Information processing method, information processing device, information processing program, generation method of learned model, and learned model

Abstract

To provide an information processing method, an information processing device, and an information processing program making it possible to know the degree of possibility of occurrence of a symptom of a respiratory disease in the future, and to provide a learned model that can be used for them, and its generation method.SOLUTION: A processor 11 acquires data based on measurement data showing a respiratory state of a person to be measured from the person to be measured, acquires first information showing action contents of the person to be measured during a future period later than the measurement timing of the measurement data, inputs the data and the first information to a learned model 13, and performs processing for allowing the person to be measured to obtain information showing the degree of possibility of occurrence of a symptom of a respiratory disease from the learned model later than the future period.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The technology disclosed herein relates to an information processing method, an information processing device, an information processing program, a method for generating a trained model, and a trained model. [Background technology]

[0002] Patent Document 1 describes a service providing device that includes a processor that acquires behavioral information indicating the duration of a facility user's presence at each location within the facility and environmental information for each location within the facility, and generates display data for displaying the environment within the facility in association with the user's behavior based on the behavioral information and the environmental information. This service providing device predicts the occurrence of symptoms such as asthma in the user based on the behavioral information and the environmental information. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Patent Publication No. 2021-196911 Summary of the Invention [Problem to be solved by the invention]

[0004] The technology disclosed herein aims to provide an information processing method, an information processing device, and an information processing program that can determine the likelihood of respiratory disease symptoms occurring in the future, as well as a trained model that can be used for these and a method for generating the same. [Means for solving the problem]

[0005] The technology of the present disclosure is as follows: Note that, although the corresponding components in the following embodiments are shown in parentheses, the present disclosure is not limited to these.

[0006] (1) A processor (processor 11) Obtain data (measurement processed data) based on measurement data indicating the respiratory state of the subject (user) measured from the subject, obtain first information indicating the behavior content of the subject in a future period (predetermined period) after the measurement timing of the measurement data, input the data and the first information into a learned model (learned model 13), and perform a process of obtaining from the learned model information indicating the likelihood that the subject will develop symptoms of a respiratory disease after the future period.

[0007] (2) The information processing method according to (1), wherein the processor obtains, as the first information, medication behavior information indicating the medication behavior of the subject in the future period.

[0008] (3) The information processing method according to (2), wherein the processor further obtains, as the first information, exercise information regarding the exercise of the subject in the future period.

[0009] (4) The information processing method according to (3), wherein the exercise information includes information on the frequency of exercising under specific circumstances.

[0010] (5) The information processing method according to any one of (1) to (4), wherein the processor further obtains, as the first information, information regarding the sleep of the subject, information regarding the drinking of the subject, information regarding the smoking of the subject, or information regarding the bathing of the subject in the future period.

[0011] (6) The information processing method according to any one of (1) to (4), wherein The information processing method in which the processor inputs a plurality of different pieces of the first information into the learned model and performs a process of obtaining the information corresponding to each of the plurality of pieces of the first information.

[0012] (7) Data based on measurement data indicating the respiratory state of the subject (user) measured from the subject is acquired, first information indicating the behavior content of the subject in a future period (predetermined period) after the measurement timing of the measurement data is acquired, An information processing apparatus (information processing server 10) including a processor (processor 11) that inputs the data and the first information into a learned model (learned model 13) and performs a process of obtaining information indicating the likelihood that the subject will develop symptoms of a respiratory disease after the future period from the learned model.

[0013] (8) Data based on measurement data indicating the respiratory state of the subject (user) measured from the subject is acquired, first information indicating the behavior content of the subject in a future period (predetermined period) after the measurement timing of the measurement data is acquired, A process of inputting the data and the first information into a learned model (learned model 13) and obtaining information indicating the likelihood that the subject will develop symptoms of a respiratory disease after the future period from the learned model, An information processing program that causes a processor (processor 11) to execute the process.

[0014] (9) The processor (processor 11) plural pieces of data (measurement processed data) based on measurement data indicating the respiratory state of the first subject (monitoring user) measured from the first subject, information indicating the behavior content of the first subject in a period (predetermined period) after the measurement timing of the measurement data, and onset presence / absence information (third data) indicating whether or not the first subject developed symptoms of a respiratory disease after the period are acquired as learning data. A method for generating a learned model that causes a program to execute machine learning based on the plurality of learning data, and inputs data (measurement processed data) based on measurement data indicating the respiratory state of the second subject (user) measured from the second subject, and first information indicating the behavior content of the second subject in a future period (predetermined period) after the measurement timing of the measurement data, and outputs information indicating the likelihood that the second subject will develop symptoms of a respiratory disease after that future period (learned model 13).

[0015] (10) A learned model (learned model 13) in which machine learning is performed using, as learning data, data (measurement processed data) based on measurement data indicating the respiratory state of the first subject (monitoring user) measured from the first subject, first information indicating the behavior content of the first subject in a period (predetermined period) after the measurement timing of the measurement data, and onset presence / absence information (third data) indicating whether or not the first subject developed symptoms of a respiratory disease after that future period, A learned model that causes a processor (processor 11) to execute a process of inputting data (measurement processed data) based on measurement data indicating the respiratory state of the second subject (user) measured from the second subject, and first information indicating the behavior content of the second subject in a future period (predetermined period) after the measurement timing of the measurement data, and outputting information indicating the likelihood that the second subject will develop symptoms of a respiratory disease after that future period.

Effects of the Invention

[0016] According to the present disclosure, it is possible to know the likelihood of developing symptoms of a respiratory disease in the future.

Brief Description of the Drawings

[0017]

Figure 1

Modes for Carrying Out the Invention

[0018] (Overview of the information processing method of the present disclosure) The information processing method is such that a processor acquires data based on measurement data indicating the respiratory state of the measured person measured from the measured person, acquires first information indicating the content of the actions of the measured person in a future period after the measurement timing of the measurement data, inputs the data and the first information into a learned model, and obtains, from the learned model, information indicating the likelihood that the measured person will exhibit symptoms of a respiratory disease after the future period.

[0019] The content of the actions includes the content of actions consciously performed by the measured person (such as taking medicine, exercising, sleeping, drinking alcohol, smoking, or taking a bath, etc.). Medication history, exercise history, sleep history, drinking history, smoking history, bath history, etc. can affect the improvement of symptoms. Thus, in addition to the data based on the measurement data measured from the measured person, by inputting the future action content of the measured person into the learned model, the likelihood of the occurrence of symptoms of the respiratory disease of the measured person in the future can be obtained from the output of the learned model.

[0020] For example, when it is found that the likelihood of symptom occurrence in the future is high by inputting specific action content, the measured person can take corresponding measures such as consciously taking medicine with reference to the action content, avoiding going to places with a large amount of allergens, getting more sleep, etc. For a doctor, when the likelihood of onset is high, corresponding measures such as changing the amount of medicine to be prescribed or increasing the frequency of hospital visits can be taken. Thus, by making it possible to predict the occurrence of future symptoms, it becomes possible to efficiently perform the treatment of respiratory diseases.

[0021] Hereinafter, a configuration example of a system including an apparatus that executes the information processing method according to the technology of the present disclosure will be described.

[0022] (System configuration) FIG. 1 is a schematic diagram showing the schematic configuration of system 100. The users targeted by system 100 are users who have been diagnosed with a respiratory disease by a doctor and have received a prescription for a therapeutic drug, but it is also possible to use the system even if the user has not received a prescription for a therapeutic drug. System 100 includes an information processing server 10, a user terminal 30 such as a smartphone owned by a user (the person to be measured), and a measuring device 40 owned by the user. The user terminal 30 and the information processing server 10 are connected to a network 20 such as the Internet and are configured to be able to communicate with each other.

[0023] The measuring device 40 is a device for measuring measurement data indicating the respiratory state of the user from the user. The measurement data indicating the respiratory state of the user is, for example, respiratory sound data (e.g., data of time vs. amplitude value), or vital capacity data (e.g., data of exhaled volume vs. flow volume (flow volume curve)).

[0024] The measuring device 40 can include, for example, a sensor unit including a microphone, and in a state where the sensor unit is applied to the user's chest, the measuring device 40 can be a device that measures and records the user's respiratory sound data with the microphone. The measuring device 40 can also be a peak flow meter that measures and records the user's vital capacity data (flow volume curve). The measuring device 40 and the user terminal 30 can be connected by wired communication or wireless communication, and the measurement data measured by the measuring device 40 is transmitted to the user terminal 30.

[0025] The information processing server 10 includes a processor 11 and a storage unit 12. The storage unit 12 includes, in addition to a work memory such as a RAM (Random Access Memory), a non-temporary storage medium such as a hard disk or a flash memory. An information processing program for the processor 11 of the information processing server 10 to execute an information processing method is stored in the storage unit 12.

[0026] The processor 11 is a general-purpose processor such as a CPU (Central Processing Unit) that executes software (program) to perform various functions, a programmable logic device (PLD) that is a processor whose circuit configuration can be changed after manufacture, such as an FPGA (Field Programmable Gate Array), or a dedicated electric circuit such as an ASIC (Application Specific Integrated Circuit) that is a processor having a circuit configuration dedicated to executing specific processing. The processor 11 may be composed of one processor, or may be composed of a combination of two or more processors of the same type or different types (for example, a plurality of FPGAs, or a combination of a CPU and an FPGA). More specifically, the hardware structure of the processor 11 is an electric circuit (circuitry) that combines circuit elements such as semiconductor elements. When the processor 11 is composed of a plurality of processors, the installation locations of these plurality of processors do not have to be within the same device, and may be provided in each of a plurality of devices distributed via the network 20.

[0027] The learned model 13 is stored in the storage unit 12. The learned model 13 is generated by executing machine learning on a learning model configured by a program using teacher data (learning data). The learned model 13 may be generated by the processor 11 of the information processing server 10, or may be generated by a processor of a computer different from the information processing server 10. Hereinafter, assuming that the processor 11 generates the learned model 13, the generation method of the learned model 13 will be described.

[0028] (Method for generating learned model) Processor 11 acquires a sample data group from storage unit 12. Each sample data included in this sample data group contains medication history data and activity history data (hereinafter referred to as first data) recorded by a monitor user different from the user who uses system 100 over a predetermined period (for example, one week, two weeks, or one month, etc.) using a terminal having the same function as user terminal 30 (hereinafter referred to as a monitor terminal). Further, each sample data contains measurement data (hereinafter referred to as second data) obtained by the monitor user measuring measurement data indicating the respiratory state using the same device as measurement device 40 at the start timing of the above-mentioned predetermined period (for example, the start date of this predetermined period or the day before). Further, each sample data contains third data which is data indicating whether or not it has been determined by a doctor's examination that there are symptoms of a respiratory disease after the end of the above-mentioned predetermined period (for example, the day after the day when this predetermined period ended). Thus, sample data including the first data, the second data, and the third data is generated for each of a large number of monitor users and recorded in storage unit 12 as a sample data group.

[0029] Note that a management application capable of managing various types of information related to the life of the monitor user is installed in the monitor terminal. The management application records medication history data indicating the medication history of the monitor user and activity state data indicating the activity state of the monitor user in the memory.

[0030] The activity state of the monitored user includes the states of activities other than taking medicine that the monitored user consciously performs (such as exercise, sleep, drinking alcohol, smoking, or taking a bath, etc.). The activity state data includes at least one of exercise history data related to exercise, sleep history data related to sleep, drinking history data related to drinking alcohol, smoking history data related to smoking, and bath history data related to taking a bath. The management application records the medicine-taking history data and activity state data of the monitored user by using, for example, the position detection function, activity amount detection function, sleep state detection function, heart rate (pulse rate) detection function, input function for various information, and acquisition function for various information from an external server via the network 20, which are installed on the monitor terminal. The management application can also record the activity state data of the monitored user by using the information detected by a wearable terminal connectable to the monitor terminal.

[0031] When the monitored user takes the therapeutic medicine prescribed by the doctor, the management application is operated to input the date and time of taking the medicine. Thereby, the management application records this input information as medicine-taking history data in the memory.

[0032] When the monitored user holds the monitor terminal or wears a wearable terminal or the like connectable to the monitor terminal and exercises, the management application acquires the exercise history data of the monitored user and records it in the memory. The exercise history data of the monitored user includes, for example, the intensity of the exercise (unit: METs) when the exercise is performed, the period during which the exercise is performed, the place where the exercise is performed (either indoors or outdoors), and the environment of the place where the exercise is performed (such as temperature or amount of allergens, etc.). The place and the environment indicate the situation where the exercise is being performed. The exercise history data of the monitored user may include the amount of activity such as the number of steps or calories consumed per unit period (such as one day, etc.). The amount of allergens is defined, for example, by the amount of pollen scattering or PM2.5, etc.

[0033] The intensity of exercise, the number of steps, and the calories burned, etc. can be obtained by the activity detection function of the monitor terminal or a wearable terminal. The location where the exercise was performed can be obtained by the location detection function of the monitor terminal or the wearable terminal. The environment of the location where the exercise was performed can be obtained by the acquisition function of various information of the monitor terminal.

[0034] When the monitored user sleeps while holding the monitor terminal or wearing a wearable terminal that can be connected to the monitor terminal, the state of that sleep is monitored by the monitor terminal or the wearable terminal, and a sleep score indicating the sleep time and the quality of sleep is recorded on the monitor terminal. The management app acquires this recorded sleep time and sleep score and records them in the memory as sleep history data.

[0035] When the monitored user drinks alcohol, the management app is operated to input the date and time of drinking and the amount of alcohol consumed, etc. Thereby, the management app records this input information in the memory as drinking history data.

[0036] When the monitored user smokes, the management app is operated to input the date and time of smoking and the number of cigarettes smoked, etc. Thereby, the management app records this input information in the memory as smoking history data.

[0037] When the monitored user takes a bath, the management app is operated to input the date and time of taking a bath and the bath time. Thereby, the management app records this input information in the memory as bath history data. In this way, the first data recorded by the monitor terminal is used as one of the sample data.

[0038] Based on the medication history data among the first data in each sample data of the sample data group acquired from the storage unit 12, the processor 11 generates medication history information regarding the medication history of the monitored user in a predetermined period. The medication history information is information obtained by processing the medication history data, and for example, it is the number of times of taking medicine (i.e., the medication frequency) during a predetermined period, or the number of times of taking medicine for each divided period when the predetermined period is divided into a plurality of periods (information indicating the distribution of the timings when medication was taken in the predetermined period), etc.

[0039] Based on the activity state data among the first data in each sample data of the sample data group acquired from the storage unit 12, the processor 11 generates activity state information regarding the activity state of the monitored user in a predetermined period. The medication history information and the activity state information constitute information indicating the behavior history of the monitored user.

[0040] As one of the activity state information, the processor 11 generates exercise history information regarding the exercise history of the monitored user in a predetermined period. The exercise history information is information obtained by processing the exercise history data included in the activity state data, and for example, it includes first exercise amount information of the exercise amount during a predetermined period, or second exercise amount information of the frequency of exercise performed under a specific situation during a predetermined period, etc. The exercise history information may be composed of a combination of the first exercise amount information and the second exercise amount information.

[0041] For the first exercise amount information, for example, the amount of physical activity (MET·hour) at a predetermined intensity (for example, 3 METs) or more is used. For example, if exercise at an intensity of 3 METs is performed for 1 hour and exercise at an intensity of 5 METs is performed for 1 hour during a predetermined period, then (3 METs × 1 hour) + (5 METs × 1 hour) = 8 [MET·hour] is generated as the first exercise amount information of the exercise amount. As the first exercise amount information of the exercise amount, the number of times or the time of physical activity at a predetermined intensity or more during a predetermined period may be generated. Also, as the first exercise amount information of the exercise amount, the cumulative number of steps or the cumulative calorie consumption during a predetermined period, etc., may be generated.

[0042] As the second amount of momentum information, for example, indoor exercise information indicating the number of times or the duration of exercise of a predetermined intensity or more indoors during a predetermined period, and outdoor exercise information indicating the number of times or the duration of exercise of a predetermined intensity or more outdoors during a predetermined period are generated.

[0043] The outdoor exercise information may be further subdivided. For example, the outdoor exercise information may be generated by being divided into first outdoor exercise information indicating the number of times or the duration of exercise of a predetermined intensity or more in a state where the amount of allergens is equal to or more than a predetermined level, and second outdoor exercise information indicating the number of times or the duration of exercise of a predetermined intensity or more in a state where the amount of allergens is less than the predetermined level.

[0044] In addition to or instead of the first outdoor exercise information and the second outdoor exercise information, third outdoor exercise information indicating the number of times or the duration of exercise of a predetermined intensity or more in a state where the temperature is equal to or more than a predetermined level, and fourth outdoor exercise information indicating the number of times or the duration of exercise of a predetermined intensity or more in a state where the temperature is less than the predetermined level may be generated. Indoor, outdoor, outdoor with an allergen amount of a predetermined level or more, outdoor with an allergen amount less than the predetermined level, outdoor with a temperature of a predetermined level or more, and outdoor with a temperature less than the predetermined level each constitute a specific situation.

[0045] Also, as one of the activity state information, the processor 11 generates sleep history information regarding the sleep history of the monitored user during a predetermined period. The sleep history information is information obtained by processing the sleep history data included in the activity state data, and is, for example, the sleep time per day during a predetermined period.

[0046] Also, as one of the activity state information, the processor 11 generates drinking history information regarding the drinking history of the monitored user during a predetermined period. The drinking history information is information obtained by processing the drinking history data included in the activity state data, and is, for example, the number of times of drinking, the amount of drinking, or the average amount of drinking per day during a predetermined period.

[0047] In addition, as one of the activity state information, the processor 11 generates smoking history information regarding the smoking history of the monitored user during a predetermined period. The smoking history information is information obtained by processing the smoking history data included in the activity state data, and is, for example, the number of cigarettes smoked during a predetermined period or the average number of cigarettes smoked per day.

[0048] In addition, as one of the activity state information, the processor 11 generates bathing history information regarding the bathing history of the monitored user during a predetermined period. The bathing history information is information obtained by processing the bathing history data included in the activity state data, and is, for example, the number of baths, the bathing time, or the average bathing time per day during a predetermined period.

[0049] Based on the second data (measurement data) in each sample data of the sample data group acquired from the storage unit 12, the processor 11 generates measurement processed data obtained by processing this second data into a form suitable for learning.

[0050] As the measurement processed data, feature values extracted from the measurement data (respiratory sound data or vital capacity data) are used. Feature values that can produce a significant difference between a person without a respiratory disease and a person with a respiratory disease can be appropriately adopted.

[0051] When the measurement data is respiratory sound data, for example, the processor 11 generates frequency - amplitude value data from the respiratory sound data (time - amplitude value data), selects a plurality of predetermined frequencies (frequencies at which the difference in amplitude is particularly large between a person with a respiratory disease and a person without one) in this data, and obtains the amplitude values corresponding to each of these frequencies as the measurement processed data.

[0052] Alternatively, the processor 11 may process the respiratory sound data to generate frequency - sound pressure data, and acquire, as the measurement processed data, the ratio between the peak value of the sound pressure in this data and the width between the frequencies corresponding to the minimum sound pressure values on both sides of the peak value.

[0053] Alternatively, the processor 11 may process the breath sound data to generate three-dimensional data of frequency, sound pressure, and time, and select, from this three-dimensional data, the frequency during the period when the sound pressure reaches the maximum value as measurement and processing data. Further, the processor 11 may obtain, as measurement and processing data, characteristic values (for example, the number of times the amplitude becomes a certain level or more and the time thereof, etc.) that can be directly extracted from the breath sound data without performing frequency analysis on the breath sound data.

[0054] It is known that the vital capacity data (flow-volume curve) measured by a peak flow meter has a difference in shape between a person without a respiratory disease and a person with a respiratory disease. Therefore, when the measurement data is vital capacity data, the processor 11 extracts a flow rate (for example, a flow rate corresponding to 50% or 25% of the maximum value of the exhaled volume, etc.) that can identify the difference in the shape from the flow-volume curve as the vital capacity data, and obtains this flow rate as measurement and processing data.

[0055] Alternatively, the processor 11 may extract the flow rates corresponding to each of a plurality of ratios (for example, 90%, 80%, 70%, etc.) with respect to the maximum value of the exhaled volume from the flow-volume curve, and use these as measurement and processing data.

[0056] Alternatively, the processor 11 generates data with the absolute expiratory volume on the horizontal axis and the flow rate on the vertical axis from the vital capacity data. In this data, it is known that there is a difference in the centroid position of the absolute expiratory volume between a person without a respiratory disease and a person with a respiratory disease. Therefore, the processor 11 may specify the centroid value of the absolute expiratory volume from this absolute expiratory volume vs. flow rate data, and use this centroid value as measurement and processing data.

[0057] The processor 11 uses, as teacher data, a data set of the first data (medication history information, activity state information (at least one of exercise history information, sleep history information, drinking history information, smoking history information, and bathing history information)) corresponding to the monitored user generated as described above, the measurement and processed data (second data), and the third data (judgment result of the presence or absence of symptoms by a doctor) corresponding to the monitored user, and causes a machine learning based on a plurality of data sets corresponding to a plurality of monitored users to be executed on a learning model to generate a learned model 13.

[0058] When the learned model 13 is input with measurement and processed data generated based on measurement data measured from an arbitrary user and first information indicating the future behavior content of the user in a future period after the measurement timing of the measurement data, it learns various parameters so as to output information indicating the likelihood that the user will develop symptoms of a respiratory disease after that future period (the next day after the end of that period, etc.). The method of machine learning is not particularly limited, and for example, any method such as logistic regression, decision tree, random forest, gradient boosting decision tree, neural network, etc. can be used.

[0059] (Use of the learned model) First, the user measures measurement data with the measurement device 40 while the user terminal 30 and the measurement device 40 are connected. The measurement data measured by the measurement device 40 and the measurement date and time information thereof are acquired by a predetermined application installed in the user terminal 30. When the predetermined application acquires the measurement data, it transmits the measurement data to the information processing server 10.

[0060] When the processor 11 acquires the measurement data transmitted from the user terminal 30, it generates the above-described measurement processed data based on the measurement data. Further, the processor 11 acquires first information indicating the future action content of the user in a future period until a predetermined time (a time of the same length as the above-mentioned predetermined period in each sample data) elapses from the measurement date and time of the measurement data. This first information is information corresponding to at least one of the above-described medication history information and activity history information.

[0061] A plurality of first information with different contents are stored in the storage unit 12, and the processor 11 may acquire at least one predetermined first information from the storage unit 12, or may acquire at least one first information selected by the user in the above-described predetermined application.

[0062] For example, it is assumed that information indicating the medication behavior in the above-described future period is used as the first information. In this case, for example, the first information is composed of the number of times of taking medicine X (information corresponding to the medication history information) in the future period. For example, the processor 11 acquires three pieces of first information with X = 0, X = 5, and X = 10.

[0063] Next, the processor 11 inputs a set of any one of the three pieces of first information and the measurement processed data into the learned model 13, and performs a process of acquiring, as an output of the learned model 13, information (for example, onset probability, etc.) indicating the likelihood that the user will develop symptoms of a respiratory disease after the above-described future period. The processor 11 repeats this process by changing the first information input to the learned model 13 to obtain the onset probability for each of the three patterns of the number of times of taking medicine.

[0064] When the processor 11 acquires the onset probability for each number of times of taking medicine, it transmits this onset probability to the management application of the user terminal 30. The management application performs a process of displaying the received onset probability on the display device of the user terminal 30 or outputting it as voice from the speaker.

[0065] As a result, the user can know the onset probability when no medication is taken in the future period from the current time (the timing when the measurement by the measuring device 40 is performed), the onset probability when five doses of medication are taken in this future period, and the onset probability when ten doses of medication are taken in this future period.

[0066] By comparing the onset probabilities for each number of doses obtained in this way, it is possible to determine how to take the medication to reduce the future onset probability. The user can determine the appropriate actions regarding the medication to be taken in the future based on the display data and the like displayed on the user terminal 30. If the user terminal 30 is a terminal managed by a doctor, this display data can be used as a reference when the doctor determines the future treatment policy.

[0067] As the first information input to the learned model 13, in addition to the information indicating the medication behavior, information regarding exercise can also be used. For example, the first information is composed of a combination of the number of times Y (information corresponding to indoor exercise information) of exercise with a predetermined intensity or more indoors and the number of times Z (information corresponding to outdoor exercise information) of exercise with a predetermined intensity or more outdoors. In this case, for example, the processor 11 acquires four pieces of the first information: the first information where Y = 0 and Z = 0, the first information where Y = 1 and Z = 0, the first information where Y = 0 and Z = 1, and the first information where Y = 1 and Z = 1.

[0068] Next, the processor 11 performs a process of inputting a set of any one of the four acquired pieces of the first information and the measurement and processing data into the learned model 13 and obtaining, as the output of the learned model 13, the onset probability of the user after the future period. The processor 11 performs this process four times by changing the first information input to the learned model 13 to obtain the onset probabilities for each of the four exercise patterns during the future period.

[0069] When the processor 11 obtains the four onset probabilities, it transmits these onset probabilities to the management app of the user terminal 30. The management app performs processes such as displaying the received onset probabilities on the display device of the user terminal 30 or outputting them as audio from the speaker. As a result, the user can know how the onset probability changes depending on how much indoor exercise and outdoor exercise are performed from the current time.

[0070] Also, the first information may be composed of the combinations of the above-mentioned number of times X, number of times Y, and number of times Z. In this case, the processor 11 obtains a plurality of first information with different combinations of the numerical values of X, Y, and Z, and inputs each obtained first information and the measurement and processing data into the learned model 13 to obtain a plurality of onset probabilities. As a result, the user can know how to take medicine and exercise from the current time to reduce the possibility of onset. In the above example, the number of times X constitutes the medicine-taking behavior information, and the number of times Y and the number of times Z constitute the information on the frequency of exercising under specific circumstances.

[0071] Also, the first information may be information regarding the user's sleep, information regarding the user's drinking, information regarding the user's smoking, or information regarding the user's bathing in the above future period.

[0072] For example, the first information can be the average sleep time T per day (information corresponding to the sleep history information) in the future period. For example, the processor 11 obtains a plurality of first information with different values of the average sleep time T, inputs a set of any of these and the measurement and processing data into the learned model 13, and performs a process of obtaining the onset probability after the above future period as the output of the learned model 13. The processor 11 repeats this process while changing the first information input to the learned model 13 to obtain the onset probability for each of the plurality of sleep patterns. As a result, the user can know how the onset probability changes depending on how much sleep is taken from the current time.

[0073] Further, the first information can be the average alcohol consumption amount AM (information corresponding to drinking history information) in a future period. For example, the processor 11 acquires a plurality of first information with different values of the average alcohol consumption amount AM, inputs a set of any of these and the measurement and processing data into the learned model 13, and performs a process of acquiring the onset probability after the above future period as the output of the learned model 13. The processor 11 repeats this process while changing the first information input to the learned model 13 to obtain the onset probability for each of a plurality of drinking patterns. Thereby, the user can know how the onset probability changes depending on how much drinking is done from the current time.

[0074] Further, the first information can be the average number of cigarettes smoked S (information corresponding to smoking history information) in a future period. For example, the processor 11 acquires a plurality of first information with different values of the average number of cigarettes smoked S, inputs a set of any of these and the measurement and processing data into the learned model 13, and performs a process of acquiring the onset probability after the above future period as the output of the learned model 13. The processor 11 repeats this process while changing the first information input to the learned model 13 to obtain the onset probability for each of a plurality of smoking patterns. Thereby, the user can know how the onset probability changes depending on how much smoking is done from the current time.

[0075] Further, the first information can be the number of bath times B (information corresponding to bath history information) in a future period. For example, the processor 11 acquires a plurality of first information with different values of the number of bath times B, inputs a set of any of these and the measurement and processing data into the learned model 13, and performs a process of acquiring the onset probability after the above future period as the output of the learned model 13. The processor 11 repeats this process while changing the first information input to the learned model 13 to obtain the onset probability for each of a plurality of bath patterns. Thereby, the user can know how the onset probability changes depending on how much bathing is done from the current time.

[0076] The first information can also be composed of a combination of two or more of the above-mentioned number of times X, number of times Y, number of times Z, average sleep time T, average alcohol consumption AM, average number of cigarettes smoked S, and number of baths B.

[0077] (Modified example of the learned model) The learned model 13 was obtained by performing machine learning using the data sets of the first data, the second data, and the third data as teacher data. However, machine learning may also be executed by further including user information in this data set.

[0078] User information is information regarding the attributes (such as gender, age, physical build (BMI), or nationality, etc.) of the monitored user who is the source of the measurement data included in the sample data. The processor 11 generates medication history information, activity state information, and measurement processed data based on each sample data in the sample data group, and uses these, user information, and the third data as teacher data, and performs machine learning on the learning program based on this teacher data to generate the learned model 13. Thereby, when measurement processed data, user information, and the first information indicating future action content for an arbitrary user are input, a learned model 13 can be generated that outputs the probability of occurrence of symptoms of a respiratory disease in the future for that user. In this way, by further using user information, the possibility of future onset can be predicted with high accuracy.

[0079] In the above description, it is assumed that the processor 11 generates the measurement processed data, but it is not limited to this. For example, a terminal-side processor that executes a predetermined application on the user terminal 30 may generate the measurement processed data and transmit it to the information processing server 10. Also, when the terminal-side processor generates the measurement processed data, a configuration may be adopted in which the learned model 13 is incorporated into the predetermined application. In this case, the terminal-side processor inputs the first information read from the memory and the measurement processed data into the learned model 13, and obtains the onset probability from the learned model 13.

Explanation of reference numerals

[0080] 10 Information processing server 11 Processor 12 Memory unit 13 Model 20 Network 30 User terminal 40 Measuring device 100 System

Claims

1. A processor obtains data based on measurement data indicating the respiratory state of the subject measured from the subject, obtains first information indicating the content of the subject's behavior in a future period after the measurement timing of the measurement data, inputs the data and the first information into a learned model, and performs a process of obtaining information indicating the likelihood that the subject will exhibit symptoms of a respiratory disease after the future period from the learned model. An information processing method.

2. The information processing method according to claim 1, wherein the processor obtains medication behavior information indicating the subject's medication behavior in the future period as the first information. An information processing method.

3. The information processing method according to claim 2, wherein the processor further obtains exercise information related to the subject's exercise in the future period as the first information. An information processing method.

4. The information processing method according to claim 3, wherein the exercise information includes information on the frequency of exercising under specific circumstances. An information processing method.

5. The information processing method according to any one of claims 1 to 4, wherein the processor further obtains information related to the subject's sleep, information related to the subject's drinking, information related to the subject's smoking, or information related to the subject's bathing in the future period as the first information. An information processing method.

6. The information processing method according to any one of claims 1 to 4, wherein the processor inputs a plurality of different pieces of the first information into the learned model and performs a process of obtaining the information corresponding to each of the plurality of pieces of the first information. An information processing method.

7. An information processing apparatus comprising a processor that obtains data based on measurement data indicating the respiratory state of the subject measured from the subject, obtains first information indicating the content of the subject's behavior in a future period after the measurement timing of the measurement data, inputs the data and the first information into a learned model, and performs a process of obtaining information indicating the likelihood that the subject will exhibit symptoms of a respiratory disease after the future period from the learned model.

8. A processor obtains data based on measurement data indicating the respiratory state of the subject measured from the subject, obtains first information indicating the content of the subject's behavior in a future period after the measurement timing of the measurement data, obtains first information indicating the content of the subject's behavior in a future period after the measurement timing of the measurement data, An information processing program that causes a processor to perform a process of inputting the data and the first information into a learned model and obtaining, from the learned model, information indicating the likelihood that the subject will develop symptoms of a respiratory disease after the future period.

9. The processor acquires a plurality of pieces of learning data including data based on measurement data indicating the respiratory state of the first subject measured from the first subject, information indicating the behavior history of the first subject in a period after the measurement timing of the measurement data, and onset presence / absence information indicating whether or not the first subject developed symptoms of a respiratory disease after that period, causes a program to perform machine learning based on the plurality of pieces of learning data, and uses, as inputs, data based on measurement data indicating the respiratory state of the second subject measured from the second subject and first information indicating the behavior details of the second subject in a future period after the measurement timing of the measurement data, to obtain a learned model that outputs information indicating the likelihood that the second subject will develop symptoms of a respiratory disease after the future period.

10. A learned model in which machine learning has been performed using, as learning data, data based on measurement data indicating the respiratory state of the first subject measured from the first subject, information indicating the behavior history of the first subject in a period after the measurement timing of the measurement data, and onset presence / absence information indicating whether or not the first subject developed symptoms of a respiratory disease after that period, wherein the learned model causes a processor to perform a process of using, as inputs, data based on measurement data indicating the respiratory state of the second subject measured from the second subject and first information indicating the behavior details of the second subject in a future period after the measurement timing of the measurement data, and outputting information indicating the likelihood that the second subject will develop symptoms of a respiratory disease after the future period.

Images