sensor that measures data other than bioimpedance.

[0035] The communication unit 24 transmits measurement data including data measured by the first sensor 22 and data measured by the second sensor 23 to the information processing device 1. The communication unit 24 transmits the measurement data, for example, by wireless communication. For example, the measurement data is transmitted to the information processing device 1 via a communication network (not shown) such as the Internet. The control unit 21 causes the communication unit 24 to transmit the measurement data to the information processing device 1. That is, the wearable device 2 transmits the measurement data to the information processing device 1.

[0036] FIG. 3 is a block diagram showing an example of the internal configuration of the information processing device 1. The information processing device 1 is configured using a computer such as a server device. The information processing device 1 includes a calculation unit 11, a memory 12 that stores temporary data generated during calculations, a storage unit 13, a reading unit 14, and a communication unit 15. The calculation unit 11 is, for example, a processor and is configured using a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), or a multi-core CPU. The calculation unit 11 may also be configured using a quantum computer. The memory 12 is, for example, a RAM. The storage unit 13 is nonvolatile and is, for example, a hard disk or nonvolatile semiconductor memory. The reading unit 14 reads information from a recording medium 10 such as an optical disk or a portable memory. The communication unit 15 communicates with the outside of the information processing device 1. Specifically, the communication unit 15 communicates with the wearable device 2 and the terminal device 3 via a communication network (not shown).

[0037] The calculation unit 11 causes the reading unit 14 to read a computer program (program product) 131 recorded on the recording medium 10, and stores the read computer program 131 in the storage unit 13. The calculation unit 11 executes processing to realize the functions of the information processing device 1 in accordance with the computer program 131. The computer program 131 may be stored in the storage unit 13 in advance, or may be downloaded from outside the information processing device 1. In this case, the information processing device 1 does not need to include the reading unit 14.

[0038] The computer program 131 can be deployed to run on a single computer, or on multiple computers located at one site or distributed across multiple sites and interconnected by a communications network. That is, the information processing device 1 may be configured with multiple computers, and the computer program 131 may be executed on multiple computers connected via a communications network. The information processing device 1 may be configured using a cloud server.

[0039] The processing of each step described below for executing the information processing method can be executed by multiple computers. The processing of each step can also be executed by different computers. The processing of each step can also be executed using a virtual machine. The processing of each step may be executed by multiple calculation units. The processing of each step may be executed by different calculation units. Data referenced during processing may be stored in multiple computers. For example, data referenced in processing executed on one computer may be stored on another computer.

[0040] The information processing device 1 receives the measurement data transmitted from the wearable device 2 via the communication unit 15. The information processing device 1 generates first biological information related to hemodynamics based on data related to the hemodynamics of the patient 5 included in the measurement data. The first biological information is information representing a physiological condition related to hemodynamics. The first biological information includes, for example, a pulse waveform, pulse rate, pulse variability, blood pressure, respiratory rate, stroke volume, oxygen saturation, or vascular resistance. The information processing device 1 also generates second biological information related to compensatory mechanisms based on data related to compensatory mechanisms associated with heart failure included in the measurement data. The second biological information is information representing a physiological condition related to compensatory mechanisms associated with heart failure. The second biological information includes, for example, fat-free mass, or information related to the amount of water in the body, such as extracellular fluid volume, extracellular water change rate, or extracellular water ratio.

[0041] The storage unit 13 stores a biological information DB (database) 133 that records first biological information and second biological information of the patient 5. The biological information DB 133 records the first biological information and second biological information for each patient 5. Furthermore, the biological information DB 133 records the first biological information in association with the time point at which data related to hemodynamics was obtained, and records the second biological information in association with the time point at which data related to compensatory mechanisms associated with heart failure was obtained.

[0042] The storage unit 13 stores a patient DB 134 that records personal information of the patient 5. The personal information of the patient 5 includes, for example, medical history, whether or not pacing is being used, age, sex, height, weight, NYHA classification, tricuspid regurgitation grade, whether or not chronic atrial fibrillation is present, and baseline blood concentrations of BNP or NT-proBNP (N-terminal proBNP). The personal information may also include the adjustment size of a belt or other attachment that attaches the wearable device 2 to the patient 5's arm.

[0043] The information processing device 1 includes a trained model 132 used to derive an index related to the worsening of heart failure based on the first biological information and the second biological information. The trained model 132 is realized by the calculation unit 11 executing processing in accordance with the computer program 131. The storage unit 13 stores data necessary to realize the trained model 132. For example, the trained model 132 is realized using a neural network. The trained model 132 may also be realized using a model other than a neural network, such as a support vector machine.

[0044] The trained model 132 may be configured using hardware. For example, the trained model 132 may be configured by hardware including a processor and a memory that stores necessary programs and data. Alternatively, the trained model 132 may be realized using a quantum computer. Alternatively, the trained model 132 may be provided outside the information processing device 1, and the information processing device 1 may execute processing using the external trained model 132. For example, the trained model 132 may be realized using multiple computers connected via a communication network, or may be realized using the cloud.

[0045] FIG. 4 is a conceptual diagram showing an example of the function of the trained model 132. First biological information, second biological information, and personal information of the patient 5 are input to the trained model 132. The trained model 132 is trained in advance to output an index related to the worsening of heart failure when the first biological information, second biological information, and personal information are input. The first biological information, second biological information, and personal information input to the trained model 132 may each be singular or plural. In this embodiment, deriving an index related to the worsening of heart failure of a patient based on biological information means inputting the biological information into the trained model 132 and outputting an index related to the worsening of heart failure of the patient from the trained model 132.

[0046] The index related to the worsening of heart failure is an index related to the progression of the worsening of heart failure in patient 5, and indicates the possibility that the patient's heart failure is in an acute worsening state or the degree of the worsening of heart failure. In the present embodiment, the index related to the worsening of heart failure is an estimated BNP value. The estimated BNP value is an estimated value of the blood concentration of BNP in patient 5. The blood concentration of BNP is a value obtained by component analysis of the patient's blood, and it is known that the higher the blood concentration of BNP, the higher the possibility that the heart failure is in an acute worsening state or the degree of the worsening of heart failure. Therefore, it is possible to evaluate the worsening of heart failure in patient 5 based on the estimated BNP value. Furthermore, NT-proBNP, which is generated by cleavage of BNP from a BNP precursor, is also known to indicate the possibility that the heart failure is in an acute worsening state or the degree of the worsening of heart failure, similar to BNP. Therefore, the estimated NT-proBNP value can also be used as an index related to the worsening of heart failure. The trained model 132 is trained in advance using training data. A method for generating the trained model 132 will be described later.

[0047] FIG. 5 is a block diagram showing an example of the internal configuration of the terminal device 3. The terminal device 3 is a computer such as a personal computer, a tablet computer, or a smartphone. The user of the terminal device 3 is a medical professional 6 such as a doctor. The terminal device 3 includes a calculation unit 31, a memory 32, a reading unit 33 that reads information from a recording medium 30 such as an optical disc or a portable memory, a storage unit 34, an operation unit 35, a display unit 36, and a communication unit 37. The calculation unit 31 is configured using, for example, a CPU, a GPU, or a multi-core CPU. The calculation unit 31 may also be configured using a quantum computer. The memory 32 stores temporary data generated in conjunction with calculations. The memory 32 is, for example, a RAM. The storage unit 34 is non-volatile, for example, a hard disk or non-volatile semiconductor memory.

[0048] The calculation unit 31 causes the reading unit 33 to read the computer program 341 recorded on the recording medium 30, and stores the read computer program 341 in the storage unit 34. The calculation unit 31 executes processing required for the terminal device 3 in accordance with the computer program 341. The computer program 341 may be stored in the storage unit 34 in advance, or may be downloaded from outside the terminal device 3. In this case, the terminal device 3 does not need to be equipped with the reading unit 33.

[0049] The operation unit 35 receives input of information such as text by receiving operations from the user. The operation unit 35 is, for example, a touch panel. The display unit 36 ​​displays images. The display unit 36 ​​is, for example, a liquid crystal display or an EL display (Electroluminescent Display). The operation unit 35 and the display unit 36 ​​may be integrated. The communication unit 37 communicates with the information processing device 1 via the communication network 4.

[0050] The information processing system 100 performs processing to generate first biological information and second biological information based on data measured by the wearable device 2. Fig. 6 is a flowchart showing an example of the processing procedure for generating first biological information and second biological information performed by the information processing system 100. Hereinafter, steps are abbreviated as S. The calculation unit 11 of the information processing device 1 performs the following processing in accordance with the computer program 131.

[0051] The wearable device 2 measures data related to hemodynamics and data related to compensatory mechanisms associated with heart failure for the body of the patient 5 (S11). In S11, the first sensor 22 measures data related to hemodynamics for the body of the patient 5. The data measured by the first sensor 22 is, for example, a change in blood flow over time. The second sensor 23 measures data related to compensatory mechanisms associated with heart failure for the body of the patient 5. The data measured by the second sensor 23 is, for example, bioimpedance. The control unit 21 causes the communication unit 24 to transmit measurement data including the data measured by the first sensor 22 and the data measured by the second sensor 23 to the information processing device 1. That is, the wearable device 2 transmits the measurement data to the information processing device 1. The information processing device 1 receives the transmitted measurement data via the communication unit 15.

[0052] The information processing device 1 generates first biological information and second biological information based on the measurement data (S12). In S12, the calculation unit 11 generates the first biological information based on data measured by the first sensor 22 included in the measurement data. The data measured by the first sensor 22 is data related to the hemodynamics of the patient 5, such as a change in blood flow over time. The first biological information is biological information related to the hemodynamics of the patient 5, such as a pulse waveform, pulse rate, pulse variability, blood pressure, respiratory rate, stroke volume, oxygen saturation, or vascular resistance. The calculation unit 11 may generate multiple types of first biological information. The calculation unit 11 generates the first biological information by calculating the first biological information based on a single measurement data measured by the first sensor 22 within a certain period (e.g., five minutes). For example, the calculation unit 11 calculates, as pulse variability, the standard deviation of the pulse rate in a single measurement data that is a measurement result for five minutes. For example, the calculation unit 11 performs a spectrum analysis on the pulse waveform in one measurement data set, which is the result of a five-minute measurement, and determines the intensity of a specific frequency band as pulse variability.

[0053] In S12, the calculation unit 11 generates second biological information based on data measured by the second sensor 23 included in the measurement data. The data measured by the second sensor 23 is data related to the compensatory mechanism of the patient 5 due to heart failure, such as bioimpedance. The second biological information is biological information related to the compensatory mechanism of the patient 5 due to heart failure, such as fat-free mass, or information related to the amount of water in the patient 5's body, such as extracellular fluid volume, extracellular water change rate, or extracellular water ratio. The calculation unit 11 may generate multiple types of second biological information. The calculation unit 11 generates the second biological information by calculating the second biological information based on a single measurement data measured by the second sensor 23. For example, if the data measured by the second sensor 23 is bioimpedance, the single measurement data used to generate the second biological information is data obtained by a multi-frequency bioelectrical impedance method, in which measurements are taken while changing the frequency of the current passed through the body, for example, from 2.5 to 350 kHz. For example, the calculation unit 11 calculates the extracellular fluid volume from one measurement data thus obtained.

[0054] The information processing device 1 then calculates moving average values ​​of the first biometric information and the second biometric information within a predetermined time window (S13). The biometric information DB 133 stores first biometric information and second biometric information generated in the past. The calculation unit 11 reads from the biometric information DB 133 a predetermined number of first biometric information and second biometric information most recently generated by processing prior to S12. Alternatively, the calculation unit 11 reads from the biometric information DB 133 first biometric information and second biometric information generated by processing prior to S12 and associated with a time point included in the predetermined time window. The calculation unit 11 calculates the moving average values ​​of the first biometric information and the second biometric information by calculating the average of multiple pieces of first biometric information and multiple pieces of second biometric information using the read first biometric information and the first biometric information and the second biometric information generated in S12. The time window width used to calculate the moving average value of the first biometric information and the moving average value of the second biometric information may be different. The process of calculating the moving average values ​​of the first biometric information and the second biometric information in S13 is executed every time the process of generating the first biometric information and the second biometric information in S12 is executed.

[0055] The process of calculating the moving average values ​​of the first biometric information and the second biometric information may be executed at predetermined time intervals regardless of the timing of the process of generating the first biometric information and the second biometric information. In this case, the information processing device 1 reads the first biometric information and the second biometric information from the biometric information DB 133 at predetermined time intervals (for example, every hour), and calculates the moving average values ​​of the first biometric information and the second biometric information using the read first biometric information and the second biometric information.

[0056] The information processing device 1 stores the first biological information and the second biological information (S14). In S14, the calculation unit 11 records the first biological information and the second biological information generated in S12 and the moving average values ​​of the first biological information and the second biological information calculated in S13 in the biological information DB 133. At this time, the calculation unit 11 records the first biological information and the second biological information in the biological information DB 133 in association with the time point when the hemodynamics of the patient 5 was observed and the time point when the compensatory mechanism of the patient 5 associated with heart failure was observed. For example, the information indicating the time point is information indicating the year, month, day, and time. The generated first biological information and the second biological information and the moving average values ​​of the first biological information and the second biological information are recorded in the biological information DB 133 in association with the information indicating the time point.

[0057] The calculation unit 11 associates pulse rate variability, which is first biological information generated based on a single measurement data set in S12, with the time (date and time) when the single measurement data was measured and records the associated pulse rate variability in the biological information DB 133. The calculation unit 11 also associates extracellular water volume, which is second biological information generated based on a single measurement data set in S12, with the time (date and time) when the single measurement data was measured and records the extracellular water volume in the biological information DB 133. The extracellular water volume is an amount equivalent to extracellular fluid volume. The calculation unit 11 may calculate an extracellular water ratio, which is the ratio of extracellular fluid volume to total body water volume, as second biological information from the single measurement data in S12, and record the ratio in the biological information DB 133 with the time (date and time) when the single measurement data was measured. The calculation unit 11 may also calculate an extracellular water change rate, which is second biological information from the single measurement data in S12, and record the ratio in the biological information DB 133 with the time (date and time) when the single measurement data was measured. The extracellular water change rate is the ratio of the extracellular water volume calculated from one measurement to the reference value of the extracellular water volume.

[0058] For example, the control unit 21 of the wearable device 2 identifies the time points when the first sensor 22 measured data and the time points when the second sensor 23 measured data, and generates information indicating the time points when each piece of data was measured. The wearable device 2 transmits measurement data including the data measured by the first sensor 22 and the data measured by the second sensor 23, and information indicating the time points when each piece of data was measured, to the information processing device 1. In S14, the calculation unit 11 records the generated first biological information and the moving average value of the first biological information in the biological information DB 133, associating them with the time points when the first sensor 22 measured the data. In addition, the calculation unit 11 records the generated second biological information and the moving average value of the second biological information in the biological information DB 133, associating them with the time points when the second sensor 23 measured the data.

[0059] Specifically, the calculation unit 11 acquires the measurement time (date and time) at which a single measurement data set was recorded in the biometric information DB 133 and associated with each of the multiple pieces of first biometric information used to calculate the moving average value. The calculation unit 11 identifies a time point located at a predetermined time position in the time window width used to calculate the moving average value, for example, a time point located at the center time position in the time window width, from the multiple measurement time points (date and time) corresponding to the multiple pieces of acquired first biometric information, and records the generated moving average value of the first biometric information in association with the identified time point in the biometric information DB 133. Similarly, the calculation unit 11 acquires the measurement time (date and time) at which a single measurement data set was recorded in the biometric information DB 133 and associated with each of the multiple pieces of second biometric information used to calculate the moving average value. The calculation unit 11 identifies a time point located at a predetermined time position in the time window width used to calculate the moving average value, for example, a time point located at the center time position in the time window width, from the multiple measurement time points (date and time) corresponding to the multiple pieces of acquired second biometric information, and records the generated moving average value of the second biometric information in association with the identified time point in the biometric information DB 133. Furthermore, the calculation unit 11 may associate the calculated moving average value of the first biological information with the latest measurement time point (date and time) among a plurality of measurement time points (date and time) corresponding to the plurality of pieces of acquired first biological information, and record the same in the biological information DB 133. Similarly, the calculation unit 11 may associate the calculated moving average value of the second biological information with the latest measurement time point (date and time) among a plurality of measurement time points (date and time) corresponding to the plurality of pieces of acquired second biological information, and record the same in the biological information DB 133.

[0060] Alternatively, the information processing device 1 may use the time when the measurement data was received in S11, the time when the first biological information and the second biological information were generated in S12, or the time when the moving average value was calculated in S13. The calculation unit 11 identifies these time points and records the generated first biological information and second biological information and the moving average values ​​of the first biological information and second biological information in association with information indicating the identified time points in the biological information DB 133. These time points are all based on the time points when the measurement data was measured by the first sensor 22 and the second sensor 23 used to generate the first biological information and the second biological information.

[0061] Hereinafter, the first biological information and the second biological information associated with a certain time point will be referred to as the first biological information and the second biological information at that time point. The first biological information at a certain time point is the first biological information related to hemodynamics at that time point. The second biological information at a certain time point is the second biological information related to compensatory mechanisms at that time point.

[0062] After S14 is completed, the information processing device 1 ends the process of generating the first biological information and the second biological information. The information processing system 100 intermittently or continuously repeats the processes of S11 to S14. That is, data related to the hemodynamics of the patient 5 and data related to compensatory mechanisms associated with heart failure are intermittently or continuously measured, and the first biological information and the second biological information are generated and stored, and moving average values ​​of the first biological information and the second biological information are calculated and stored intermittently or continuously. As a result, a time series of the first biological information and the second biological information and a time series of the moving average values ​​of the first biological information and the second biological information are recorded.

[0063] The information processing device 1 performs a process of deriving an index related to worsening heart failure based on the first biological information and the second biological information. FIG. 7 is a flowchart showing an example of the procedure of the process of deriving an index related to worsening heart failure performed by the information processing device 1. The information processing device 1 acquires the first biological information and the second biological information (S21). In S21, the calculation unit 11 acquires the first biological information and the second biological information by reading the first biological information and the second biological information from the biological information DB 133. At this time, the calculation unit 11 acquires moving average values ​​of the first biological information and the second biological information. The first biological information and the second biological information obtained from the patient 5 engaged in daily activities fluctuate due to various factors, such as changes in the patient 5's movements or environment. By using the moving average values ​​as the values ​​of the first biological information and the second biological information to derive the index related to worsening heart failure, the influence of fluctuations in the first biological information and the second biological information on the index is suppressed.

[0064] In order to prevent fluctuations in the first biological information and the second biological information from affecting the index, the values ​​of the first biological information and the second biological information may be values ​​obtained by convolution processing of data related to hemodynamics measured by the first sensor 22 and data related to compensatory mechanisms associated with heart failure measured by the second sensor 23. Alternatively, the values ​​of the first biological information and the second biological information used to derive the index related to the worsening of heart failure may be moving medians of the first biological information and the second biological information over a predetermined time window width. The moving medians are calculated as appropriate, and, similar to the moving average values, a predetermined time position within the time window width used to calculate each moving median, such as the central time position within the time window width, or the most recent measurement time point is identified. Similar to the moving average values, the moving medians are associated with the identified time points and recorded in the biological information DB 133.

[0065] The information processing device 1 derives an index related to worsening heart failure at a certain time point. This time point is referred to as the evaluation time point. For example, the time point at which the information processing device 1 performs a process to derive an index related to worsening heart failure is referred to as the evaluation time point. In S21, the calculation unit 11 acquires first biological information at a first time point that is a specific period of time before the evaluation time point, and acquires second biological information at a second time point that is a specific period of time before the evaluation time point and different from the first time point.

[0066] Specifically, the calculation unit 11 obtains the moving average value or moving median value of the first biometric information at a first time point that is a specific period before the evaluation time point, based on the time point (date and time) associated with each value from among the moving average values ​​or moving median values ​​of the first biometric information recorded in the biometric information DB 133. Also, the calculation unit 11 obtains the moving average value or moving median value of the second biometric information at a second time point that is a specific period before the evaluation time point and different from the first time point, based on the time point (date and time) associated with each value from among the moving average values ​​or moving median values ​​of the second biometric information recorded in the biometric information DB 133.

[0067] When the patient's heart failure worsens, hemodynamics changes, and the first biological information related to hemodynamics changes. However, in response to the hemodynamic changes, hormones such as BNP are secreted in the body, and after a certain amount of time, the hemodynamics returns to normal due to the action of the hormones. Therefore, the first biological information at a first time point, which is earlier than the evaluation time point, may be more significantly affected by the state of heart failure than the first biological information at the evaluation time point. By using the first biological information at the first time point, an index accurately reflecting the state of heart failure can be derived. Such changes in hemodynamics due to the action of hormones associated with the worsening of heart failure are gradual. Therefore, the first time point is a time point that occurs a specific period of time that is more than one hour before the evaluation time point. Note that NT-proBNP, which does not have hormonal function (biological activity), is also produced simultaneously with BNP by cleavage of BNP precursors. Therefore, even when an estimated NT-proBNP value is used as an index related to the worsening of heart failure, the first time point is a time point that occurs a specific period of time that is more than one hour before the evaluation time point.

[0068] Compensatory mechanisms associated with heart failure also return to normal after a certain period of time due to hormones. Therefore, the second biological information at a time point earlier than the evaluation time point may be more significantly affected by the state of heart failure than the second biological information at the evaluation time point. However, changes in compensatory mechanisms due to hormones occur more slowly than changes in hemodynamics. Therefore, the second biological information at a second time point earlier than the first time point may be most significantly affected by the state of heart failure. By using the first biological information at the second time point, an index accurately reflecting the state of heart failure can be derived. Such changes in compensatory mechanisms due to the action of hormones, associated with the worsening of heart failure, are more gradual than changes in hemodynamics. Therefore, the second time point is a time point that is a specific period of time that is more than one hour before the evaluation time point and that is also a specific period of time that is more than one hour before the first time point. In addition, based on the relationship between BNP and NT-proBNP described above, even when the estimated NT-proBNP value is used as an index for the worsening of heart failure, the second time point is a time point that occurs a specific period that is more than one hour before the evaluation time point, and is also a time point that occurs a specific period that is more than another hour before the first time point.

[0069] In S21, the information processing device 1 may acquire the first biological information and the second biological information by a method other than reading them from the biological information DB 133. For example, the information processing device 1 may acquire the first biological information and the second biological information by generating the first biological information and the second biological information from measurement data. For example, a medical professional 6 such as a doctor may operate the operation unit 35 of the terminal device 3, the first biological information and the second biological information may be transmitted from the terminal device 3 to the information processing device 1, and the information processing device 1 may acquire the first biological information and the second biological information by receiving the first biological information and the second biological information. For example, the medical professional 6 may operate the operation unit 35 of the terminal device 3, specify any of the first biological information and the second biological information from the biological information DB 133, and the information processing device 1 may acquire the first biological information and the second biological information by receiving the specified first biological information and the second biological information.

[0070] The information processing device 1 acquires the personal information of the patient 5 (S22). In S22, the calculation unit 11 acquires the personal information of the patient 5 by reading the personal information of the patient 5 from the patient DB 134. The information processing device 1 may acquire the personal information of the patient 5 by other methods. For example, the personal information of the patient 5 may be transmitted from the terminal device 3 to the information processing device 1, and the information processing device 1 may receive the personal information, thereby acquiring the personal information of the patient 5.

[0071] The information processing device 1 inputs the acquired first biological information, second biological information, and personal information to the trained model 132 (S23). In S23, the calculation unit 11 inputs the first biological information, second biological information, and personal information to the trained model 132 and causes the trained model 132 to execute processing. In response to the input of the first biological information, second biological information, and personal information, the trained model 132 outputs an estimated BNP value as an index related to the worsening of heart failure.

[0072] The information processing device 1 acquires an estimated BNP value as an index related to the worsening of heart failure (S24). In S24, the calculation unit 11 acquires the estimated BNP value output by the trained model 132 and stores the acquired estimated BNP value in the storage unit 13. At this time, the estimated BNP value is stored in association with the evaluation time point. For example, the calculation unit 11 stores the estimated BNP value in association with information indicating the evaluation time point, such as the date and time, in the storage unit 13. In this way, the calculation unit 11 uses the trained model 132 to derive the estimated BNP value, which is an index related to the worsening of heart failure at the evaluation time point, based on the moving average or moving median of the acquired first biological information at the first time point and the moving average or moving median of the acquired second biological information at the second time point. The calculation unit 11 stores the derived estimated BNP value in association with the evaluation time point (date and time) in the storage unit 13.

[0073] In S21 to S24, the calculation unit 11 may derive an estimated BNP value, which is an index related to the worsening of heart failure, using a plurality of first biological information values ​​and a plurality of second biological information values ​​within a predetermined time window. In this case, the calculation unit 11 acquires, from the plurality of first biological information values ​​recorded in the biological information DB 133, a plurality of first biological information values ​​that are at a first time point that is a specific period before the evaluation time point and that is within the predetermined time window, based on the time point (date and time) associated with each value. Furthermore, from the plurality of second biological information values ​​recorded in the biological information DB 133, the calculation unit 11 acquires, from the plurality of second biological information values ​​recorded in the biological information DB 133, a plurality of second biological information values ​​that are at a second time point that is a specific period before the evaluation time point, different from the first time point, and that is within the predetermined time window, based on the time point (date and time) associated with each value.

[0074] The calculation unit 11 then derives an estimated BNP value, which is an index related to the worsening of heart failure at the time of evaluation, from the values ​​of the multiple first biological information and the values ​​of the multiple second biological information. To derive such an index, for example, a trained model is used, which utilizes a convolutional neural network to train the model to derive an index related to the worsening of heart failure at the time of evaluation from the values ​​of the multiple first biological information at a first time point that is a specific period of time before the time of evaluation and within a specific time window, and the values ​​of the multiple second biological information at a second time point that is a specific period of time before the time of evaluation, different from the first time point, and within a specific time window. The calculation unit 11 inputs the values ​​of the multiple first biological information, the values ​​of the multiple second biological information, and personal information of the patient 5 into the trained model, causes the trained model to execute processing, and obtains the estimated BNP value output by the trained model.

[0075] In S23 and S24, the calculation unit 11 may derive an estimated BNP value, which is an index related to the worsening of heart failure, using values ​​of multiple types of biological information. FIG. 8 is a conceptual diagram showing an example of the function of the trained model 132 using values ​​of multiple types of biological information. Multiple types of first biological information, multiple types of second biological information, and personal information of the patient 5 are input to the trained model 132. FIG. 8 shows pulse variability, pulse rate, blood pressure, and respiratory rate as examples of multiple types of first biological information. The first time points for each type of first biological information may be different from each other. Other types of first biological information may be used, and the number of types of first biological information may be two, three, or more than four. FIG. 8 shows extracellular water volume and extracellular water change rate as examples of multiple types of second biological information. The second time points for each type of second biological information may be different from each other.

[0076] The trained model 132 is trained in advance to output an index (estimated BNP value) related to the worsening of heart failure when multiple types of first biological information, multiple types of second biological information, and personal information are input. The calculation unit 11 inputs multiple types of first biological information, multiple types of second biological information, and personal information of the patient 5 to the trained model 132, causes the trained model 132 to execute processing, and acquires the estimated BNP value output by the trained model 132. In S23 to S24, the calculation unit 11 may use the value of one type of second biological information. In this case, the trained model 132 is trained in advance to output an index related to the worsening of heart failure when multiple types of first biological information, one type of second biological information, and personal information are input. Furthermore, in S23 to S24, the calculation unit 11 may use the value of one type of first biological information. In this case, the trained model 132 is pre-trained to output an index related to the worsening of heart failure when one type of first biometric information, multiple types of second biometric information, and personal information are input.

[0077] The information processing device 1 calculates a representative value of the estimated BNP value within a predetermined period of time (S25). In S25, the calculation unit 11 uses the plurality of estimated BNP values ​​stored in the storage unit 13 to calculate a representative value of the plurality of estimated BNP values ​​associated with the plurality of time points included in the predetermined period of time. For example, the calculation unit 11 calculates the median or average of the plurality of estimated BNP values ​​as the representative value. In S25, the calculation unit 11 calculates a representative value of the estimated BNP value within a specific period of time, such as one day or one hour. The calculation unit 11 may also calculate a representative value of the estimated BNP value within the most recent predetermined period of time. The calculation unit 11 may also calculate representative values ​​of the estimated BNP value within each of multiple periods of time, such as a representative value of the estimated BNP value within one day and a representative value of the estimated BNP value within one hour. The calculation unit 11 stores the representative value of the estimated BNP value in the storage unit 13 in association with information indicating the period.

[0078] In S25, the information processing device 1 may calculate representative values ​​of the first biological information and the second biological information, similar to the representative value of the estimated BNP value. The representative value is an average or median of the first biological information and the second biological information within a specific period. The calculation unit 11 calculates the representative values ​​of the first biological information and the second biological information within the specific period based on a plurality of pieces of first biological information and second biological information recorded in the biological information DB 133 in association with a plurality of time points included in the specific period. The calculation unit 11 stores the representative values ​​of the first biological information and the second biological information in the storage unit 13 in association with information indicating the period.

[0079] The information processing device 1 outputs the first biological information, the second biological information, and the estimated BNP value (S26). In S26, the calculation unit 11 transmits the latest estimated BNP value, the representative value of the latest estimated BNP value, and the first biological information and second biological information used to derive the latest estimated BNP value from the communication unit 15 to the terminal device 3. The terminal device 3 receives the estimated BNP value, the representative value of the estimated BNP value, the first biological information, and the second biological information via the communication unit 37 and displays them on the display unit 36. In this way, the first biological information, the second biological information, and the estimated BNP value are output.

[0080] The first biological information, the second biological information, and the estimated BNP value output using the terminal device 3 are confirmed by a medical professional 6 such as a doctor. The medical professional 6 can evaluate the worsening of heart failure in the patient 5 based on the confirmed estimated BNP value. For example, the medical professional 6 can determine the degree of worsening of heart failure based on the estimated BNP value. Furthermore, the medical professional 6 can determine the need for treatment, the content of the treatment, the content of accurate guidance to be given to the patient 5, etc., depending on the degree of worsening of heart failure.

[0081] The medical professional 6 can check the representative estimated BNP value and evaluate the state of heart failure of the patient 5 based on the representative estimated BNP value. By using the representative estimated BNP value, it is possible to suppress variations in the estimated BNP value due to factors such as the movements of the patient 5 and evaluate a representative state of heart failure over a predetermined period of time. In addition, the medical professional 6 can check the first biological information and the second biological information and evaluate the physiological state of the patient 5 based on the first biological information and the second biological information.

[0082] In S26, the information processing device 1 may output a graph showing changes over time in the first biological information, the second biological information, and the estimated BNP value. The calculation unit 11 generates a graph showing changes over time in the first biological information, the second biological information, and the estimated BNP value based on the multiple estimated BNP values ​​and multiple representative values ​​stored in the storage unit 13, and the multiple pieces of first biological information and second biological information recorded in the biological information DB 133. The calculation unit 11 transmits the generated graph from the communication unit 15 to the terminal device 3, and the terminal device 3 receives the graph via the communication unit 37. The terminal device 3 displays the graph on the display unit 36.

[0083] FIG. 9 is a diagram showing a first example of graphs representing time changes in first biological information, second biological information, and estimated BNP values. The first graph from the top represents time changes in estimated BNP values, with the horizontal axis representing time and the vertical axis representing estimated BNP values. The units of estimated BNP values ​​are pg / mL. The second graph from the top represents time changes in extracellular water volume, which is second biological information, with the horizontal axis representing time and the vertical axis representing the value of extracellular water volume. The units of extracellular water volume are kg. The third graph from the top represents time changes in pulse rate variability, which is first biological information, with the horizontal axis representing time and the vertical axis representing the value of pulse rate variability. The time indicated by the horizontal axes of the three graphs is the same. Personal information such as the name, age, and gender of patient 5 is attached to the graphs.

[0084] The first graph from the top shows the time change of the estimated BNP value in a box-and-whisker plot. Each box and whisker represents the time change of the estimated BNP value over a day. Each box and whisker represents the maximum, third quartile, median, first quartile, and minimum of the estimated BNP value over a day. Since FIG. 9 contains 14 boxes and whiskers, it shows the time change of the estimated BNP value over 14 days. Furthermore, FIG. 9 shows the time change of the moving average value of the extracellular water content, which is the second biological information, and the moving average value of the pulse rate variability, which is the first biological information. The moving average value displayed on the graph is a different value from the moving average value used to derive the estimated BNP value, and is a moving average value over multiple days.

[0085] 10 is a diagram showing a second example of a graph showing the time changes of the first biological information, the second biological information, and the estimated BNP value. As in FIG. 9, the first graph from the top shows the time changes of the estimated BNP value, the second graph shows the time changes of the extracellular water content, which is the second biological information, and the third graph shows the time changes of the pulse rate variability, which is the first biological information. The intervals between vertical lines in the diagram indicate one day. The graph showing the time changes of the estimated BNP value is a line graph showing the representative value of the estimated BNP value for each day and connecting the representative values ​​of the estimated BNP value with a line. The graph showing the time changes of the extracellular water content is a line graph showing the representative value of the extracellular water content for each day and connecting the representative values ​​of the extracellular water content with a line. The graph showing the time changes of the pulse rate variability is a line graph showing the representative value of the pulse rate variability for each day and connecting the representative values ​​of the pulse rate variability with a line. For example, the representative value is the average or median value for one day. The horizontal axes of the three graphs are aligned to the same time.

[0086] FIG. 11 is a diagram showing a third example of a graph showing temporal changes in first biological information, second biological information, and estimated BNP values. The graph shown in FIG. 11 is similar to the graph shown in FIG. 9. Furthermore, in the graph shown in FIG. 11, when an estimated BNP value on the graph is selected by placing a cursor over a point on the graph showing temporal changes in estimated BNP values, the selected estimated BNP value and the evaluation time point associated with the estimated BNP value are displayed. YYYY / MM / DD shown in FIG. 11 is information expressing the evaluation time point in year, month, and day. XX shown in FIG. 11 is the estimated BNP value. Furthermore, in the graph showing temporal changes in the second biological information, the range of time points associated with the second biological information used to derive the estimated BNP value is highlighted. In the graph showing temporal changes in the first biological information, the range of time points associated with the first biological information used to derive the estimated BNP value is highlighted. If a moving average or moving median of the first and second biological information is used to derive the estimated BNP value, the range of time points included within the time window used to calculate the moving average or moving median is highlighted.

[0087] The user of the terminal device 3 operates the operation unit 35, whereby an instruction to adjust the cursor position is input to the terminal device 3, and the calculation unit 31 adjusts the cursor position according to the input instruction. An estimated BNP value on the graph is selected according to the cursor position, and the calculation unit 31 displays a pop-up image including the selected estimated BNP value and the evaluation time point on the display unit 36, superimposed on the graph. Furthermore, the calculation unit 31 performs a process of highlighting on the graph the range of time points associated with the first biological information and the second biological information used to derive the estimated BNP value according to the selected estimated BNP value. At this time, the calculation unit 31 may perform a process of requesting the information processing device 1 for information indicating the selected estimated BNP value and evaluation time point, and the range of time points associated with the first biological information and the second biological information used to derive the estimated BNP value. Furthermore, the information processing device 1 may transmit the information requested by the terminal device 3 to the terminal device 3.

[0088] The estimated BNP value corresponding to the cursor position is selected, and the calculation unit 31 displays the selected estimated BNP value. When the cursor is positioned over one of the boxes and whiskers on the graph, a representative value of the estimated BNP value for one day is selected, and the calculation unit 31 may display the representative value of the estimated BNP value for one day. The representative value is, for example, the median value for one day. Even when a line graph such as that shown in FIG. 10 is displayed, the calculation unit 31 may perform processing to display the estimated BNP value according to the cursor position. By displaying the estimated BNP value and the evaluation time point, the user can confirm the estimated BNP value at each evaluation time point. Furthermore, by highlighting the range of time points associated with the first biological information and the second biological information, the user can confirm when the first biological information and the second biological information used to derive the estimated BNP value were obtained. Note that the operation of selecting an estimated BNP value on the graph may also be the operation of positioning the cursor on a scale mark on the horizontal axis shown on the graph (e.g., a scale mark indicating one day). In addition, if the graph of estimated BNP values ​​shows the representative value of the estimated BNP value for one day or the box and whiskers for each day, the estimated BNP value on the graph may be selected by placing the cursor over the range divided into each day on the graph.

[0089] 9 to 11, the time indicated on the horizontal axis of each of the graphs showing the time changes of the output first biological information, second biological information, and estimated BNP value is consistent. The time indicated on the horizontal axis of each graph is based on the evaluation time point, not the first or second time point. That is, the graphs show the estimated BNP value based on the evaluation time point, and also show the first biological information and second biological information based on the evaluation time point.

[0090] The medical professional 6 checks the graphed first biological information, second biological information, and time-dependent changes in the estimated BNP value. The medical professional 6 can evaluate the state of heart failure of the patient 5 based on the time-dependent changes in the estimated BNP value. For example, the medical professional 6 can determine whether the heart failure is worsening or whether the heart failure state is stable based on the time-dependent changes in the estimated BNP value. The medical professional 6 can also check the time-dependent changes in the first biological information and second biological information, and evaluate the physiological state of the patient 5 based on the time-dependent changes in the first biological information and second biological information.

[0091] After S26 is completed, the information processing device 1 ends the process of deriving an index related to the worsening of heart failure. The information processing device 1 executes the processes of S21 to S26 as needed. For example, the information processing system 100 intermittently or continuously repeats the processes of S11 to S14, and the information processing device 1 intermittently or continuously executes the processes of S21 to S26. That is, the information processing device 1 may intermittently or continuously acquire the first biological information and the second biological information and intermittently or continuously derive the estimated BNP value. This allows the time changes of the first biological information, the second biological information, and the estimated BNP value to be obtained, and enables the evaluation of the worsening of heart failure based on the time changes of the first biological information, the second biological information, and the estimated BNP value.

[0092] The processes of S21 to S25 and S26 may be executed separately. For example, the processes of S21 to S25 may be executed intermittently or continuously, and S26 may be executed at different times. For example, when the calculated estimated BNP value exceeds a predetermined threshold, the information processing device 1 may output the estimated BNP value to warn of worsening heart failure. Alternatively, the information processing device 1 may output the estimated BNP value in response to an instruction from the medical professional 6. For example, the medical professional 6 operates the operation unit 35 of the terminal device 3 to specify one of the patients 5, and the information processing device 1 outputs the estimated BNP value of the specified patient. The medical professional 6 can check the estimated BNP value of any patient 5 and evaluate the worsening of heart failure.

[0093] In the processes of S21 to S26, the information processing device 1 derives an estimated BNP value based on the first biological information, the second biological information, and the personal information of the patient 5. The degree of exacerbation of heart failure may vary depending on the personal information of the patient 5, such as medical history, whether or not pacing is being performed, age, gender, or baseline blood BNP concentration value. Therefore, by using the personal information of the patient 5, an appropriate estimated BNP value according to the patient 5's condition can be obtained, enabling an appropriate assessment of the exacerbation of heart failure. Note that the information processing device 1 may be configured to perform a process of deriving an estimated BNP value based on the first biological information and the second biological information without using the personal information of the patient 5. In this configuration, the trained model 132 is trained in advance to output an estimated BNP value when the first biological information and the second biological information are input. In this configuration, the information processing device 1 may omit the process of S22.

[0094] In the processing of S21, the information processing device 1 may adjust the first time point or the second time point depending on the patient 5. For example, in a patient 5 undergoing pacing, the pulse rate is stable, so hormone-induced changes in hemodynamics and compensatory mechanisms may occur more slowly. Therefore, the information processing device 1 determines whether or not pacing is being performed based on the personal information of the patient 5, and sets the first time point and the second time point to earlier time points for a patient 5 undergoing pacing. By using the first biological information and the second biological information from earlier times, a more accurate estimated BNP value can be obtained.

[0095] The information processing device 1 may adjust the first time point or the second time point depending on the value of the first biological information or the second biological information. For example, if the pulse rate included in the first biological information is smaller than a predetermined threshold, the information processing device 1 may set the first time point and the second time point to earlier time points. The information processing device 1 may adjust the first time point or the second time point depending on an instruction from the medical professional 6. For example, the medical professional 6 operates the operation unit 35 of the terminal device 3 to specify the first time point or the second time point, and the information processing device 1 acquires the first biological information at the specified first time point or the second biological information at the specified second time point. By these methods, the first time point or the second time point is adjusted depending on the patient 5, and a more accurate estimated BNP value depending on the condition of the patient 5 can be obtained.

[0096] In the description of the processes of S21 to S24, an example has been shown in which the time point at which the information processing device 1 performs the process of deriving an estimated BNP value is set as the evaluation time point, but other time points may also be set as the evaluation time point. The information processing device 1 may set a time point earlier than the time point at which the process of deriving an estimated BNP value is performed as the evaluation time point and perform the process of deriving an estimated BNP value at the evaluation time point. The information processing device 1 may set a time point in the future than the time point at which the process of deriving an estimated BNP value is performed as the evaluation time point and perform the process of deriving an estimated BNP value at the evaluation time point. For example, the information processing device 1 may set the current time point as the first time point and perform the process of deriving an estimated BNP value in the future.

[0097] Next, a method for generating the trained model 132 will be described. The trained model 132 is generated by learning. Learning of the trained model 132 is performed by a learning device 7. FIG. 12 is a block diagram showing an example of the internal functional configuration of the learning device 7. The learning device 7 is a computer such as a server device or a personal computer. The learning device 7 includes a calculation unit 71, a memory 72, a reading unit 73, a storage unit 74, an operation unit 75, a display unit 76, and an interface unit 77. The calculation unit 71 is configured using, for example, a CPU, a GPU, or a multi-core CPU. The calculation unit 71 may also be configured using a quantum computer. The memory 72 stores temporary data generated in conjunction with calculations. The memory 72 is, for example, a RAM. The reading unit 73 reads information from a recording medium 70 such as an optical disk or a portable memory. The storage unit 74 is non-volatile, for example, a hard disk or a non-volatile semiconductor memory.

[0098] The calculation unit 71 causes the reading unit 73 to read the computer program 741 recorded on the recording medium 70, and stores the read computer program 741 in the storage unit 74. The calculation unit 71 executes processing required for the learning device 7 in accordance with the computer program 741. The computer program 741 may be a computer program product. The computer program 741 may be downloaded from outside the learning device 7, or may be pre-stored in the learning device 7. In these cases, the learning device 7 does not need to be equipped with the reading unit 73.

[0099] The computer program 741 can be deployed to run on a single computer, or on multiple computers located at one site or distributed across multiple sites and interconnected by a communications network. That is, the learning device 7 may be configured with multiple computers, and the computer program 741 may run on multiple computers connected via a communications network. The learning device 7 may be configured using a cloud server.

[0100] The operation unit 75 receives input of information by receiving operations from the user. The operation unit 75 is, for example, a touch panel, a pen tablet, a keyboard, or a pointing device. The display unit 76 displays information. The display unit 76 is, for example, a liquid crystal display or an EL display. The operation unit 75 and the display unit 76 may be integrated. The interface unit 77 receives input of data. The learning device 7 may be realized by an information processing device 1.

[0101] The memory unit 74 stores a learning model that is the basis of the trained model 132 and training data for training the learning model. The training data includes the first biological information, the second biological information, the patient's personal information, and a correct value of an index related to the worsening of heart failure. The correct value of the index is a value related to a component in the blood that changes in response to the worsening of heart failure, and can be obtained by a blood test or the like. For example, the correct value of the index is a test value of the patient's blood concentration of BNP obtained by a blood test.

[0102] The first biological information, second biological information, patient personal information, and test values ​​of BNP blood concentrations included in the training data are associated with each other. More specifically, the first biological information at the first time point, the second biological information at the second time point, patient personal information, and test values ​​of BNP blood concentrations obtained at the evaluation time point are associated with each other and recorded in the training data. The training data records the first biological information, second biological information, personal information, and test values ​​of BNP blood concentrations obtained for multiple patients. That is, the training data includes multiple data sets consisting of the first biological information, second biological information, personal information, and test values ​​of BNP blood concentrations. Each piece of information included in the training data is input by the user operating the operation unit 75 or via the interface unit 77 and stored in the memory unit 74.

[0103] The learning model that is the basis of the trained model 132 is realized by the calculation unit 71 executing information processing in accordance with the computer program 741. The storage unit 74 stores data necessary to realize the learning model. Note that the learning model may be configured by hardware. The learning model may also be realized using a quantum computer.

[0104] The learning device 7 executes a trained model generation method. More specifically, the learning device 7 generates a trained model 132 by training the trained model using training data. FIG. 13 is a flowchart showing an example of the processing procedure executed by the learning device 7. The learning device 7 executes the following processing by the calculation unit 71 executing information processing in accordance with the computer program 741. The learning device 7 acquires training data by the calculation unit 71 reading out training data stored in the memory unit 74 (S31). In S31, the learning device 7 may acquire training data by reading out training data stored outside the learning device 7 via the interface unit 77.

[0105] The learning device 7 then generates a trained model 132 that outputs an indicator of worsening heart failure when the first biological information, the second biological information, and the patient's personal information are input through learning based on the training data (S32). In S32, the calculation unit 71 inputs the first biological information, the second biological information, and the patient's personal information included in the training data into the learning model and causes the learning model to execute processing. The learning model performs calculations in response to the input of the first biological information, the second biological information, and the patient's personal information, and outputs an estimated value of the test value of the blood BNP concentration, i.e., an estimated BNP value, as an indicator of worsening heart failure. The calculation unit 71 adjusts the calculation parameters of the learning model so as to reduce the error between the estimated BNP value output by the learning model and the test value of the blood BNP concentration associated with the first biological information, the second biological information, and the patient's personal information input to the learning model. For example, the calculation unit 71 adjusts the parameters using an error backpropagation algorithm.

[0106] The calculation unit 71 performs machine learning of the learning model by repeating processing using multiple data sets including the first biological information, the second biological information, the patient's personal information, and the test value of the BNP blood concentration included in the training data, and adjusting the parameters of the learning model. By adjusting the calculation parameters of the learning model in this manner, the calculation unit 71 generates a trained model 132. The calculation unit 71 stores the adjusted final parameters in the memory unit 74. After S32 is completed, the learning device 7 ends the processing.

[0107] The generated trained model 132 is provided in the information processing device 1. For example, the final parameters of the adjusted trained model are input to the information processing device 1 and stored in the memory unit 13. The trained model 132 is realized by the calculation unit 11 performing information processing using the stored parameters. The information processing device 1 performs the processes of S21 to S26 using the trained model 132, making it possible to easily obtain an estimated BNP value based on the first biological information, the second biological information, and the patient's personal information.

[0108] The trained model 132 may be configured to output an index related to the worsening of heart failure when the first biological information and the second biological information are input without using the patient's personal information. In this configuration, the trained model 132 is trained in advance using training data including the first biological information, the second biological information, and correct values ​​of the index. Even in this configuration, the information processing device 1 can easily obtain an estimated BNP value by performing the processes of S21 to S26 using the trained model 132.

[0109] In the processes of S21 to S26, the estimated BNP value is derived using the trained model 132, but the information processing device 1 may be configured to derive the estimated BNP value using a method other than the method using the trained model 132. For example, the information processing device 1 may be configured to derive the estimated BNP value based on the first biological information and the second biological information using a model based on Bayesian estimation. In the processes of S21 to S26, the estimated BNP value is derived, but the information processing device 1 may derive an index other than the estimated BNP value as an index related to the worsening of heart failure. For example, the information processing device 1 may derive an estimated value of the blood concentration of NT-proBNP (estimated NT-proBNP value) as an index related to the worsening of heart failure.

[0110] As an index of worsening heart failure, an estimated rate of change in BNP values ​​or an estimated rate of change in NT-proBNP values ​​derived from estimated BNP values ​​or estimated NT-proBNP values ​​at multiple time points may be used. Alternatively, as an index of worsening heart failure, a value related to a blood component other than BNP or NT-proBNP that changes in response to worsening heart failure may be used. An example of such a value related to a blood component is the blood concentration of atrial natriuretic peptide.

[0111] As described above in detail, the information processing device 1 acquires first biological information and second biological information based on the measurement data obtained by the wearable device 2, and derives an index related to the worsening of heart failure based on the first biological information and the second biological information. When the patient's 5 heart failure worsens, hemodynamics changes, and the first biological information related to hemodynamics changes. Therefore, it is possible to derive an index related to the worsening of heart failure based on the first biological information. Furthermore, when the patient's 5 heart failure worsens, a compensatory mechanism is activated, and the second biological information related to the compensatory mechanism associated with heart failure changes. Therefore, it is possible to derive an index related to the worsening of heart failure based on the second biological information.

[0112] Pulse variability, which is the first biological information, fluctuates sensitively with the worsening of heart failure. Water content, which is the second biological information, also fluctuates sensitively with the worsening of heart failure. By using pulse variability and water content as the first biological information and the second biological information, an index that is particularly sensitive to the worsening of heart failure can be derived.

[0113] In this embodiment, since the index related to the worsening of heart failure is derived based on both of two types of biological information, namely, the first biological information and the second biological information, the variation of the index is suppressed and a more reliable index is obtained compared to when the index is derived based on one type of biological information. By using the more reliable index, the medical professional 6 can evaluate the worsening of heart failure in the patient 5 more accurately than before.

[0114] Conventional methods for assessing the worsening of heart failure based on the blood concentration of BNP require blood tests to measure the blood concentration of BNP, making it difficult to frequently assess the worsening of heart failure. In this embodiment, an index related to the worsening of heart failure is obtained from the measurement data obtained by the wearable device 2, making it possible to frequently obtain the index and constantly monitor the degree of worsening of heart failure. This also makes it possible to detect serious worsening of heart failure early. Because the patient 5 only needs to wear the wearable device 2, the burden on the patient 5 is reduced compared to methods that require blood tests.

[0115] The wearable device 2 may include sensors other than the first sensor 22 and the second sensor 23, and data measured by each sensor may be used. For example, the wearable device 2 may include a temperature sensor or an acceleration sensor. The temperature sensor may measure the body temperature of the patient 5, and the information processing device 1 may determine whether the patient 5 is suffering from an infectious disease. Based on the measurement results of the acceleration sensor, the information processing device 1 can determine the activity state of the patient 5. For example, the information processing device 1 can determine that the patient 5 is active when acceleration is high, and determine that the patient 5 is sleeping when acceleration is low. It has been reported that the pulse rate or respiratory rate is more highly correlated with the worsening of heart failure at night. By using the first biological information during sleep, the information processing device 1 can derive an index that more accurately represents the degree of worsening heart failure.

[0116] The information processing device 1 may generate activity amount information representing the patient's activity amount based on the measurement results of the acceleration sensor, and may also use the activity amount information to derive an index related to the worsening of heart failure. The activity amount information may be, for example, the cumulative number of steps taken over a predetermined time period (e.g., one hour). The activity amount information may be, for example, calculated from the cumulative number of steps taken over a first time period (e.g., 30 minutes) to obtain a representative value of the cumulative number of steps taken over a second time period (e.g., three hours) longer than the first time period. The representative value may be, for example, the average, median, or mode of the cumulative number of steps taken over multiple first time periods included in the second time period. The acceleration sensor may be a triaxial acceleration sensor. The activity amount information may be information other than the number of steps. The intensity of physical activity expressed in units of METs may be calculated from a composite acceleration calculated based on the triaxial acceleration data, and exercise (EX) calculated by integrating the intensity of physical activity over time may be used as the activity amount information.

[0117] The wearable device 2 measures acceleration using an acceleration sensor. The information processing device 1 performs processes similar to those in S12 to S14 to generate activity amount information based on the measurement results and records the activity amount information in the biological information DB 133 in association with a third time point at which acceleration was measured. The information processing device 1 acquires the activity amount information in S21 and inputs the activity amount information to the trained model 132 along with the first biological information, the second biological information, and personal information in S23. FIG. 14 is a conceptual diagram showing an example of the function of the trained model 132 that uses the activity amount information. The trained model 132 receives the first biological information, the second biological information, personal information of the patient 5, and the activity amount information. FIG. 14 shows pulse variability, pulse rate, blood pressure, and respiratory rate as examples of multiple types of first biological information, and extracellular water volume and extracellular water change rate as examples of multiple types of second biological information. The third time point associated with the activity amount information input to the trained model 132 is a predetermined time point relative to the first time point, the second time point, and the evaluation time point. The predetermined time point relative to the first time point, the second time point, and the evaluation time point is a time point at a predetermined position on the time axis relative to the first time point, the second time point, and the evaluation time point, for example, a time point at a predetermined time difference on the time axis relative to the first time point, the second time point, and the evaluation time point. The third time point may be the same time point as any of the first time point, the second time point, and the evaluation time point.

[0118] The trained model 132 in this form is pre-trained to output an index (estimated BNP value) related to the worsening of heart failure when first biological information, second biological information, personal information, and activity amount information are input. The calculation unit 11 inputs the first biological information, second biological information, personal information of the patient 5, and activity amount information to the trained model 132, causes the trained model 132 to execute processing, and acquires the estimated BNP value output by the trained model 132. When heart failure begins to worsen, physical function declines, which ultimately affects the patient's activity level. Therefore, by deriving an estimated BNP value, which is an index related to the worsening of heart failure, using activity amount information in addition to the first biological information and second biological information, it is possible to more accurately evaluate the worsening of heart failure in the patient 5.

[0119] In the present embodiment, an example has been shown in which the outputted indices are checked by the medical professional 6, but the outputted indices may also be checked by someone other than the medical professional 6. For example, the indices relating to the worsening of heart failure may be checked by the patient 5. For example, the terminal device 3 may be a smartphone carried by the patient 5. Alternatively, the wearable device 2 may have the functions of the terminal device 3, and the indices may be output using the wearable device 2. The patient 5 can check the indices relating to the worsening of his or her own heart failure and evaluate the worsening of heart failure himself or herself.

[0120] In the present embodiment, an example has been shown in which the information processing device 1 generates the first biological information and the second biological information based on data measured by the wearable device 2. The information processing system 100 may be configured such that the wearable device 2 generates the first biological information and the second biological information, and the information processing device 1 acquires the first biological information and the second biological information from the wearable device 2. Alternatively, the wearable device 2 may have the functions of the information processing device 1 and the terminal device 3, and all processing may be performed solely by the wearable device 2.

[0121] The present invention is not limited to the above-described embodiments, and various modifications are possible within the scope of the claims. In other words, embodiments obtained by combining technical means modified appropriately within the scope of the claims are also included in the technical scope of the present invention.

[0122] The matters described in each embodiment can be combined with each other. Furthermore, the independent claims and dependent claims described in the claims can be combined with each other in any and all combinations, regardless of the reference format. Furthermore, the claims do not use a format in which a claim references two or more other claims (multiple claim format), but this is not limited to this. They may be written using a multiple claim format or a format in which multiple claims (multi-multi claim) reference at least one other multiple claim.

[0123] REFERENCE SIGNS LIST 100 Information processing system 1 Information processing device 10 Recording medium 11 Calculation unit 13 Storage unit 131 Computer program 132 Trained model 2 Wearable device 22 First sensor 23 Second sensor 3 Terminal device 4 Communication network 5 Patient 6 Medical worker 7 Learning device

Claims

1. A computer program that causes a computer to execute a process of acquiring first biological information related to a patient's hemodynamics at a first time point and second biological information related to the compensatory mechanisms associated with heart failure of the patient at a second time point different from the first time point, and deriving an index related to the progression of the patient's heart failure at an evaluation time point that is later than the first and second time points based on the acquired first and second biological information.

2. The computer program according to claim 1, wherein the second point in time is a point in time earlier than the first point in time.

3. The computer program according to claim 1, wherein the first biological information includes pulse rate variability, and the second biological information includes the amount of water in the body.

4. The computer program according to claim 1, characterized in that the computer is caused to execute the process of intermittently or continuously acquiring the first biometric information and the second biometric information, and intermittently or continuously deriving the index.

5. The computer program according to claim 1, characterized in that the computer is caused to execute a process of calculating a representative value of the index within a predetermined period of time from the derived multiple indexes.

6. The computer program according to claim 1, characterized in that the computer is caused to execute a process of outputting a graph showing the change over time of the index.

7. The computer program according to claim 1, characterized in that it causes a computer to execute a process of outputting the index and the first biometric information and the second biometric information used to derive the index.

8. The computer program according to claim 1, characterized in that it causes a computer to execute a process of deriving the index based on the first biological information, the second biological information, and personal information of the patient.

9. The computer program according to claim 1, wherein the index is an estimated value of the blood concentration of BNP (Brain Natriuretic Peptide) or NT-proBNP (N-terminal proBNP).

10. The computer program according to claim 1, characterized in that the computer is caused to execute a process of inputting the acquired first biometric information and second biometric information into a trained model that outputs the index when the first biometric information and the second biometric information are input, and obtaining the index output by the trained model, thereby deriving the index.

11. The computer program of claim 1, wherein the first biological information includes pulse variability, pulse rate, blood pressure, and respiratory rate, and the first time point is a different time point with respect to at least two of the pulse variability, pulse rate, blood pressure, and respiratory rate.

12. The computer program of claim 1, characterized in that the computer is caused to execute a process of acquiring activity amount information representing the patient's activity amount at a predetermined time point relative to the first time point, the second time point, and the evaluation time point, and deriving the index based on the first biological information, the second biological information, and the activity amount information.

13. The computer program according to claim 1, characterized in that the computer is caused to execute a process of outputting a graph showing the change in the index over time, a graph showing the change in the first biological information over time, and a graph showing the change in the second biological information over time, with the times shown by each graph aligned to be the same.

14. The computer program according to claim 1, characterized in that it causes a computer to perform the following processes: calculate a representative value of the index within a predetermined period of time from the derived multiple indices; calculate a representative value of the first biometric information and the second biometric information within the period from the acquired multiple first biometric information and the second biometric information; and output a graph showing the change over time in the representative value of the index, a graph showing the change over time in the representative value of the first biometric information, and a graph showing the change over time in the representative value of the second biometric information, with the times shown on each graph being the same.

15. An information processing method comprising: acquiring first biological information relating to a patient's hemodynamics at a first time point; and second biological information relating to the compensatory mechanisms associated with the patient's heart failure at a second time point different from the first time point; and deriving, based on the acquired first biological information and second biological information, an index relating to the degree of progression of the patient's heart failure at an evaluation time point that is later than the first and second time points.

16. An information processing device comprising a calculation unit, which acquires first biological information relating to the hemodynamics of a patient at a first time point and second biological information relating to compensatory mechanisms associated with heart failure of the patient at a second time point different from the first time point, and derives an index relating to the degree of progression of the patient's heart failure at an evaluation time point which is later than the first time point and the second time point based on the acquired first biological information and second biological information.

17. An information processing system comprising: a wearable device; and an information processing device, wherein the wearable device has a first sensor that measures data necessary to generate first biological information related to the hemodynamics of a patient wearing the wearable device; and a second sensor that measures data necessary to generate second biological information related to compensatory mechanisms associated with heart failure, and the information processing device has a calculation unit that generates the first biological information at a first time point based on data measured by the first sensor, generates the second biological information at a second time point different from the first time point based on data measured by the second sensor, and derives an index related to the progression of the patient's heart failure at an evaluation time point that is later than the first time point and the second time point based on the generated first biological information and second biological information.

18. The information processing system described in claim 17, characterized in that the wearable device measures data intermittently or continuously using the first sensor and the second sensor, and the calculation unit generates the first biometric information and the second biometric information intermittently or continuously.

19. A method for generating a trained model, comprising: acquiring training data including first biological information relating to a patient's hemodynamics at a first time point, second biological information relating to the compensatory mechanisms associated with the patient's heart failure at a second time point different from the first time point, and a correct value of an index relating to the progression of the patient's heart failure at an evaluation time point that is later than the first time point and the second time point; and generating a trained model that outputs the index when the first biological information and the second biological information are input by learning based on the training data.

20. A method for generating a trained model as described in claim 19, characterized in that the correct value of the index contained in the training data is a value related to a component in blood, and the index output by the trained model is an estimated value related to the component in blood.

Images