Estimation device, trained model, estimation method, trained model generation method, information processing device, information processing method, and computer program

Abstract

This estimation device is for estimating biological-related condition information of a subject, the estimation device comprising: an estimation unit that inputs input information to a trained model and acquires output information from the trained model; and a storage unit that stores the output information. The biological-related condition information includes information indicating the mental and physical condition of the subject or information indicating the environmental condition that affects the mental and physical condition of the subject. The input information includes at least mental and physical recovery ability information indicating the degree of speed of mental and physical recovery of the subject following completion of exercise. The output information includes the biological-related condition information.

Description

Estimation device, learning model, estimation method, learning model generation method, information processing device, information processing method, and computer program

[0001] The present disclosure relates to an estimation device, a learning model, an estimation method, a learning model generation method, an information processing device, an information processing method, and a computer program.

[0002] One of the objectives of the information processing system described in Patent Document 1 is to understand the frailty state of a subject. To achieve this objective, the information processing system generates recovery index information that indicates how the subject is recovering from the response to the occurrence of exercise intensity, based on information on changes in biological data (particularly heart rate and blood pressure) from a moving state to a resting state.

[0003] Patent No. 7065550

[0004] However, in the above-mentioned Patent Document 1, for example, it is not clear how specifically it is desirable to generate the above-mentioned recovery index information using the heart rate, which is one of the above-mentioned biological data of the subject, and in other words, there is a problem that it is not always possible to obtain reliable recovery index information. For this reason, there is also a problem in Patent Document 1 that even if recovery index information is used, highly reliable information cannot always be obtained.

[0005] The object of the present disclosure is to provide an estimation device, a learning model, an estimation method, a learning model generation method, and a computer program that can obtain highly reliable information by effectively utilizing more reliable information on the physical and mental recovery ability of a subject than conventional information for estimating the physical and mental recovery of the subject.

[0006] Another object of the present disclosure is to provide an information processing device, an information processing method, and a computer program that can calculate more reliable mental and physical recovery ability information than conventional methods in order to estimate a subject's mental and physical recovery.

[0007] According to the present disclosure, an estimation device can be provided that estimates bio-related state information of a subject, the estimation device comprising: an estimation unit that inputs input information to a learning model and obtains output information from the learning model; and a memory unit that stores the output information, wherein the bio-related state information includes information representing the subject's mental and physical state or information representing the state of the environment that affects the subject's mental and physical state, the input information includes at least mental and physical recovery ability information that indicates the degree of speed of the subject's mental and physical recovery from the end of exercise, and the output information includes the bio-related state information.

[0008] According to the present disclosure, a learning model can be provided that causes a computer to function to estimate bio-related state information of a subject, the learning model causing the computer to function to input input information and output output information, the bio-related state information including information representing the subject's mental and physical state or information representing the state of the environment that affects the subject's mental and physical state, the input information including at least mental and physical recovery ability information indicating the degree of speed of the subject's mental and physical recovery from the end of exercise, and the output information including the bio-related state information.

[0009] According to the present disclosure, an estimation method can be provided for estimating bio-related state information of a subject, the estimation method comprising the steps of inputting input information into a learning model and acquiring output information from the learning model to which the input information has been input, wherein the bio-related state information comprises information representing the mental and physical state of the subject, or information representing the state of the environment that affects the mental and physical state of the subject, the input information comprises at least mental and physical recovery ability information indicating the degree of speed of the subject's mental and physical recovery from the end of exercise, and the output information comprises the bio-related state information.

[0010] According to the present disclosure, it is possible to provide a computer program that causes a computer to execute the above estimation method.

[0011] According to the present disclosure, a learning model generation method can be provided, comprising the steps of: acquiring a learning dataset; and performing learning using the learning dataset to generate a learning model that outputs output information when input information is input, wherein the learning dataset includes at least mental and physical recovery ability information indicating the degree of speed of the learning subject's mental and physical recovery from the end of exercise, and biological-related state information of the learning subject, wherein the biological-related state information of the learning subject includes information representing the learning subject's mental and physical state, or information representing an environmental state that affects the learning subject's mental and physical state, wherein the input information includes at least mental and physical recovery ability information indicating the degree of speed of the subject's mental and physical recovery from the end of exercise, and the output information includes the biological-related state information of the subject, and the biological-related state information of the subject includes information representing the subject's mental and physical state, or information representing an environmental state that affects the subject's mental and physical state.

[0012] According to the present disclosure, a computer program can be provided that causes a computer to execute the above learning model generation method.

[0013] According to the present disclosure, an information processing device can be provided that includes a processing unit that calculates mental and physical recovery ability information that indicates the degree of speed of the subject's mental and physical recovery since the end of exercise based on the subject's heart rate change and exercise heart rate, and a memory unit that stores the mental and physical recovery ability information, wherein the heart rate change indicates the difference between the subject's exercise heart rate and resting heart rate.

[0014] According to the present disclosure, an information processing method can be provided for calculating mental and physical recovery ability information indicating the degree of speed of a subject's mental and physical recovery since the end of exercise, the information processing method including the steps of: (1) acquiring information on the subject's heart rate; and (2) calculating the mental and physical recovery ability information based on the subject's heart rate change and heart rate during exercise, wherein the heart rate change indicates the difference between the subject's heart rate during exercise and resting heart rate.

[0015] According to the present disclosure, it is possible to provide a computer program that causes a computer to execute the above information processing method.

[0016] According to the present disclosure, it is possible to provide an estimation device, a learning model, an estimation method, a learning model generation method, and a computer program that can obtain highly reliable information by effectively utilizing more reliable information on the physical and mental recovery ability of a subject than conventional information for estimating the physical and mental recovery of the subject.

[0017] According to the present disclosure, it is possible to provide an information processing device, an information processing method, and a computer program that are capable of calculating more reliable mental and physical recovery ability information than conventional methods for estimating the mental and physical recovery of a subject.

[0018] The mind-body estimation system SSS of embodiment 1 will be described. The configuration of the wearable device WT of embodiment 1 is shown. The configuration of the mind-body estimation device SS of embodiment 1 is shown. A flowchart (basic) showing the operation of the mind-body estimation system SSS of embodiment 1. A flowchart (detailed) showing the operation of the mind-body estimation system SSS of embodiment 1. Mathematical formulas for showing the operation of the mind-body estimation system SSS of embodiment 1. A diagram (part 1) showing the operation of the mind-body estimation system SSS of embodiment 1. A diagram (part 2) showing the operation of the mind-body estimation system SSS of embodiment 1. A diagram (part 3) showing the operation of the mind-body estimation system SSS of embodiment 1. A diagram (part 4) showing the operation of the mind-body estimation system SSS of embodiment 1. A diagram (part 5) showing the operation of the mind-body estimation system SSS of embodiment 1. A diagram (part 6) showing the operation of the mind-body estimation system SSS of embodiment 1. A diagram (part 7) showing the operation of the mind-body estimation system SSS of embodiment 1. FIG. 8 is a diagram (part 8) showing the operation of the mind-body estimation system SSS of embodiment 1. FIG. 9 is a diagram (part 9) showing the operation of the mind-body estimation system SSS of embodiment 1. FIG. 10 is a flowchart (details) showing the operation of the mind-body estimation system SSS of embodiment 2. FIG. 11 is a diagram (part 1) showing the operation of the mind-body estimation system SSS of embodiment 2. FIG. 12 is a diagram (part 2) showing the operation of the mind-body estimation system SSS of embodiment 2. FIG. 13 is a diagram (part 3) showing the operation of the mind-body estimation system SSS of embodiment 2. FIG. 14 is a diagram (part 1) showing the operation of embodiment 3. FIG. 15 is a diagram (part 2) showing the operation of embodiment 3. FIG. 16 shows the hardware configuration of the mind-body estimation system SSS of embodiments 1 to 3. FIG. 17 shows the hardware configuration based on software realization of the mind-body estimation system SSS of embodiments 1 to 3. FIG. 18 is a block diagram showing an example of the configuration of an information processing system according to embodiment 4. FIG. 10 is a diagram showing an example of a training dataset according to embodiment 4. FIG. 11 is a diagram showing examples of input information and output information according to embodiment 4. FIG. 12 is a block diagram showing an example of the configuration of an estimation system according to embodiment 4. FIG. 13 is a flowchart showing an example of an estimation method executed by the estimation system according to embodiment 4. FIG. 14 is a block diagram showing an example of the configuration of an estimation device according to embodiment 4.FIG. 1 is a diagram schematically showing an example of a deep neural network according to embodiment 4. FIG. 2 is a diagram schematically showing another example of a deep neural network according to embodiment 4. FIG. 3 is a diagram schematically showing yet another example of a deep neural network according to embodiment 4. FIG. 4 is a flowchart showing an example of an estimation process according to embodiment 4. FIG. 5 is a block diagram showing an example of the configuration of a learning device according to embodiment 4. FIG. 6 is a flowchart showing an example of a learning method by the learning device according to embodiment 4. FIG. 7 is a block diagram showing an example of the configuration of a training data creation device according to embodiment 4. FIG. 8 is a flowchart showing an example of a training data creation method by the training data creation device according to embodiment 4.

[0019] Preferred embodiments of the present disclosure will be described in detail below with reference to the accompanying drawings. In this specification and drawings, components having substantially the same functional configurations are designated by the same reference numerals, and redundant description will be omitted.

[0020] Embodiments of a mind-body estimation system according to the present disclosure will be described (embodiments 1 to 3).

[0021] First Embodiment A mind-body estimation system SSS according to a first embodiment will be described.

[0022] <Configuration of First Embodiment> As shown in Fig. 1, the mind-body estimation system SSS of the first embodiment includes wearable devices WT1 to WTm (m is an integer of 2 or more) and a mind-body estimation device SS. The wearable devices WT1 to WTm and the mind-body estimation device SS are connected to each other via a network NW (e.g., the Internet) as shown in Fig. 1. The mind-body estimation device SS corresponds to an example of an "information processing device" of the present disclosure.

[0023] In the following, to facilitate explanation and understanding, for example, the same name and multiple symbols may be collectively referred to as one name and one symbol, and for example, wearable terminals WT1 to WTm may be collectively referred to as wearable terminal WT.

[0024] The wearable device WT is equipped with a function for measuring, for example, the heart rate HR of the subject HI (e.g., as shown in FIG. 7), and is attachable to the arm, head, etc. The wearable devices WT1 to WTm are used, for example, by subjects HI1 to HIm, who are subjects for whom mental and physical recovery (particularly recovery of cardiac function) is to be estimated; for example, wearable device WT1 is used by subject HI1, wearable device WT2 is used by subject HI2, and so on, and wearable device WTm is used by subject HIm. The subjects correspond to an example of a "subject" in the present disclosure.

[0025] The mind-body estimation device SS is used by an administrator KA who manages the mind-body estimation device SS.

[0026] <Configuration of Wearable Terminal WT> FIG. 2 shows the configuration of the wearable terminal WT according to the first embodiment.

[0027] As shown in FIG. 2, the wearable terminal WT of the first embodiment includes an input / output unit NS(WT), a processing unit SY(WT), a storage unit KI(WT), and a communication unit TU(WT).

[0028] The input / output unit NS (WT) includes a conventionally known sensor or the like, and acquires, for example, the heart rate HR (for example, as shown in FIG. 7) of the subject HI.

[0029] The processing unit SY (WT) performs, for example, processing related to the heart rate HR of the subject HI.

[0030] The storage unit KI (WT) stores, for example, data necessary for processing by the processing unit SY (WT).

[0031] The communication unit TU (WT) performs communication via the network NW (shown in FIG. 1), and transmits, for example, the heart rate HR of the subject HI to the mind-body estimation device SS.

[0032] <Configuration of Mind-Body Estimation Apparatus SS> FIG. 3 shows the configuration of the mind-body estimation apparatus SS of the first embodiment.

[0033] As shown in FIG. 3, the mind-body estimation device SS of the first embodiment has an input / output unit NS(SS), a processing unit SY(SS), a memory unit KI(SS), and a communication unit TU(SS).

[0034] The input / output unit NS (SS) is used by the administrator KA to, for example, input data for controlling the operation of the mind-body estimation device SS, and to output data for monitoring the operation.

[0035] The processing unit SY (SS) performs, for example, processing related to estimation of the mind and body of the subject HI.

[0036] The storage unit KI(SS) stores, for example, data necessary for processing by the processing unit SY(SS).

[0037] The communication unit TU (SS) communicates via the network NW, and receives, for example, the heart rate HR of the subject HI from the wearable device WT.

[0038] <Operation of First Embodiment> The operation of the mind-body estimation system SSS of the first embodiment will be described.

[0039] FIG. 4 is a flowchart (basic) showing the operation of the mind-body estimation system SSS of the first embodiment.

[0040] FIG. 5 is a flowchart (details) showing the operation of the mind-body estimation system SSS of the first embodiment.

[0041] FIG. 6 shows mathematical expressions for illustrating the operation of the mind-body estimation system SSS of the first embodiment.

[0042] 7 to 15 are diagrams showing the operation of the mind-body estimation system SSS of the first embodiment.

[0043] The operation of the mind-body estimation system SSS of the first embodiment will be described with reference to FIGS.

[0044] <Basic Operation of Mind-Body Estimation System SSS> The basic operation of the mind-body estimation system SSS of the first embodiment will be described mainly with reference to FIG.

[0045] In the following, for ease of explanation and understanding, it is assumed that the subject whose mind and body is to be estimated is the subject HI1 among the subjects HI1 to HIm (shown in FIG. 1).

[0046] Step S1: The wearable device WT1 (shown in FIGS. 1 and 2) of the subject HI1 acquires the heart rate HR of the subject HI1 for a predetermined period (e.g., 1 minute, 3 minutes, or 5 minutes). The wearable device WT1 transmits the acquired heart rate HR to the mind-body estimation device SS (shown in FIGS. 1 and 3) via the network NW.

[0047] When the mind-body estimation device SS receives the heart rate HR from the wearable device WT1, it calculates the heart rate HR(t) (shown in FIG. 7) which is the heart rate HR per unit time (t) (for example, one minute).

[0048] 7, the mind-body estimation device SS calculates, for example, heart rate HR(20:59), which is the heart rate HR at 20:59 (1 minute), heart rate HR(21:00),..., heart rate HR(21:54), which is the heart rate HR at 21:54 (1 minute), and heart rate HR(21:55), which is the heart rate HR at 21:55 (1 minute). In this way, the mind-body estimation device SS obtains information on heart rate HR(t).

[0049] Step S2: The mind-body estimation device SS calculates a heart rate change ΔHR(t), which is the difference in the heart rate HR(t) of the subject HI1 between adjacent unit times. More specifically, as shown in Figure 8, the mind-body estimation device SS calculates a heart rate change ΔHR(t), which is the difference between the heart rate HR(t-1) of the subject HI1 during a first unit time (t-1) when the subject HI1 is exercising (for example, low-intensity exercise such as walking that the subject HI1 can actually perform), and the heart rate HR(t) during a second unit time (t) following the first unit time (t-1) when the subject HI1 is at rest.

[0050] As a result, as shown in Figure 8, the mind-body estimation device SS calculates a heart rate change amount ΔHR(12:01) = -15 times, which is the difference between the heart rate HR(12:00), which is the heart rate at 12:00 (corresponding to (t-1)), and the heart rate HR(12:01), which is the heart rate at 12:01 (corresponding to t); and similarly, calculates a heart rate change amount ΔHR(16:31) = -40 times, which is the difference between the heart rate HR(16:30), which is the heart rate at 16:30 (corresponding to (t-1)), and the heart rate HR(16:31), which is the heart rate at 16:31 (corresponding to t).

[0051] Here, the distinction between "when exercising" (i.e., when subject HI1 is in the first state) and "when at rest" (i.e., when subject HI1 is in the second state) is merely subjective, in other words, it is not uniquely determined numerically, for example, on the conventionally known METs.

[0052] However, as an example, the mind and body estimation device SS determines whether the subject HI1 is "exercising" or "resting" based on the exercise intensity (e.g., METs) of the subject HI1 or the number of steps taken by the subject HI1. In this case, the mind and body estimation device SS acquires information on the exercise intensity (e.g., METs) or the number of steps taken from the wearable device WT1.

[0053] For example, the mind-body estimation device SS determines that the subject HI1 is exercising when METs exceeds a predetermined value, and determines that the subject HI1 is resting when METs is equal to or less than the predetermined value. In this case, the predetermined value is, for example, "1," but is not particularly limited thereto. In this case, METs is, for example, METs per unit time. In addition, for example, the mind-body estimation device SS determines that the subject HI1 is exercising when the number of steps exceeds a predetermined value, and determines that the subject HI1 is resting when the number of steps is equal to or less than the predetermined value. In this case, the predetermined value is, for example, "0," but is not particularly limited thereto. In this case, the number of steps is, for example, the number of steps per unit time.

[0054] Furthermore, "exercise" indicates that the subject HI1 is moving, and whether or not the subject HI1 is exercising is determined by the exercise intensity or the number of steps. Therefore, the degree of movement in "exercise" can be determined arbitrarily based on the magnitude of the above-mentioned predetermined value. This also applies to "rest." In other words, based on the predetermined value, a state of not exercising is a resting state, and a state of not resting is a state of exercise.

[0055] Step S3: The mind-body estimation device SS approximates the relationship between the heart rate HR(t-1) and the heart rate change ΔHR(t) using a model function MF, where the model function MF corresponds to, for example, the curve equation KH (shown in FIGS. 6 and 17) and the line equation CH (shown in FIGS. 6 and 17).

[0056] When analyzing the relationship between the heart rate HR(t-1) and the amount of change in heart rate ΔHR(t) according to the distribution law "t distribution," the curve equation KH and linear equation CH shown in FIG. 6 are used. On the other hand, when analyzing according to the distribution law "Shifted-Wald distribution" (hereinafter referred to as "SW distribution"), the curve equation KH and linear equation CH (shown in FIG. 17) are used.

[0057] Here, "Y" in the distribution BP(t-1) (shown in FIGS. 6 and 17) corresponds to the amount of change in heart rate ΔHR(t), and "X" corresponds to the heart rate HR(t-1).

[0058] The subscript "i" in the distribution BP(t-1), curve equation KH, straight line equation CH, etc. (shown in Figures 6 and 17) indicates "1 to m" for subjects HI1 to HIm, and the subscript "j" indicates the jth subject of the same subject HI (for example, "1j" indicates the jth subject for subject HI1).

[0059] <t-distribution> For the distribution BP(t-1) (shown in FIG. 6), "μi,j" indicates the expected value, "σ" indicates the variance, and "ν" indicates the degree of freedom (the degree to which outliers are not tolerated). The expected value "μi,j" is represented by the curve equation KH (shown in FIG. 6), and "zi,j" in the curve equation KH is represented by the straight line equation CH (shown in FIG. 6). "A" in the curve equation KH indicates the difference between the upper and lower limit values ​​of the curve represented by the curve equation KH (hereinafter referred to as "fluctuation range A"). The curve equation KH represents, for example, a logistic curve. "b0j" and "b1j" in the straight line equation CH respectively indicate the intercept and slope of the line represented by the straight line equation CH.

[0060] <SW Distribution> For the distribution BP(t-1) (shown in FIG. 17), "γi,j" indicates the skewness of the distribution toward positive values, "αi,j" indicates the position of the peak in the distribution, and "θi,j" indicates the lower limit (lower limit value) of the possible values. "θi,j" is expressed by the curve equation KH (shown in FIG. 17), and "zi,j" in the curve equation KH is expressed by the straight line equation CH (shown in FIG. 17). "A" in the curve equation KH indicates the difference (amplitude) between the upper and lower limit values ​​of the curve shown by the curve equation KH. The curve equation KH represents, for example, a logistic curve. "b0j" and "b1j" in the straight line equation CH indicate the intercept and slope of the line shown by the straight line equation CH, respectively.

[0061] Note that "γi,j" is expressed by the related equation 1 (shown in FIG. 17), and "αi,j" is expressed by the related equation 2 (shown in FIG. 17). "c0j" and "c1j" in equation 1 represent the intercept and slope of the linear equation "c0j + Xi,j · c1j" in equation 1, respectively. "d0j" and "d1j" in equation 2 represent the intercept and slope of the linear equation "d0j + Xi,j · d1j" in equation 2, respectively.

[0062] 4 and 8, the explanation will be continued. In step S3, the mind-body estimation device SS approximates the relationship between, for example, the heart rate HR(12:00)=80 beats / minute ("12:00" corresponds to (t-1), as shown in FIG. 8) and the amount of change in heart rate ΔHR(12:01)=-15 beats ("12:01" corresponds to t, as shown in FIG. 8) using a model function MF (for example, the curve equation KH and the linear equation CH shown in FIG. 6), and similarly approximates the relationship between, for example, the heart rate HR(16:30)=90 beats / minute ("16:30" corresponds to (t-1), as shown in FIG. 8) and the amount of change in heart rate ΔHR(16:31)=-40 beats ("16:31" corresponds to t, as shown in FIG. 8) using a model function MF (same as above).

[0063] Step S4: The mind-body estimation device SS acquires parameter values ​​PV that realize the model function MF. More specifically, the mind-body estimation device SS acquires parameter values ​​PV (e.g., the slope b1j in the linear equation CH (shown in FIGS. 6 and 17)) that realize the curve equation KH and linear equation CH described above (shown in FIGS. 6 and 17). In addition to the slope b1j, the mind-body estimation device SS may also acquire an intercept b0j and / or a fluctuation range A as the parameter values ​​PV. The slope b1j indicates the degree of speed of recovery of the heart rate (e.g., heart rate) of the subject HI1. For example, the larger the absolute value of the slope b1j for the subject HI1, the faster the recovery of the heart rate (e.g., heart rate) of the subject HI1. The intercept b0j indicates the degree of magnitude of the heart rate of the subject HI1 at rest. For example, the larger the intercept b0j, the higher the resting heart rate of subject HI1. The fluctuation range A indicates the extent of the maximum recovery range (maximum recovery amount) of the heart rate of subject HI1 in the range from the lower limit to the upper limit of the heart rate HR(t-1). For example, the larger the fluctuation range A, the greater the maximum recovery range of the heart rate of subject HI1.

[0064] The parameter value PV is an example of mental and physical recovery ability information. Specifically, the processes (information processing method) of steps S1 to S4 are executed by the processing unit SY (SS).

[0065] <Operation (Details) of Mind-Body Estimation System SSS> The operation (details) of the mind-body estimation system SSS of the first embodiment will be described mainly with reference to FIG.

[0066] The essence of the mind-body estimation system SSS of embodiment 1 is to "approximate the relationship between the heart rate HR(t-1) and the amount of change in heart rate ΔHR(t) with a curve equation KH," and thereby "obtain, by Bayesian estimation or the like, parameter values ​​(e.g., slope b1j) that minimize the overall error between the relationship and the curve equation KH." For the sake of ease of explanation and understanding, a somewhat detailed discussion of whether or not a necessary and sufficient number of heart rates HR(t-1) exist for the same heart rate HR(t-1), for example, with reference to FIG. 10, will be omitted.

[0067] FIG. 5 shows the operation of the mind-body estimation system SSS of the first embodiment in more detail than FIG.

[0068] Regarding the correspondence between FIG. 5 and FIG. 4, steps S30A to S30E (shown in FIG. 5) are details of step S3 (shown in FIG. 4), and step S40A (shown in FIG. 5) is details of step S4 (shown in FIG. 4).

[0069] In the following, for ease of explanation and understanding, it is assumed that the subject whose mind and body is to be estimated is the subject HI1 out of the subjects HI1 to HIm (shown in FIG. 1), as described above.

[0070] Step S10: As in step S1, the mind-body estimation device SS calculates the heart rate HR(t) (shown in FIG. 7) which is the heart rate HR per unit time (t) (for example, one minute).

[0071] Step S20: As in step S2, the mind-body estimation device SS calculates the heart rate change ΔHR(t), which is the difference between the heart rate HR(t-1) within the first unit time (t-1) and the heart rate HR(t) within the second unit time (t), for subject HI1, as shown in Figure 8.

[0072] Step S30A: The mind-body estimation device SS assigns the relationship between the heart rate HR(t-1) and the heart rate change ΔHR(t) to a two-dimensional coordinate space 2ZK, as shown in Fig. 9. Here, "assigning" means to plot, or to draw, mark, or place a point.

[0073] More specifically, as shown or suggested in Figures 8 and 9, the mind-body estimation device SS assigns a relationship in a two-dimensional coordinate space 2ZK between, for example, heart rate HR(12:00) = 80 beats / minute ("12:00" corresponds to (t-1)) and heart rate change amount ΔHR(12:01) = -15 beats ("12:01" corresponds to t), and similarly assigns a relationship between, for example, heart rate HR(16:30) = 90 beats / minute ("16:30" corresponds to (t-1)) and heart rate change amount ΔHR(16:31) = -40 beats ("16:31" corresponds to t).

[0074] Step S30B: The mind-body estimation device SS creates a distribution BP(t-1) of the heart rate change ΔHR(t) for each heart rate HR(t-1) in the two-dimensional coordinate space 2ZK. The mind-body estimation device SS creates this distribution in accordance with the "t-distribution" (shown in FIG. 6).

[0075] More specifically, as shown in Figure 10, the mind-body estimation device SS creates a distribution BP(t-1) = 90 beats / min in a two-dimensional coordinate space 2ZK, for example, for a heart rate HR(t-1) = 90 beats / min, for five heart rate changes ΔHR(t) (shown by ●) corresponding to the heart rate HR(t-1) = 90 beats / min, i.e., heart rate changes ΔHR(t) = -41, -38, -37, -36, -32, by following the distribution rule "t distribution." Similarly, the mind-body estimation device SS creates a distribution BP(t-1) = 45 beats / min, 46 beats / min, . . . , 99 beats / min, 100 beats / min in the two-dimensional coordinate space 2ZK for other heart rates HR(t-1) = 45 beats / min, 46 beats / min, . . . , 99 beats / min, 100 beats / min by following the distribution law "t-distribution" as shown in Figure 10.

[0076] Step S30C: As shown in FIG. 11, the mind-body estimation device SS acquires the expected value μ(t-1) (shown by ●) in the two-dimensional coordinate space 2ZK based on the distribution BP(t-1) (shown in FIG. 10).

[0077] More specifically, as shown in Figure 11, the mind-body estimation device SS acquires the expected value μ(t-1) (shown by ●) for the distribution BP(t-1) = 90 times / minute in a two-dimensional coordinate space 2ZK, and similarly acquires the expected value μ(t-1) (shown by multiple ●) for the distribution BP(t-1) = 45 times / minute, 46 times / minute, ..., 99 times / minute, and 100 times / minute.

[0078] Step S30D: As shown in Figure 12, in a two-dimensional coordinate space 2ZK, the mind-body estimation device SS creates a virtual curve KK (shown by a dotted line) that optimally passes through the multiple expected values ​​μ(t-1) (in other words, at a position as close as possible to the position of each expected value μ(t-1)) based on the expected value μ(t-1) for each heart rate HR(t-1), in other words, multiple expected values ​​μ(t-1) (shown by multiple ●).

[0079] Step S30E: As shown in FIG. 13, the mind-body estimation device SS approximates the curve equation KH (shown in FIG. 6, shown in solid line) to the virtual curve KK (shown in dotted line) in the two-dimensional coordinate space 2ZK.

[0080] Step S40A: The mind-body estimation device SS acquires the slope b1j (shown in FIGS. 6, 14, and 15) that realizes the straight-line equation CH (shown in FIG. 6) that corresponds to the curve equation KH (shown in FIG. 6). More specifically, the mind-body estimation device SS acquires the slope b1j for realizing the straight-line equation CH that corresponds to the curve equation KH that is approximated to the virtual curve KK, using, for example, the least squares method, maximum likelihood estimation, EM algorithm, Bayesian estimation, variational Bayes, gradient descent, or backpropagation.

[0081] Specifically, the processes (information processing method) of steps S10 to S40A in FIG. 5 are executed by the processing unit SY (SS).

[0082] Here, the slope b1j shown in Figures 14 and 15 is an example to facilitate explanation and understanding; in other words, it has no relation whatsoever to the curve drawn by the virtual curve KK shown in Figures 14 and 15 (including tangents at each point on the curve, etc.).

[0083] Effect of First Embodiment As described above, the mind-body estimation system SSS of the first embodiment can acquire the slope b1j (shown in FIGS. 6, 14, and 15) of the straight-line equation CH (shown in FIG. 6) that realizes the curve equation KH (shown in FIG. 6 and shown in solid lines) that is approximated to the virtual curve KK (shown in dotted lines) that appropriately passes through multiple "expected values ​​μ(t-1)" in multiple distributions BP(t-1) that follow the distribution law "t-distribution," which indicates the relationship between the heart rate HR(t-1) and the heart rate change ΔHR(t). More specifically, the slope b1j can be acquired as an index for estimating the mind-body recovery of the subject HI1, with higher reliability than conventional recovery index information (described in Patent Document 1).

[0084] In addition to the above-described effects, the mind-body estimation system SSS of embodiment 1 can obtain the slope b1j while taking into consideration individual differences unique to the subject HI1, and can obtain the slope b1j as an index that can be used even when the subject HI1 is assumed to be performing another exercise (e.g., high-intensity exercise such as swimming) that is different from the exercise (e.g., walking described above) and that the subject HI1 would not or would not be able to actually perform. Furthermore, for example, by extending (extrapolating) the curve obtained by the curve equation KH in the direction in which the heart rate HR(t-1) increases, the amount of change in heart rate ΔHR(t) at a heart rate HR(t-1) that is not measured or cannot be measured can be estimated.

[0085] Second Embodiment A mind-body estimation system SSS according to a second embodiment will be described.

[0086] <Configuration of Second Embodiment> The mind-body estimation system SSS of the second embodiment has the same configuration as the mind-body estimation system SSS of the first embodiment (shown in FIGS. 1 to 3).

[0087] <Operation of Embodiment 2> The basic operation of the mind-body estimation system SSS of embodiment 2 is similar to the basic operation of the mind-body estimation system SSS of embodiment 1 (shown in FIG. 4). However, the detailed operation of the mind-body estimation system SSS of embodiment 2 differs from the detailed operation of the mind-body estimation system SSS of embodiment 1 (shown in FIG. 5). Unlike embodiment 1, which uses an "expected value μ" based on the distribution law "t distribution," embodiment 2 uses a "lower limit θ" based on the distribution law "SW distribution." This difference is clear from a comparison between FIG. 15 of embodiment 1 and FIG. 21 of embodiment 2. In embodiment 1, as shown in FIG. 15, the curve equation KH passes "approximately through the center" of the area showing the relationship between the amount of change in heart rate ΔHR(t) and the heart rate HR(t-1) (the area marked with a black circle in the figure), whereas in embodiment 2, as shown in FIG. 21, the curve equation KH passes "approximately through the lower limit" of the area.

[0088] FIG. 16 is a flowchart (details) showing the operation of the mind-body estimation system SSS of the second embodiment.

[0089] 17 to 21 are diagrams showing the operation of the mind-body estimation system SSS of the second embodiment.

[0090] The operation (details) of the mind-body estimation system SSS of the second embodiment will be described mainly with reference to FIG.

[0091] The mind-body estimation system SSS of embodiment 2, like the mind-body estimation system SSS of embodiment 1, essentially "approximates the relationship between the heart rate HR(t-1) and the amount of change in heart rate ΔHR(t) with a curve equation KH," and thereby "determines, by Bayesian estimation or the like, parameter values ​​(e.g., slope b1j) that minimize the error between the relationship and the curve equation KH overall." As with embodiment 1, for the sake of ease of explanation and understanding, a discussion that may seem to be somewhat detailed, such as whether or not a necessary and sufficient number of heart rates HR(t-1) exist for the same heart rate HR(t-1), will be omitted, for example, with reference to FIG. 18.

[0092] In the following, for ease of explanation and understanding, it is assumed that the subject whose mind and body is to be estimated is the subject HI1 among the subjects HI1 to HIm (shown in FIG. 1), as in the first embodiment.

[0093] Prior to step S30F: The mind-body estimation device SS performs operations similar to steps S10, S20, and S30A of embodiment 1. Specifically, the mind-body estimation device SS first calculates a heart rate HR(t), which is the heart rate HR per unit time (t) (e.g., one minute), as shown in FIG. 7 , as in step S10. The mind-body estimation device SS then calculates a heart rate change ΔHR(t), which is the difference between the heart rate HR(t-1) within a first unit time (t-1) and the heart rate HR(t) within a second unit time (t), as shown in FIG. 8 , as in step S20. Finally, the mind-body estimation device SS maps the relationship between the heart rate HR(t-1) and the heart rate change ΔHR(t) to a two-dimensional coordinate space 2ZK, as shown in FIG. 9 , as in step S30A.

[0094] Step S30F: As in step S30B of embodiment 1, mind-body estimation device SS creates a distribution BP(t-1) of heart rate change ΔHR(t) for each heart rate HR(t-1) in two-dimensional coordinate space 2ZK. Unlike step S30B of embodiment 1, which follows the distribution rule "t-distribution," mind-body estimation device SS creates this distribution according to the distribution rule "SW-distribution" as shown in FIG. 18. Even when using the distribution rule "SW-distribution," mind-body estimation device SS creates a distribution BP(t-1)=45 beats / min, 46 beats / min, 99 beats / min, and 100 beats / min, as shown in FIG. 18, in the same way as in embodiment 1 (shown in FIG. 10) which uses the distribution rule "t-distribution."

[0095] Step S30G: The mind-body estimation device SS acquires a lower limit value θ(t-1) in the two-dimensional coordinate space 2ZK based on the distribution BP(t-1) (shown in FIG. 10) in the same manner as in step S30C in embodiment 1, but differs from step S30C in embodiment 1 in that it acquires a lower limit value θ(t-1). As a result, as shown in FIG. 19, the mind-body estimation device SS acquires lower limit values ​​θ(t-1) (shown by multiple ●) for the distribution BP(t-1)=45 times / minute, 46 times / minute, . . . , 99 times / minute, and 100 times / minute.

[0096] Step S30H: As in step S30D of embodiment 1, the mind-body estimation device SS creates a virtual curve KK (shown by a dotted line) that optimally passes through multiple lower limit values ​​θ(t-1) (shown by multiple ●) in a two-dimensional coordinate space 2ZK, as shown in Figure 20.

[0097] After step S30H: The mind-body estimation device SS performs the same operations as steps S30E and S40A in embodiment 1. More specifically, as in step S30E, the mind-body estimation device SS first approximates a curve equation KH (shown in FIG. 17, shown by a solid line) to a virtual curve KK (shown by a dotted line) in a two-dimensional coordinate space 2ZK, as shown in FIG. 13. Next, as in step S40A, the mind-body estimation device SS obtains a slope b1j (shown in FIGS. 17, 20, and 21) that realizes a straight-line equation CH (shown in FIG. 17) corresponding to the curve equation KH.

[0098] Specifically, the processes (information processing method) of steps S30F to S30H in FIG. 16 are executed by the processing unit SY (SS).

[0099] Effect of Second Embodiment As described above, in the mind-body estimation system SSS of the second embodiment, even when a "lower limit value θ" based on the distribution law "SW distribution" is used, it is possible to obtain the slope b1j of the straight-line equation CH (shown in FIG. 17) that realizes the curve equation KH (shown in FIG. 17), just as in the first embodiment, which uses an "expected value μ" based on the distribution law "t distribution." More specifically, it is possible to obtain a slope b1j that is more reliable as an index for estimating the mind-body recovery of the subject HI1 than conventional recovery index information (described in Patent Document 1).

[0100] In addition to the above effects, the mind-body estimation system SSS of embodiment 2 can obtain the above slope b1j while taking into consideration individual differences unique to the subject HI1, as with the mind-body estimation system SSS of embodiment 1, and can obtain the above slope b1j as an index that can be used even when it is assumed that the subject HI1 performs another exercise (for example, a high-intensity exercise such as swimming) that is different from the exercise (such as the walking described above) and that the subject HI1 would not actually perform or would not actually be able to perform. Otherwise, the effects of embodiment 2 are similar to those of embodiment 1.

[0101] Supplementary Explanation of First and Second Embodiments In the first and second embodiments, the distribution laws are "t distribution" and "SW distribution." Instead of following the "t distribution" and "SW distribution," other distribution laws (for example, normal distribution (Gaussian distribution)) or exponentially modified Gaussian distribution (ex-Gaussian distribution)) may be followed.

[0102] In the first and second embodiments, as described above, the slope b1j in the linear equation CH (shown in FIGS. 6 and 17) is acquired as an index for estimating the physical and mental recovery of subject HI1. Instead of acquiring the slope b1j, for example, the fluctuation range A in the curve equation KH (shown in FIGS. 6 and 17) may be acquired. In other words, the mind and body estimation device SS acquires one or more of the slope b1j, the fluctuation range A, and the intercept b0j as an index for estimating the physical and mental recovery of subject HI1 (mental and physical recovery ability information).

[0103] Third Embodiment A mind-body estimation system SSS according to a third embodiment will be described.

[0104] <Configuration of Third Embodiment> The mind-body estimation system SSS of the third embodiment has the same configuration as the mind-body estimation system SSS of the first embodiment (shown in FIGS. 1 to 3).

[0105] 22 and 23 are diagrams showing the operation of embodiment 3. In embodiment 3, in the mind-body estimation system SSS, the mind-body estimation device SS (processing unit SY(SS)) estimates the degree of mind-body recovery of the subject based on mind-body recovery ability information (e.g., slope b1j). An example is given below.

[0106] As shown in Fig. 22, the mind-body estimation system SSS of embodiment 3 observes the slope b1j (shown in, for example, Figs. 6, 15, 17, and 21) over time for the same subject HI, and more specifically, observes the slope b1j for "May," "June," and "July" for "subject HI1." This makes it possible to estimate changes over time in the mind-body recovery of subject HI1 (for example, whether the progress of recovery is stable or unstable, or whether the progress of recovery is gradually accelerating or gradually slowing down).

[0107] In contrast to the above, the mind-body estimation system SSS of embodiment 3 simultaneously observes the slopes b1j for multiple subjects HI, and more specifically, observes the slopes b1j for "May" for, for example, "subject HI1," "subject HI2," and "subject HI3," as shown in Fig. 23. This makes it possible to estimate individual differences in the mind-body recovery of subjects HI1, HI2, and HI3 (for example, whether the progress of recovery among the three subjects is similar, or whether only subject HI1 is recovering slowly, or whether only subject HI3 is recovering quickly).

[0108] <Hardware Configuration of the Embodiments> FIG. 24 shows the hardware configuration of the mind-body estimation system SSS of the first to third embodiments.

[0109] To perform the above-described functions, the mind-body estimation system SSS of the first to third embodiments includes a processing circuit SYO, as shown in FIG. 24, and may further include an input circuit NYU and an output circuit SYU as necessary.

[0110] The processing circuit SYO is dedicated hardware that realizes the functions of the wearable terminal WT and the processing units SY(WT) and SY(SS) of the mind-body estimation device SS (shown in FIGS. 2 and 3).

[0111] The processing circuit SYO is, for example, a single circuit, a complex circuit, a programmed processor, a parallel programmed processor, an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), or a combination thereof.

[0112] The input circuit NYU and the output circuit SYU exchange inputs and outputs related to the operation of the processing circuit SYO with, for example, the wearable terminal WT and the outside of the mind-body estimation device SS.

[0113] <Hardware Configuration Based on Software Realization of the Embodiments> FIG. 25 shows the hardware configuration based on software realization of the mind-body estimation system SSS of the first to third embodiments.

[0114] As shown in FIG. 25, the mind-body estimation system SSS of the first to third embodiments includes a processor PRO and a memory circuit KIO, and may further include an input circuit NYU and an output circuit SYU as required.

[0115] The processor PRO is a CPU (also called a central processing unit, processing device, arithmetic unit, microprocessor, microcomputer, or DSP (Digital Signal Processing)) that executes programs. The processor PRO realizes the functions of the wearable terminal WT and the processing unit SY(WT) and processing unit SY(SS) of the mind-body estimation device SS (shown in FIGS. 2 and 3).

[0116] The processor PRO realizes the above-mentioned functions by software, firmware, or a combination of software and firmware. The software and firmware are written as programs and stored in the memory circuit KIO.

[0117] The processor PRO realizes the above-mentioned functions by reading and executing the above-mentioned programs from the memory circuit KIO. The above-mentioned programs can also be said to cause a computer to execute the procedures and methods of the wearable terminal WT and the processing unit SY(WT) and processing unit SY(SS) of the mind-body estimation device SS.

[0118] Here, the memory circuit KIO is, for example, a non-volatile or volatile semiconductor memory such as RAM (Random Access Memory), ROM (Read Only Memory), flash memory, EPROM (Erasable Programmable Read Only Memory), EEPROM (Electrically Erasable Programmable Read-Only Memory), etc., as well as a magnetic disk, a flexible disk, an optical disk, a compact disk, a mini disk, a DVD (Digital Versatile Disc), etc.

[0119] Of the functions of the wearable terminal WT, the processing unit SY(WT) of the mind-body estimation device SS, and the processing unit SY(SS), some of the functions may be realized by a processing circuit SYO (shown in Figure 24), while other functions may be realized by a processor PRO (shown in Figure 25).

[0120] As described above, the functions of the wearable terminal WT, the processing unit SY(WT) of the mind-body estimation device SS, and the processing unit SY(SS) can be realized by hardware, software, firmware, or a combination of these.

[0121] The input circuit NYU and the output circuit SYU exchange inputs and outputs related to the operation of the processor PRO with, for example, the wearable terminal WT and the outside of the mind-body estimation device SS.

[0122] Summary of Embodiments 1 to 3 As described above with reference to FIGS. 1 to 25 , the processing unit SY (SS) calculates physical and mental recovery capacity information based on the subject's heart rate change ΔHR(t) and exercise heart rate HR(t-1). The physical and mental recovery capacity information indicates the degree of speed of recovery of the subject's heart rate from the end of exercise. In other words, the physical and mental recovery capacity information indicates the degree of speed of recovery of the subject's heart rate from the end of exercise to rest. Specifically, the physical and mental recovery capacity information includes parameter values ​​PV that realize a model function MF that approximates the relationship between the subject's heart rate change ΔHR(t) and exercise heart rate HR(t-1). The model function MF includes a linear equation CH. Preferably, the parameter value PV indicates the slope b1j of the line represented by the linear equation CH. The greater the absolute value of the slope of the linear equation CH, the greater the speed of recovery of the subject's heart rate.

[0123] Specifically, the processing unit SY(SS) calculates a heart rate change ΔHR(t), which is the difference between the exercise heart rate HR(t-1), which is the heart rate during a first unit of time while the subject is exercising, and the resting heart rate HR(t), which is the heart rate during a second unit of time following the first unit of time while the subject is at rest. The processing unit SY(SS) then approximates the relationship between the heart rate change ΔHR(t) and the exercise heart rate HR(t-1) using a model function MF. Furthermore, the processing unit SY(SS) acquires parameter values ​​PV that realize the model function MF as mental and physical recovery capacity information. The memory unit KI(SS) stores the mental and physical recovery capacity information.

[0124] More specifically, the model function MF includes a curve equation KH that approximates the relationship between the heart rate change ΔHR(t) and the exercise heart rate HR(t-1). The parameter value PV indicates the slope b1j of the line expressed by the linear equation CH that constitutes the input variables "zi, j" of the curve equation KH. The curve equation KH is a function fitted to a virtual curve KK (FIGS. 12 and 20). The virtual curve KK is a virtual curve obtained based on representative values ​​of the heart rate change ΔHR(t) obtained for each actual measurement of the exercise heart rate HR(t-1). Each representative value corresponds to the actual measurement of the exercise heart rate HR(t-1) and is obtained based on a distribution law (e.g., t-distribution or SW distribution) from the distribution of the actual measurement of the heart rate change ΔHR(t). The representative value is, for example, the expected value μ obtained from the t-distribution (see FIG. 6) or the lower limit value θ obtained from the SW distribution (see FIG. 17).

[0125] The parameter value PV serving as the mind-body recovery ability information may include, for example, one or more of the slope b1j, the fluctuation range A of the curve equation KH, and the intercept b0j of the straight line equation CH.

[0126] Next, an information processing system according to a fourth embodiment of the present disclosure will be described.

[0127] Fourth Embodiment An information processing system according to a fourth embodiment of the present disclosure uses a learning model to estimate a subject's biological-related state information from at least the subject's mental and physical recovery ability information. The subject's biological-related state information includes information representing the subject's mental and physical state, or information representing an environmental state that affects the subject's mental and physical state. The learning model is constructed by performing learning using a training dataset. The training dataset includes at least the training subject's mental and physical recovery ability information and the training subject's biological-related state information. The training subject's biological-related state information includes information representing the training subject's mental and physical state, or information representing an environmental state that affects the subject's mental and physical state.

[0128] The subject corresponds to an example of a "subject" in the present disclosure. The training subject corresponds to an example of a "training subject" in the present disclosure.

[0129] The learning subject's mental and physical recovery ability information and the subject's mental and physical recovery ability information according to the fourth embodiment are the mental and physical recovery ability information described in the first to third embodiments. In the first to third embodiments, the mind and body estimation system SSS calculated more reliable mental and physical recovery ability information than conventional methods for estimating the subject's mental and physical recovery. In the fourth embodiment, the information processing system effectively utilizes more reliable mental and physical recovery ability information than conventional methods for estimating the subject's mental and physical recovery to obtain highly reliable biological-related state information.

[0130] An information processing system 1 according to a fourth embodiment of the present disclosure will be described with reference to FIGS. 26 to 39 . FIG. 26 is a block diagram showing an example configuration of the information processing system 1. As shown in FIG. 26 , the information processing system 1 includes a training data creation device 2, a learning device 3, and an estimation device 4. Each of the training data creation device 2, the learning device 3, and the estimation device 4 is a computer. The estimation device 4 includes a preprocessing unit 41, an estimation unit 42, and a learning model TM1. The estimation device 4 may further include a postprocessing unit 43.

[0131] The learning data creation device 2 acquires biological state raw data A1 and correct answer information Z1 of the learning subject. The biological state raw data A1 is raw data indicating the biological state of the learning subject. A learning subject is a subject from whom raw data for learning is acquired. A learning subject is typically a human. The biological state raw data A1 includes vital raw data and behavioral raw data of the learning subject. The vital raw data is raw data indicating the vital signs of a living organism. The behavioral raw data is raw data indicating the behavior of a living organism. The correct answer information Z1 indicates biological-related state information.

[0132] The training data creation device 2 creates a training data set F1 based on the biological state raw data A1 and the correct answer information Z1. The training data set F1 includes feature information G1 and a correct answer label B1 of the training subject. The feature information G1 includes at least the physical and mental recovery ability information D1 of the training subject. The feature information G1 is an explanatory variable. The feature information G1 is created based on the biological state raw data A1. The correct answer label B1 is a target variable. The correct answer label B1 is created based on the correct answer information Z1. The correct answer label B1 includes the biological-related state information M1.

[0133] The learning device 3 executes a machine learning algorithm to perform learning using a learning dataset F1. The learning device 3 generates a learning model TM1 by repeatedly performing learning using a plurality of learning datasets F1. The learning device 3 performs supervised learning. The learning model TM1 outputs output information L2 when input information K2 is input. Typically, the learning model TM1 is a trained model. Furthermore, the learning model TM1 is a computer program.

[0134] On the other hand, the estimation device 4 uses the learning model TM1 to estimate the biological-related state information M2 of the subject. The subject is typically a human being.

[0135] Specifically, the preprocessing unit 41 performs preprocessing on the subject's biological condition raw data A2 and generates input information K2 as a result of the preprocessing. The biological condition raw data A2 is raw data indicating the subject's biological condition. The biological condition raw data A2 includes the subject's vital raw data and behavioral raw data. The vital raw data is raw data indicating the vital signs of the subject. The behavioral raw data is raw data indicating the behavior of the subject.

[0136] The preprocessing unit 41 includes at least a mind-body estimation unit 411. The mind-body estimation unit 411 calculates the subject's mind-body recovery ability information D2 based on the subject's raw vital data and raw behavioral data. Therefore, the input information K2 includes at least the subject's mind-body recovery ability information D2.

[0137] Specifically, the mind and body estimation unit 411 calculates the mind and body recovery ability information D2 in the same manner as the mind and body estimation device SS in any of the first to third embodiments. In other words, the mind and body estimation unit 411 has the same functions as the mind and body estimation device SS (processing unit SY(SS)). Note that the pre-processing unit 41 may acquire the subject's mind and body recovery ability information D2 from the mind and body estimation device SS without providing the mind and body estimation unit 411. In this case, for example, the information processing system 1 may include the mind and body estimation device SS.

[0138] The estimation unit 42 inputs input information K2 to the learning model TM1. As a result, the learning model TM1 outputs output information L2. The input information K2 is an explanatory variable. The output information L2 is a target variable. The output information L2 includes the subject's biological-related state information M2. The estimation unit 42 acquires the output information L2 from the learning model TM1. The post-processing unit 43 may perform post-processing on the output information L2 and generate output information N2, which is the result of the post-processing. The output information N2 includes the biological-related state information M3 after post-processing.

[0139] Hereinafter, when there is no need to distinguish between the biological-related condition information M2 and M3, the biological-related condition information M2 and M3 of the subject may be referred to as "biological-related condition information MX."

[0140] In the fourth embodiment, the estimation device 4 inputs at least the subject's physical and mental recovery ability information into the learning model TM1, thereby outputting the biological-related state information MX with high estimation accuracy. In this case, for example, the estimation device 4 does not require diagnosis and evaluation by a medical professional, testing using a medical device such as a medical image diagnostic device, or collection and testing of bodily fluids such as blood. The medical professional may be, for example, a doctor or nurse. In addition, the estimation device 4 automatically and continuously acquires the biological state raw data A2 from a wearable device (e.g., a wearable terminal WT) and / or a mobile terminal. Therefore, the biological-related state information MX can be continuously obtained while reducing the burden on the subject.

[0141] Next, the correlation between the learning subject's physical and mental recovery ability information D1 and the organism-related state information M1 in the learning dataset F1 will be described. FIG. 27 is a diagram showing an example of the learning dataset F1. As shown in FIG. 27, the learning dataset F1 includes feature information G1 and a correct label B1 of the learning subject. The feature information G1 includes at least the physical and mental recovery ability information D1. Meanwhile, the correct label B1 includes the organism-related state information M1. In the correct label B1, the organism-related state information M1 is, for example, an actual measurement value (e.g., a diagnosis by a doctor, an examination result using a medical device, a biometric test result using a biomarker or the like, or an evaluation result using various indices).

[0142] Specifically, the biological state information M1 includes information M100 representing the learning subject's physical and mental state (hereinafter referred to as "physical and mental state information M100"), or information M110 representing the state of the environment that affects the learning subject's physical and mental state (hereinafter referred to as "physical and mental environment information M110"). Meanwhile, the physical and mental recovery capacity information D1 is information indicating the degree of speed of the learning subject's physical and mental recovery from the end of exercise. In other words, the physical and mental recovery capacity information D1 indicates the degree of speed of the learning subject's physical and mental recovery from the end of exercise until the time of rest. Typically, the physical and mental recovery capacity information D1 indicates the degree of speed of the learning subject's heart rate recovery from the end of exercise. In other words, the physical and mental recovery capacity information D1 indicates the degree of speed of the learning subject's heart rate recovery from the end of exercise until the time of rest. Hereinafter, the speed of physical and mental recovery may be referred to as "physical and mental recovery speed," and the speed of heart rate recovery may be referred to as "heart rate recovery speed."

[0143] The mental and physical recovery rate can be affected by the state of various mental and physical functions. In particular, the heart rate recovery rate can be more affected by the state of various mental and physical functions. For example, if the mental and physical recovery rate (especially the heart rate recovery rate) is relatively low, there is a possibility that the state of mental and physical functions is poor or that there is a disorder in the mental and physical functions. Therefore, it can be inferred that there is a correlation between the mental and physical recovery capacity information D1 and the mental and physical condition information M100.

[0144] Furthermore, the mental and physical recovery rate may be indirectly affected by the state of the living body's surrounding environment. In particular, the heart rate recovery rate may be indirectly affected by the state of the living body's surrounding environment. For example, if the state of the living body's surrounding environment is poor, it will have a negative impact on mental and physical function. As a result, the mental and physical recovery rate (especially the heart rate recovery rate) may decrease. Therefore, it can be inferred that there is a correlation between the mental and physical recovery ability information D1 and the mental and physical environment information M110.

[0145] Therefore, according to the fourth embodiment, the learning model TM1 constructed by learning using the learning dataset F1 (mental and physical recovery capacity information D1 and organism-related state information M1) can output (generate) the subject's biological-related state information M2 (FIG. 28) when the subject's mental and physical recovery capacity information D2 (FIG. 28) is input. In other words, the estimation device 4 can estimate the subject's biological-related state information M2 by using the learning model TM1.

[0146] In particular, the mental and physical recovery capacity information D1 and D2 are indices for evaluating the dynamic process (response capacity) of "recovery from stress." For example, heart rate can be affected by the state of various mental and physical functions. However, heart rate itself is not an index for evaluating the dynamic process of "recovery from stress." By learning and inputting the mental and physical recovery capacity information D1 and D2 that evaluates the dynamic process as in embodiment 4, the learning model TM1 can output organism-related state information M2 that more sensitively reflects the state of various mental and physical functions and / or the state of the organism's surrounding environment.

[0147] More preferably, the physical and mental recovery capacity information D1 is calculated based on the learning subject's heart rate change and exercise heart rate. Specifically, the physical and mental recovery capacity information D1 includes parameter values ​​PV that realize a model function MF that approximates the relationship between the learning subject's heart rate change and exercise heart rate. The parameter value PV indicates the slope b1j of the line expressed by the linear equation CH that constitutes the model function MF. The greater the absolute value of the slope b1j, the greater the heart rate recovery speed.

[0148] The heart rate change indicates the difference between the learning subject's heart rate during exercise and resting heart rate. In other words, the heart rate change indicates the difference between the exercise heart rate, which is the heart rate during a first unit of time while the learning subject is exercising, and the resting heart rate, which is the heart rate during a second unit of time following the first unit of time while the learning subject is at rest. Thus, the mental and physical recovery capacity information D1 is an index that quantifies the rate at which the heart rate, which increased during exercise, decreases after exercise stops (i.e., the heart rate recovery rate). Therefore, the mental and physical recovery capacity information D1 reflects the "mental and physical state" and / or the "environmental state that affects the mental and physical state" with a high degree of reliability. Therefore, according to the fourth embodiment, the estimation device 4 can estimate the organism-related state information M2 with high accuracy by using the learning model TM1.

[0149] The mind-body recovery capability information D1 may include information obtained by performing a predetermined process on the parameter values ​​PV that realize the model function MF. The predetermined process may be, for example, a statistical process described later, and is not particularly limited.

[0150] Here, for example, the mental and physical condition information M100 of the living body-related condition information M1 may include information representing the physical condition (hereinafter referred to as "physical condition information"). Because the mental and physical recovery capacity information D1 is an index that quantifies the heart rate recovery rate, the mental and physical recovery capacity information D1 reflects the physical condition (e.g., worsening or improvement of physical illnesses and symptoms) with high sensitivity.

[0151] Physical condition information may include, for example, information regarding diseases or symptoms of the circulatory system (hereinafter referred to as "circulatory system information"), information regarding diseases or symptoms of the respiratory function system (hereinafter referred to as "respiratory function system information"), information regarding diseases or symptoms of the biocontrol system (hereinafter referred to as "biocontrol system information"), information regarding subjective symptoms related to the physical condition (hereinafter referred to as "subjective symptom information"), or information representing the physical condition that can be indicated by the results of a biopsy (hereinafter referred to as "biopsy information").

[0152] The cardiovascular information may include, for example, information describing conditions related to heart failure, coronary artery disease, arrhythmia, hypertension, or cardiac autonomic neuropathy.

[0153] For example, in the case of heart failure, the heart rate remains high and sympathetic nervous activity becomes chronic, which slows down the heart rate recovery rate. Therefore, it can be inferred that there is a correlation between the mind and body recovery ability information D1 and the onset of heart failure.

[0154] For example, in the case of coronary artery disease, a decrease in the heart rate regulating reflex and delayed vagus nerve activity may cause a delay in heart rate recovery. As a result, the heart rate recovery speed decreases. Therefore, it can be inferred that there is a correlation between the mind and body recovery ability information D1 and coronary artery disease.

[0155] For example, in the case of arrhythmia, especially tachyarrhythmia, heart rate variability is poor, making it difficult to regulate the heart rate after exercise. This delays heart rate recovery, resulting in a slower heart rate recovery rate. Therefore, it can be inferred that there is a correlation between mind and body recovery ability information D1 and arrhythmia.

[0156] For example, in the case of hypertension, it is thought that heart rate recovery slows due to arterial stiffness and increased sympathetic nervous activity. As a result, the heart rate recovery speed decreases. Therefore, it can be inferred that there is a correlation between mind and body recovery ability information D1 and hypertension.

[0157] For example, cardiac autonomic neuropathy, particularly diabetic cardiac autonomic neuropathy, can cause a decrease in heart rate recovery rate. Therefore, it can be inferred that there is a correlation between the mind and body recovery ability information D1 and cardiac autonomic neuropathy.

[0158] The respiratory function system information may include, for example, information representing a condition related to COPD (Chronic Obstructive Pulmonary Disease), asthma, pulmonary fibrosis, ILD (Interstitial Lung Disease), or high altitude acclimatization failure.

[0159] For example, in the case of COPD, shortness of breath maintains the heart rate, which is thought to remain high even after exercise. As a result, the heart rate recovery speed decreases. Therefore, it can be inferred that there is a correlation between the mind-body recovery ability information D1 and COPD.

[0160] For example, in the case of asthma, prolonged airway resistance and sympathetic nerve stimulation prolong the heart rate, which is thought to cause the heart rate to remain high even after exercise. As a result, the heart rate recovery rate decreases. Therefore, it can be inferred that there is a correlation between the mind and body recovery ability information D1 and asthma.

[0161] For example, in the case of pulmonary fibrosis or ILD, a persistently high heart rate may be observed even after exercise due to a decrease in oxygen uptake efficiency. As a result, the heart rate recovery rate decreases. Therefore, it can be inferred that there is a correlation between the mental and physical recovery ability information D1 and pulmonary fibrosis or ILD.

[0162] For example, in the case of high altitude insufficiency, compensatory tachycardia persists due to a lack of oxygen, which is thought to delay heart rate recovery. As a result, the heart rate recovery speed decreases. Therefore, it can be inferred that there is a correlation between the mind and body recovery ability information D1 and high altitude insufficiency.

[0163] The biocontrol system information may include, for example, information representing a condition related to an autonomic neuropathy or an endocrine disorder. As an example, the biocontrol system information may include information representing a condition related to CAN (cardiac autonomic neuropathy), chronic stress, overwork, autonomic imbalance, hyperthyroidism, an adrenal tumor (e.g., pheochromocytoma), a sleep disorder, or SAS (sleeping apnea syndrome).

[0164] For example, in the case of chronic stress, overwork, or autonomic imbalance, the reactivation of the parasympathetic nervous system is delayed, resulting in insufficient sympathetic convergence, which may cause the heart rate to remain high after exercise. As a result, the heart rate recovery rate decreases. Therefore, it can be inferred that there is a correlation between the mind-body recovery ability information D1 and chronic stress, overwork, or autonomic imbalance.

[0165] For example, in the case of hyperthyroidism, the basal heart rate increases and the heart rate recovery is delayed. As a result, the heart rate recovery speed decreases. Therefore, it can be inferred that there is a correlation between the mental and physical recovery ability information D1 and hyperthyroidism.

[0166] For example, in the case of an adrenal tumor, excessive secretion of noradrenaline is thought to cause tachycardia and slowed heart rate recovery. As a result, the heart rate recovery speed decreases. Therefore, it can be inferred that there is a correlation between the mind-body recovery ability information D1 and an adrenal tumor.

[0167] For example, in the case of sleep disorders or SAS, nighttime sympathetic nervous tension may persist into the daytime, hindering heart rate recovery after exercise. As a result, heart rate recovery speed decreases. Therefore, it can be inferred that there is a correlation between the mind-body recovery ability information D1 and sleep disorders and SAS.

[0168] The subjective symptom information may include, for example, information representing subjective symptoms or pre-symptomatic findings related to a physical condition. For example, the subjective symptom information may include information representing palpitations, shortness of breath, shortness of breath (dyspnea), dizziness, unsteadiness, or prolonged fatigue.

[0169] For example, when the heart rate recovery rate is reduced, the respiratory rate remains high even after exercise, which can lead to subjective symptoms such as palpitations, shortness of breath, breathlessness (dyspnea), dizziness, unsteadiness, or prolonged fatigue. Therefore, it can be inferred that there is a correlation between the mind and body recovery ability information D1 and the subjective symptom information.

[0170] The biopsy information may include, for example, information that indicates a physical condition that can be indicated by the results of a body fluid test, a test performed by a medical image diagnostic device, or a respiratory function test.

[0171] The test results for body fluids (e.g., urine) are, for example, test results for urinary cortisol or urinary albumin. For example, urinary cortisol may be elevated due to chronic stress or autonomic nervous tension, which may delay heart rate recovery due to the action of the parasympathetic nervous system, thereby reducing the heart rate recovery rate. For example, microalbuminuria is considered an early marker of autonomic nervous system disorders and cardiovascular risk, and autonomic nervous system disorders and cardiovascular diseases may reduce the heart rate recovery rate. From these results, it can be inferred that there is a correlation between the mental and physical recovery ability information D1 and information indicating a physical condition that can be indicated by the test results for body fluids (e.g., elevated urinary cortisol or microalbuminuria).

[0172] The examination results obtained by the medical imaging diagnostic device include, for example, chest X-ray examination results (e.g., increased cardiothoracic ratio, pulmonary hyperinflation, or pulmonary fibrosis) or abdominal ultrasound examination results (e.g., hepatic fatty changes). For example, an increased cardiothoracic ratio (cardiomegaly) may contribute to the detection of potential heart failure or myocardial damage. Heart failure or myocardial damage may also result in a decrease in cardiac recovery rate. Furthermore, for example, pulmonary hyperinflation or pulmonary fibrosis is used to screen for COPD or ILD, which may affect cardiac recovery after exercise. As can be seen from these findings, chest X-ray examination results can indirectly affect the mental and physical recovery capacity information D1. Therefore, it can be inferred that there is a correlation between the mental and physical recovery capacity information D1 and information indicating a physical condition that can be indicated by chest X-ray examination results (e.g., increased cardiothoracic ratio, pulmonary hyperinflation, or pulmonary fibrosis). Furthermore, for example, increases in body weight and BMI due to hepatic fatty changes are thought to maintain high heart rates after exercise in the context of metabolic syndrome. As a result, heart rate recovery speed decreases. Therefore, it can be inferred that there is a correlation between the mental and physical recovery ability information D1 and information indicating a physical condition (e.g., hepatic fatty changes) that can be indicated by the results of abdominal ultrasound examination.

[0173] Respiratory function test results include, for example, exhaled gas analysis results, breathing pattern test results, pulmonary function test results, or blood gas analysis results. For example, the concentration of nitric oxide (FeNO) in exhaled air is associated with airway inflammation. Increased respiratory load associated with airway inflammation may delay cardiac recovery. As a result, cardiac recovery speed decreases. Furthermore, changes in respiratory pattern (e.g., respiratory rate, respiratory variability, or ventilation efficiency), such as respiratory load or gas exchange efficiency, are thought to affect cardiac recovery speed. Similarly, pulmonary function such as vital capacity is thought to affect cardiac recovery speed. Furthermore, for example, a decrease in arterial oxygen partial pressure indicates insufficient oxygen supply, which delays cardiac recovery after exercise and therefore decreases cardiac recovery speed. From these results, it can be inferred that there is a correlation between the mental and physical recovery ability information D1 and information indicating physical conditions that can be indicated by respiratory function test results (e.g., abnormalities in exhaled nitric oxide concentration, abnormal breathing pattern, abnormal lung function, abnormal blood oxygen content, etc.).

[0174] The respiratory function test results include, for example, vital capacity (VC), peak expiratory flow (PEF), arterial oxygen partial pressure (PaO ２ : Partial Pressure of Oxygen), VE / VO ２ (Ventilation volume / oxygen intake), VE / VCO ２ (ventilation volume / carbon dioxide output), VO ２ / WR (oxygen intake / work), or peak oxygen intake (peak VO ２ ) and other indicators.

[0175] The respiratory function test may be the result of a test performed during an exercise test, such as a timed walking test.

[0176] The physical condition information may include, for example, information on accompanying symptoms of a physical disease or condition, or information on side effects or complications of treatment for a physical disease or condition.

[0177] Here, for example, the mental and physical condition information M100 of the biological-related condition information M1 may include information representing a mental state (hereinafter referred to as "mental state information"). Because the mental and physical recovery capacity information D1 is an index that quantitatively indicates the heart rate recovery rate, the mental and physical recovery capacity information D1 reflects the mental state (e.g., worsening or improvement of mental illness and symptoms) with high sensitivity.

[0178] The mental state information may include, for example, information representing a state related to mental illness (hereinafter referred to as "mental illness information"), or information representing a state related to mental stress (hereinafter referred to as "mental stress information").

[0179] The mental illness information may include, for example, information representing a condition related to anxiety disorder, panic disorder, depression, trauma, or post-traumatic stress disorder (PTSD).

[0180] For example, anxiety disorders and panic disorders are illnesses that involve constant tension and hypervigilance, so hypervigilance can continue even after exercise, causing the heart rate to not decrease. As a result, the heart rate recovery rate decreases. Therefore, it can be inferred that there is a correlation between the mind-body recovery ability information D1 and anxiety disorders or panic disorders.

[0181] For example, in the case of depression, the autonomic nervous system response slows down, which is thought to cause a delay in heart rate recovery after exercise. As a result, the heart rate recovery speed decreases. Therefore, it can be inferred that there is a correlation between the mind-body recovery ability information D1 and depression.

[0182] For example, trauma and PTSD are diseases in which hyperarousal is one of the main symptoms, and since the heart rate tends to remain high in response to stressful stimuli, it is thought that the heart rate recovery rate after exercise may also decrease. Therefore, it can be inferred that there is a correlation between the mind-body recovery ability information D1 and trauma or PTSD.

[0183] The mental load information may include, for example, information that represents a mental state that can be indicated by the results of a test regarding the mental state.

[0184] The test results relating to the mental state are, for example, test results of oxidative stress, immunity, or cytokines, or test results based on a psychological stress scale, a fatigue scale, or a lethargy scale.

[0185] For example, heart rate recovery rate decreases due to disturbances in the autonomic nervous system. A state in which heart rate recovery rate decreases causes shortness of breath and mental stress, such as fatigue and psychological stress related to exercise. Furthermore, mental stress can affect the results of mental state tests. As can be seen from this, it can be inferred that the mind-body recovery capacity information D1 is indirectly correlated with the results of mental state tests. In other words, it can be inferred that there is a correlation between the mind-body recovery capacity information D1 and information that indicates the mental state that can be indicated by the results of mental state tests (e.g., the mental state based on oxidative stress, immunity, cytokines, psychological stress scale, fatigue scale, or fatigue scale).

[0186] The mental state information may include, for example, information about accompanying symptoms of a mental disorder or condition, or information about side effects or complications of treatment for a mental disorder or condition.

[0187] Here, for example, the mental and physical condition information M100 of the living body-related condition information M1 may include information regarding the intake of a specific component (hereinafter referred to as "specific component intake information").

[0188] The specific component intake information may include, for example, information regarding the intake of medicines (hereinafter referred to as "medicine intake information") or information regarding the intake of luxury goods (hereinafter referred to as "luxury goods intake information").

[0189] Medication intake information may include, for example, information regarding intake of beta-agonists (bronchodilators), anticholinergics, or antihistamines.

[0190] For example, because β-stimulants induce sympathetic nerve activity, it is thought that sympathetic nerve activation continues even after exercise, maintaining a high heart rate. As a result, the heart rate recovery rate decreases. Therefore, it can be inferred that there is a correlation between the mind-body recovery ability information D1 and the use of β-stimulants.

[0191] For example, taking an anticholinergic or antihistamine drug is thought to delay the recovery of heart rate by the parasympathetic nervous system through vagus nerve block, resulting in a slower heart rate recovery rate. Therefore, it can be inferred that there is a correlation between the mind-body recovery ability information D1 and the use of an anticholinergic or antihistamine drug.

[0192] The recreational substance intake information may include, for example, information regarding the intake of caffeine, energy drinks, or nicotine.

[0193] For example, the intake of caffeine or energy drinks is thought to activate the sympathetic nervous system, causing a delay in heart rate recovery. As a result, the heart rate recovery speed decreases. Therefore, it can be inferred that there is a correlation between the mind and body recovery ability information D1 and the intake of caffeine or energy drinks.

[0194] For example, nicotine intake through smoking is thought to induce a decrease in pulmonary ventilation efficiency and sympathetic nerve activity, resulting in a delay in heart rate recovery. As a result, the heart rate recovery speed decreases. Therefore, it can be inferred that there is a correlation between the mind and body recovery ability information D1 and nicotine intake (smoking).

[0195] The specific component intake information may include, for example, information regarding side effects from the intake of medicines or luxury items.

[0196] Here, for example, the mental and physical condition information M100 of the organism-related condition information M1 may include information on the quality of life (QOL) of the learning subject (hereinafter referred to as "QOL information"). The QOL information indicates, for example, the physical condition, mental condition, or quality of life related to the intake of a specific component.

[0197] For example, it is believed that quality of life is directly affected by symptoms such as shortness of breath and fatigue that may occur due to a decrease in heart rate recovery rate. Furthermore, it is believed that quality of life is affected by the occurrence or worsening of the above-mentioned symptoms, etc., related to physical or mental state, which may affect heart rate recovery rate. Furthermore, it is believed that quality of life is affected by discontinuing the administration of the above-mentioned medicines, which may affect heart rate recovery rate, related to the intake of specific ingredients. As can be understood from these facts, it can be inferred that there is a correlation between the physical and mental recovery ability information D1 and quality of life related to physical state, mental state, or intake of specific ingredients.

[0198] Here, for example, the mental and physical environment information M110 of the organism-related state information M1 may include information on atmospheric abnormalities (hereinafter referred to as "atmospheric abnormality information") or information on a mentally stressful environment (hereinafter referred to as "stressful environment information"). The mental and physical environment information M110 is expressed, for example, by information indicating whether an abnormality in the environmental state is "present" or "absent," information indicating the degree of the environmental state in stages, or information indicating the environmental state by a score.

[0199] The atmosphere abnormality information may include, for example, information regarding a low-oxygen environment, a hot environment, harmful gases, or an abnormality in air components.

[0200] For example, in a high-altitude environment, which may be a hypoxic environment, a drop in oxygen partial pressure causes a sustained high heart rate after exercise. As a result, the heart rate recovery rate decreases. Therefore, it can be inferred that there is a correlation between the mind-body recovery ability information D1 and the fact that the learning subject is staying in a high-altitude environment.

[0201] For example, in low-oxygen environments such as tunnel excavation sites and underground construction sites, the oxygen concentration may drop below 19.5% due to insufficient ventilation. In such situations, the heart rate remains high after exercise due to the need to supply oxygen to the entire body. As a result, the heart rate recovery rate decreases. Therefore, it can be inferred that there is a correlation between the mental and physical recovery ability information D1 and the fact that the learning subject is in a low-oxygen environment such as a tunnel excavation site or underground construction site.

[0202] For example, in an enclosed space that may be a hypoxic environment, mild hypoxia may occur due to the accumulation of carbon dioxide and oxygen dilution. In such a situation, a high heart rate persists after exercise due to the supply of oxygen to the entire body. As a result, the heart rate recovery rate decreases. Therefore, it can be inferred that there is a correlation between the mind-body recovery ability information D1 and the fact that the learning subject's environment is an enclosed space.

[0203] For example, in extremely hot and humid environments (e.g., saunas, factories, or under the blazing sun), if body temperature rises and dehydration occurs, it is thought that the heart rate will remain high for a longer period after exercise. As a result, the heart rate recovery speed will decrease. Therefore, it can be inferred that there is a correlation between the mind and body recovery ability information D1 and the hot and humid environment in which the learning subject is staying.

[0204] For example, in a low-temperature environment that is an extremely hot environment (e.g., a cold warehouse or outdoors in a cold region), peripheral vasoconstriction and sympathetic nervous activation are thought to occur, causing the heart rate to remain high. As a result, the heart rate recovery speed decreases. Therefore, it can be inferred that there is a correlation between the mind and body recovery ability information D1 and the fact that the learning subject is in a low-temperature environment.

[0205] For example, in an environment where the subject is exposed to the harmful gas carbon monoxide, a decrease in oxygen transport capacity leads to a sustained high heart rate, which creates an oxygen debt and is thought to slow the heart rate recovery rate after exercise. Therefore, it can be inferred that there is a correlation between the mind-body recovery ability information D1 and the fact that the learning subject is in an environment where he or she is exposed to carbon monoxide.

[0206] For example, in an environment with a lot of fine particles, dust, or PM2.5 (an environment with abnormal air composition), increased respiratory resistance is thought to delay respiratory recovery after exercise and reduce the heart rate recovery rate. Therefore, it can be inferred that there is a correlation between the mind-body recovery ability information D1 and the fact that the learning subject is in an environment with a lot of fine particles, etc.

[0207] For example, it is presumed that an environment with a high concentration of volatile organic compounds (VOCs) (an environment with abnormal air components) induces central stimulation and sympathetic reflexes, maintaining a high heart rate. This makes it difficult for the heart rate to recover after exercise. As a result, the heart rate recovery rate decreases. Therefore, it can be presumed that there is a correlation between the mind-body recovery ability information D1 and the fact that the learning subject is in an environment with a high concentration of volatile organic compounds.

[0208] For example, in an environment with a high concentration of carbon dioxide (an environment with abnormal air components), a high heart rate is thought to persist as respiratory acidosis is corrected. This makes it difficult for the heart rate to recover after exercise. As a result, the heart rate recovery speed decreases. Therefore, it can be inferred that there is a correlation between the mind-body recovery ability information D1 and the fact that the learning subject is in an environment with a high concentration of carbon dioxide.

[0209] The stress environment information may include, for example, information about a loud environment, a low frequency sound environment, a stressful environment, or a surveillance environment.

[0210] For example, in a high-volume or low-frequency sound environment, auditory stress is thought to delay heart rate recovery. As a result, the heart rate recovery speed decreases. Therefore, it can be inferred that there is a correlation between the mind-body recovery ability information D1 and whether the learning subject is in a high-volume or low-frequency sound environment.

[0211] For example, in a highly stressful or highly monitored environment (such as a workplace environment), psychological pressure can reduce parasympathetic nervous activity, potentially disrupting heart rate recovery after exercise has stopped. Therefore, it can be inferred that there is a correlation between the mind-body recovery ability information D1 and whether the learning subject is in a stressful or monitored environment.

[0212] The mental and physical environment information M110 may include, for example, information about an event (for example, an accident) that may be caused by an abnormal atmospheric environment or a mentally stressful environment.

[0213] Here, the content of the mental and physical condition information M100 is different from the content of the mental and physical environment information M110. In other words, the content of the mental and physical condition information M100 and the content of the mental and physical environment information M110 do not overlap.

[0214] As described above, the physical and mental recovery ability information D1 is an index that quantifies the speed and responsiveness of heart rate recovery after exercise. Therefore, the physical and mental recovery ability information D1 can indicate not only a decline in respiratory and circulatory system function, but also a hypoxic environment or a high-stress environment that affects the activity of the autonomic nervous system and reduces heart rate recovery rate. Therefore, for example, if a subject's heart rate recovery rate temporarily drops significantly compared to normal, it can be inferred that the environmental conditions have deteriorated. For example, as described above, the physical and mental recovery ability information D1 can contribute to predicting the occurrence of hypoxic environments and the possibility of associated accidents at tunnel excavation sites, confined work sites, and the like.

[0215] Here, the organism-related condition information M1 indicates, for example, the results of a diagnosis or evaluation by a medical professional, the patient himself / herself, or a third party, as information indicated by a score, binary-classified information, or multi-class classified information. Diagnosis and evaluation include diagnosis and evaluation based on the results of tests using medical equipment such as a medical image diagnostic device, as well as diagnosis and evaluation based on the results of biometric tests such as biomarkers. In binary classification, the organism-related condition information M1 is, for example, information indicating the "presence" or "absence" of a disease, symptom, abnormality, or intake. In multi-class classification, the organism-related condition information M1 is, for example, information indicating the degree of a disease, symptom, abnormality, quality of life, or environmental condition in stages.

[0216] For example, in the case of COPD, in binary classification by a medical professional, the biological-related condition information M1 is information indicating "diagnosed" of COPD (e.g., 1) or information indicating "not diagnosed" of COPD (e.g., 0) based on the results of a chest X-ray taken by an X-ray examination device. For example, in the case of COPD, in multi-class classification by a medical professional, the biological-related condition information M1 indicates severity. In this case, the severity is represented by, for example, information indicating Stage I (e.g., chronic cough and phlegm, mild shortness of breath) (e.g., 0), information indicating Stage II (e.g., heavy cough and phlegm, shortness of breath) (e.g., 1), information indicating Stage III (e.g., heavy cough and phlegm, severe shortness of breath, reduced exercise ability) (e.g., 2), or information indicating Stage IV (heavy cough and phlegm, shortness of breath that significantly interferes with daily life) (e.g., 3).

[0217] Furthermore, the biological-related state information M1 may be, for example, the value of an evaluation index or test index for evaluating or testing a disease, symptom, abnormality, quality of life, or environmental state, information obtained by classifying the value into two values, or information obtained by classifying the value into multiple classes. Examples of the evaluation index or test index include a disease activity index, a biopsy test index, a severity score, a functional assessment scale, a prognosis score, and a symptom assessment scale. Examples of the symptom assessment scale include an assessment scale such as a pain scale (e.g., a VAS (Visual Analogue Scale)).

[0218] For example, in the case of COPD, in multi-class classification by a medical professional, the biological condition information M1 indicates the severity based on the value of an evaluation index or test index. In this case, the evaluation index or test index is, for example, "%FEV1." "%FEV1" is calculated based on the amount of air exhaled per second (FEV1) and the forced vital capacity (FVC). In this case, if the GOLD classification is used, the severity is represented by information indicating Stage I (%FEV1≧80%) (e.g., 0), information indicating Stage II (50%≦%FEV1<80%) (e.g., 1), information indicating Stage III (30%≦%FEV1<50%) (e.g., 2), or information indicating Stage IV (%FEV1<30%, or %FEV1<50% and chronic respiratory failure) (e.g., 3).

[0219] Furthermore, the biological-related condition information M1 may be, for example, a numerical value indicating the result of a biometric test such as a biomarker, information obtained by classifying the numerical value into two values, or information obtained by classifying the numerical value into multiple classes.

[0220] Furthermore, the biological-related state information M1 may be, for example, a comprehensive result obtained by combining the results of the above examples, and may be represented by information indicated by a score, binary-classified information, or multi-class-classified information.

[0221] As described above, the bio-related state information M1 may directly indicate the physical or environmental state in numerical values ​​(direct numerical output), may indicate the physical or environmental state by binary classification (binary classification output), or may indicate the physical or environmental state by multi-class classification (multi-class classification output).

[0222] Continuing with reference to FIG. 27 , the feature information G1 will be described. The feature information G1 may further include one or more of the following information: vital information H1 of the learning subject; behavioral information J1 of the learning subject; information R1 about the learning subject's environment (hereinafter referred to as "environmental information R1"); self- or third-party report information V1 about the learning subject's physical and mental state; and attribute information Q1 of the learning subject. This is because these pieces of information are correlated with the organism-related state information M1, which can further improve the estimation accuracy of the organism-related state information M2 by the learning model TM1. For example, the feature information G1 may include vital information H1 and behavioral information J1 in addition to the physical and mental recovery ability information D1.

[0223] First, the correlation between the vital information H1 and the like and the living body-related condition information M1 will be described.

[0224] For example, vital information H1 is an objective indicator of the physiological functional state of the learning subject and responds directly to changes in the physical state. Since information representing the physical state is reflected in these measurements, the two are closely related. Therefore, it can be inferred that there is a correlation between the vital information H1 and the physical state information of the organism-related state information M1.

[0225] For example, vital information H1 is an objective indicator that reflects the autonomic nervous activity of the learning subject, and fluctuates according to changes in mental state. Because mental state affects vital information through stress responses and emotional changes, the two are closely related. Therefore, it can be inferred that there is a correlation between vital information H1 and mental state information in the organism-related state information M1.

[0226] For example, the vital information H1 objectively indicates biological reactions due to the intake of specific components of medicines, luxury goods, etc. These components directly affect vital indicators such as heart rate, so changes in the intake status are reflected in the vital information. Therefore, it can be inferred that there is a correlation between the vital information H1 and the specific component intake information of the biological-related condition information M1.

[0227] For example, vital information H1 is an objective indicator that reflects the state of the biological functions of the learning subject and is related to changes in quality of life. A decline in quality of life leads to physical and mental stress, which manifests as abnormalities in vital information, so the two influence each other. Therefore, it can be inferred that there is a correlation between the vital information H1 and the QOL information of the biological-related condition information M1.

[0228] For example, the vital information H1 is an objective indicator of the environmental response of the living body of the learning subject, and responds sensitively to changes in the environmental condition. Atmospheric abnormalities such as hypoxia and high temperature activate the compensatory functions of the living body, which are observed as fluctuations in the vital information. Therefore, it can be inferred that there is a correlation between the vital information H1 and the atmospheric abnormality information of the living body-related condition information M1.

[0229] For example, vital information H1 is an objective indicator of the stress response of a learning subject, and changes when the subject is exposed to a mentally stressful environment. A highly stressful or noisy environment stimulates the autonomic nervous system, which is directly reflected in the vital information. Therefore, it can be inferred that there is a correlation between the vital information H1 and the stressful environment information of the organism-related state information M1.

[0230] For example, the behavioral information J1 is an objective indicator that indicates the exercise intensity and exercise pattern of the learning subject, and is closely related to the physical condition. Physical illnesses and functional disorders manifest as a decrease in exercise intensity and the number of steps taken, while appropriate physical activity improves physical function. Therefore, it can be inferred that there is a correlation between the behavioral information J1 and the physical condition information of the organism-related condition information M1.

[0231] For example, the behavioral information J1 is an objective indicator of the exercise intensity and exercise pattern of the learning subject, and reflects the mental state. Changes in mental state such as anxiety and depression appear as changes in exercise intensity and activity rhythm, and are reflected in the fluctuation pattern of the behavioral information. Therefore, it can be inferred that there is a correlation between the behavioral information J1 and the mental state information of the organism-related state information M1.

[0232] For example, behavioral information J1 is an objective indicator of the exercise intensity and exercise pattern of the learning subject, and is related to the intake status of a specific component. The intake of drugs or luxury items affects the physical activity pattern and exercise intensity, which is observed as a change in behavioral information. Therefore, it can be inferred that there is a correlation between the behavioral information J1 and the specific component intake information of the biological-related state information M1.

[0233] For example, the behavioral information J1 is an objective indicator that indicates the exercise intensity and exercise pattern of the learning subject, and is directly related to quality of life. A decline in quality of life appears as a decrease in exercise intensity and a decrease in the number of steps taken, and is clearly reflected in the behavioral information. Therefore, it can be inferred that there is a correlation between the behavioral information J1 and the QOL information of the organism-related state information M1.

[0234] For example, the behavioral information J1 is an objective indicator that indicates the exercise intensity and exercise pattern of the learning subject, and changes depending on environmental conditions. Atmospheric abnormalities such as hypoxia and extreme temperatures limit exercise intensity and physical activity and are detected as changes in the behavioral information. Therefore, it can be inferred that there is a correlation between the behavioral information J1 and the atmospheric abnormality information in the organism-related state information M1.

[0235] For example, the behavioral information J1 is an objective indicator that indicates the exercise intensity and exercise pattern of the learning subject, and is affected by the mental stress environment. A highly stressful environment or a monitoring environment changes the exercise pattern, which is manifested as a change in exercise intensity and walking state. Therefore, it can be inferred that there is a correlation between the behavioral information J1 and the stress environment information of the organism-related state information M1.

[0236] For example, the environmental information R1 is an objective indicator of the environmental conditions in which the learning subject lives, and is related to the physical condition. Changes in weather conditions and working environment affect the function of the circulatory system and respiratory system, causing fluctuations in the physical condition. Therefore, it can be inferred that there is a correlation between the environmental information R1 and the physical condition information of the organism-related condition information M1.

[0237] For example, environmental information R1 is an objective indicator of the learning subject's living environment and affects their mental state. Seasonal changes, weather conditions, and stress factors in the work environment affect mood and emotional state, resulting in changes in mental state. Therefore, it can be inferred that there is a correlation between environmental information R1 and the mental state information in the biological-related state information M1.

[0238] For example, environmental information R1 is an objective indicator of the external environmental conditions of the study subject and is related to the intake pattern of a specific component. Changes in season and work status affect the tendency to use medications and consume luxury items, which are observed as fluctuations in the intake of a specific component. Therefore, it can be inferred that there is a correlation between environmental information R1 and the specific component intake information of the organism-related condition information M1.

[0239] For example, environmental information R1 is an objective indicator of the living environment conditions of the learning subject and is closely related to quality of life. The quality of weather conditions and working environment directly affect life satisfaction and activity ability, and are manifested as fluctuations in quality of life. Therefore, it can be inferred that there is a correlation between environmental information R1 and the QOL information of the organism-related state information M1.

[0240] For example, environmental information R1 is an objective indicator of the external environmental condition of the learning subject and is directly related to atmospheric abnormalities. Season and weather information affect oxygen concentration and thermal conditions, and the working environment is related to air quality and exposure to harmful substances. Therefore, it can be inferred that there is a correlation between environmental information R1 and the atmospheric abnormality information of the organism-related condition information M1.

[0241] For example, the environmental information R1 is an objective indicator of the living and working environment of the learning subject, and is closely related to the mental stress environment. Work information and seasonal fluctuations are directly linked to mental stress factors such as noise levels and tension, and changes in the environment appear as fluctuations in mental stress. Therefore, it can be inferred that there is a correlation between the environmental information R1 and the stress environment information of the organism-related state information M1.

[0242] The self-reported information V1 from the learning subject himself / herself captures each aspect of the organism-related condition information M1 from a subjective point of view, and it can be assumed that there is a correlation between the two.

[0243] The report information W1 by a third party captures each aspect of the organism-related condition information M1 from an objective point of view, and it can be assumed that there is a correlation between the two.

[0244] The attribute information Q1 is information indicating the basic characteristics of the learning subject and indicates the basic background related to the content of the biological-related state information M1. Therefore, it can be inferred that there is a correlation between the attribute information Q1 and the biological-related state information M1.

[0245] Here, particularly when the feature information G1 further includes one or more of vital information H1, behavioral information J1, environmental information R1, report information V1, and attribute information Q1 of the learning subject, the feature information G1 (explanatory variable) and the organism-related state information M1 (objective variable) are at least different in content or timing. This is because the objective variable to be estimated and the explanatory variables used to explain the objective variable for estimation are at least different in content or timing. For example, even if the content of the feature information G1 and the organism-related state information M1 is the same, if the feature information G1 is past information and the organism-related state information M1 is current or latest information, the feature information G1 and the organism-related state information M1 are at different times.

[0246] Next, the contents of the vital information H1 and the like will be described in detail.

[0247] The vital sign information H1 includes information related to the heartbeat. In this specification, the heartbeat is not limited to a heartbeat based on an electrocardiogram waveform acquired by electrocardiography (ECG), but also includes a pulse based on a pulse waveform. In other words, in this specification, a heartbeat based on an electrocardiogram waveform and a pulse based on a pulse waveform are treated as "heartbeats." The method for acquiring the pulse waveform is not particularly limited, and the pulse waveform may be acquired by, for example, photoplethysmography (PPG).

[0248] The information related to the heart rate may be, for example, the heart rate (HR), the R-R interval, a time domain index related to the heart rate, a frequency domain index related to the heart rate, or a nonlinear index related to the heart rate. The heart rate is the number of times the heart beats within a certain period of time, and is expressed, for example, in beats per minute (bpm). A certain period of time in the heart rate may be referred to as a first certain period of time. The R-R interval is the time interval from one QRS wave to the next QRS wave in an electrocardiogram waveform. The time domain index related to the heart rate, the frequency domain index related to the heart rate, and the nonlinear index related to the heart rate are collectively referred to as heart rate variability indexes (HRV indexes). Note that in this specification, for example, the pulse rate (PR) is treated as the heart rate, the pulse interval (PI) is treated as the R-R interval, and pulse rate variability is treated as heart rate variability.

[0249] Examples of time domain indices include SDNN (Standard deviation of NN intervals), RMSSD (Root Mean Square of Successive Differences), CVRR (Coefficient of Variation of RR intervals), SDRR (Standard deviation of RR intervals), SDANN (Standard Deviation of the Average NN intervals for each 5-minute segment of a 24-hour HRV recording), SDNN index, NN50 (the number of pairs of successive NN intervals that differ by more than 50 ms), pNN50 (the proportion of NN50 divided by the total number of NN intervals), HR Max, HR Min, (HR Max - HR Min), HTI (HRV Triangular Index), or TINN (Triangular Interpolation of the NN Interval Histogram). HRV is heart rate variability. RMSSD is, in terms of heartbeats, the square root of the mean value of the squares of the differences between successive adjacent R-R intervals.

[0250] The frequency domain index is, for example, LF (power or peak of low frequency components), HF (power or peak of high frequency components), LF / HF, Total Power, LF Norm, HF Norm, ULF (power in the extremely low frequency region), or VLF (power in the very low frequency region) of heart rate variability.

[0251] Examples of nonlinear indices include entropy, SD1 (standard deviation in the direction perpendicular to y=x in a Poincaré plot), SD2 (standard deviation in the direction along y=x in a Poincaré plot), SD1 / SD2, ApEn (approximate entropy), SampEn (sample entropy), DFA α1 (Detrended Fluctuation Analysis, which describes short-term fluctuations), DFA α2 (Detrended Fluctuation Analysis, which describes long-term fluctuations), (DFA α1) / (DFA α2), CVI (Cardiac Vagal Index), or CSI (Cardiac Sympathetic Index).

[0252] The vital information H1 may include, for example, one or more of blood pressure information, respiration information, body temperature information, blood information, and electroencephalogram information. The respiration information may be, for example, a respiration rate, a respiration rate, or a respiration volume. The body temperature information may be, for example, at least one of skin temperature, body temperature, and core body temperature. The blood information may be, for example, arterial blood oxygen saturation (e.g., SpO ２ ) or blood sugar levels.

[0253] The vital information H1 may also include exercise reactivity information. The exercise reactivity information is information that indicates the degree of physical and mental response to exercise. Typically, the exercise reactivity information is information that indicates the degree of increase in heart rate after starting exercise. In this case, the exercise reactivity information can be used to evaluate the state of the sympathetic nervous system (autonomic nervous system disturbance) and lack of exercise.

[0254] The activity information J1 includes at least one of information regarding the number of steps and information regarding exercise intensity. Information regarding the number of steps and information regarding exercise intensity are types of exercise amount information. The number of steps and exercise intensity may be collectively referred to as exercise amount. Information regarding the number of steps may indicate, for example, the number of steps per minute (spm) within a certain period of time (e.g., one minute). A certain period of time in steps may be referred to as a second certain period of time. Information regarding exercise intensity may be indicated, for example, in metabolic equivalents (METs). METs indicate, for example, the number of METs within a certain period of time (e.g., one minute). A certain period of time in METs may be referred to as a third certain period of time. METs is a unit that represents the intensity of physical activity as a multiple of the resting state. Note that information regarding exercise intensity is not limited to METs and may be indicated, for example, by a relative value to maximum oxygen uptake (%VO2max) or a method based on maximum heart rate (%HRmax, %MHR).

[0255] The behavior information J1 may include, for example, one or more of information on energy consumption and body movement information. The body movement information may be, for example, the amount or frequency of a specific body movement (e.g., rolling over or struggling). The behavior information J1 may also include the time the biological state detection device (e.g., a wearable device) is worn by the learning subject.

[0256] The behavioral information J1 may also include sleep information of the learning subject. The sleep information is information related to the sleep of a living organism. The sleep information includes at least one of information related to sleep duration and information related to sleep rhythm. The sleep information may also include information related to sleep state and / or information related to medication for sleep.

[0257] Furthermore, the behavioral information J1 may include drinking information of the learning subject. The drinking information is information about drinking by a living body. The drinking information is, for example, information indicating whether or not the subject has drunk alcohol, information indicating the amount of alcohol consumed, or information indicating the subject's drinking history.

[0258] The behavior information J1 may include, for example, smoking information, which indicates whether or not the user smokes or their smoking history.

[0259] The environmental information R1 is information indicating the environment in which a living organism lives or works. The environmental information R1 includes, for example, one or more of seasonal information, weather information, and employment information. The seasonal information may be indicated by the name of the season, such as winter, or by the month, such as January. The seasonal information includes, for example, information on the amount of ultraviolet light exposure calculated from the season. The weather information includes, for example, information on weather or atmospheric pressure. The employment information includes, for example, one or more of work history, working conditions, work shifts, and the amount of organic solvent exposure due to the occupation.

[0260] The self-report information V1 includes a subjective report of the learning subject's physical and mental state. The self-report information V1 includes, for example, a self-report by the learning subject regarding their physical state, mental state, or quality of life. For example, the self-report information V1 may be in text format, may be the result of binary or multi-class classification of subjective assessments, or may be the results of a rating scale, questionnaire, or survey expressed numerically. The self-report regarding the physical state includes, but is not limited to, subjective reports regarding, for example, respiratory symptoms such as shortness of breath, dizziness or unsteadiness, fatigue, sleep (quantity and quality), amount of exercise, or the presence or severity of various physical symptoms. The self-report regarding the mental state includes, but is not limited to, subjective reports regarding, for example, anxiety or psychological stress.

[0261] The third-party report information W1 includes an objective report by a third party indicating the learning subject's physical and mental state. The third party may be, for example, a medical professional, a close relative, or a cohabitant. The third-party report information W1 includes, for example, an objective report by a third party regarding the learning subject's physical state, mental state, or quality of life. For example, the third-party report information W1 may be in text format, may be the result of binary or multi-class classification of an objective assessment, or may be the result of a rating scale, questionnaire, or survey, etc., expressed numerically. The objective report regarding the physical state includes, but is not limited to, objective reports regarding respiratory symptoms such as shortness of breath, dizziness, unsteadiness, fatigue, sleep (quantity and quality), amount of exercise, or the presence or severity of various physical symptoms. The objective report regarding the mental state includes, but is not limited to, objective reports regarding anxiety or psychological stress.

[0262] The self- or third-party report information W1 may be, for example, current or latest information, or may be past information (information reported in the past).

[0263] The attribute information Q1 of the learning subject is information indicating the attributes of the learning subject. The attribute information Q1 includes, for example, one or more of basic information, anthropometric information, medical history information, diagnostic history information, drug use history information, lifestyle information, exercise-related information, genetic information, and disability-related information. The basic information is, for example, information on age, gender, race, and / or place of residence. The anthropometric information is, for example, information on height, weight, and / or BMI. The medical history information is, for example, information on medical history, surgical history, and / or current illness history. The diagnostic history information is, for example, the results of a health check or information on diagnostic images obtained by a medical image diagnostic device. The drug use history information is, for example, current medication status and / or past medication information. The exercise-related information is, for example, information on whether or not the learning subject has an exercise habit, the type of exercise, and / or the frequency of exercise. The disability-related information is, for example, information on whether or not the learning subject has a disability, the type of disability, and / or the severity of the disability.

[0264] In the fourth embodiment, the physical and mental recovery capacity information D1, vital information H1, and behavioral information J1 may include the physical and mental recovery capacity information, vital information, and behavioral information for a period indicated by a specific condition. The specific condition indicates any of a sleep-related condition, a wake-related condition, and a condition related to a time period of a day caused by the sun. The period indicated by the sleep-related condition is, for example, a period during sleep, a predetermined period before falling asleep, or a predetermined period after falling asleep. The period indicated by the wake-related condition is, for example, a period during wake-up, a predetermined period before waking up, or a predetermined period after waking up. The period indicated by the sun-related condition related to a time period of a day is, for example, a predetermined daytime period, a predetermined nighttime period, a predetermined morning period, a predetermined daytime period, or a predetermined nighttime period. However, with regard to the physical and mental recovery capacity information D1, the period during sleep, the predetermined period after falling asleep, and the predetermined period before waking up are not included in the "period indicated by the specific condition." This is because the physical and mental recovery capacity information D1 cannot be calculated during these periods.

[0265] The reason why the mental and physical recovery capacity information D1, vital sign information H1, and behavioral information J1 for a period indicated by specific conditions are effective in estimating the organism-related state information M2 is that biological functions or states exhibit diurnal fluctuations due to mental, physical, or environmental abnormalities. For example, biological functions or states may fluctuate depending on the time of day due to mental, physical, or environmental abnormalities. Therefore, by learning the mental and physical recovery capacity information D1, vital sign information H1, and behavioral information J1 for different times of the day, it may be possible to estimate mental, physical, or environmental abnormalities with high accuracy. In particular, because sleep state, wakefulness state, and time periods due to the sun are basic elements of biological rhythms, setting a period based on these conditions is particularly effective in estimating the organism-related state information M2.

[0266] As an example, the following explains why the mental and physical recovery capacity information D1 for a period indicated by specific conditions is effective in estimating the organism-related condition information M2. In diseases such as asthma, symptoms may fluctuate throughout the day. For example, respiratory function in asthma varies depending on the time of day, and large diurnal fluctuations in respiratory function indicate unstable bronchial conditions. Decreased respiratory function is associated with the risk of attacks and the state of attacks. Meanwhile, the mental and physical recovery capacity information D1 is an index that primarily represents the functional state of the respiratory and circulatory systems. Therefore, by quantifying the heart rate recovery rate at different time periods, such as after waking up and before falling asleep, during the day and at night, or in the morning, afternoon, and night, using the mental and physical recovery capacity information D1 and learning the differences between different time periods, it is possible to evaluate the severity of asthma, the instability of symptoms, the state of attacks, or predict the risk of future attacks. In this way, by using the mental and physical recovery capacity information D1 by time period, the state and worsening tendency of respiratory diseases, including asthma, can be effectively estimated and / or predicted.

[0267] For example, the physical and mental recovery capacity information D1 may include the difference between the physical and mental recovery capacity information for a first period indicated by a specific condition and the physical and mental recovery capacity information for a second period indicated by the specific condition, where the first period and the second period are, for example, a predetermined period after waking up and a predetermined period before falling asleep, a morning period and an afternoon period, or a daytime period and a nighttime period.

[0268] Next, referring to FIG. 28 , the input information K2 input to the learning model TM1 and the output information L2 output from the learning model TM1 will be described. FIG. 28 is a diagram showing an example of the input information K2 and the output information L2. As shown in FIG. 28 , the input information K2 includes at least the subject's mental and physical recovery ability information D2. The mental and physical recovery ability information D2 is information indicating the degree of speed of the subject's mental and physical recovery since the end of exercise. Typically, the mental and physical recovery ability information D2 indicates the degree of speed of recovery of the subject's heart rate since the end of exercise. Hereinafter, the speed of mental and physical recovery may be referred to as the "mental and physical recovery rate," and the speed of heart rate recovery may be referred to as the "heart rate recovery rate." Otherwise, the mental and physical recovery ability information D2 is similar to the learning subject's mental and physical recovery ability information D1.

[0269] The input information K2 may further include one or more of the subject's vital sign information H2, the subject's behavioral information J2, information R2 about the subject's environment (hereinafter referred to as "environmental information R2"), self-report information V2 about the subject's physical and mental state, and the subject's attribute information Q2. For example, the input information K2 may include the vital sign information H2 and the behavioral information J2 in addition to the physical and mental recovery ability information D2. Note that when the input information K2 includes the vital sign information H2, the feature amount information G1 ( FIG. 27 ) includes the vital sign information H1; when the input information K2 includes the behavioral information J2, the feature amount information G1 includes the behavioral information J1; when the input information K2 includes the environmental information R2, the feature amount information G1 includes the environmental information R1; when the input information K2 includes the self-reported information V2, the feature amount information G1 includes the report information V1; and when the input information K2 includes the attribute information Q2, the feature amount information G1 includes the attribute information Q1.

[0270] The vital information H2 includes information related to the heart rate. The vital information H2 may also include exercise reactivity information. The information related to the heart rate and the exercise reactivity information in FIG. 28 are similar to the information related to the heart rate and the exercise reactivity information in FIG. 27, respectively.

[0271] The behavior information J2 includes at least one of information regarding the number of steps and information regarding exercise intensity. The behavior information J1 may further include one or more of information regarding the subject's sleep and alcohol consumption. The information regarding the number of steps, exercise intensity, sleep, and alcohol consumption in FIG. 28 are similar to the information regarding the number of steps, exercise intensity, sleep, and alcohol consumption in FIG. 27, respectively.

[0272] Environmental information R2 is information indicating the environment in which a living organism lives or works. Self-report information V2 includes a subjective report of the subject's physical and mental state. Self-report information V2 may be, for example, current or latest information, or past information (information reported in the past). Attribute information Q2 is information indicating the attributes of the subject. Furthermore, environmental information R2, self-report information V2, and attribute information Q2 in FIG. 28 are similar to environmental information R1, self-report information V1, and attribute information Q1 in FIG. 27, respectively.

[0273] At least one of the physical and mental recovery ability information D2, vital information H2, and behavior information J2 may include physical and mental recovery ability information, vital information, or behavior information for a period indicated by a specific condition. The specific condition indicates any of a condition related to sleep, a condition related to wakefulness, and a condition related to a time period of a day caused by the sun. The specific conditions for the physical and mental recovery ability information D2, vital information H2, and behavior information J2 are the same as the specific conditions for the physical and mental recovery ability information D1, vital information H1, and behavior information J1 in FIG. 27.

[0274] On the other hand, the output information L2 includes the subject's biological-related state information M2. The biological-related state information M2 includes information M200 representing the subject's mental and physical state (hereinafter referred to as "mental and physical state information M200") or information M210 representing the state of the environment that affects the subject's mental and physical state (hereinafter referred to as "mental and physical environment information M210"). The mental and physical state information M200 and mental and physical environment information M210 in Fig. 28 are similar to the mental and physical state information M100 and mental and physical environment information M110 in Fig. 27, respectively.

[0275] For example, the mental and physical condition information M200 includes information representing a physical state (physical condition information), information representing a mental state (mental state information), information regarding the intake of specific components (specific component intake information), or information regarding quality of life (QOL information). QOL information indicates a physical state, a mental state, or quality of life related to the intake of specific components. The physical condition information includes information representing a state related to a disease of the circulatory system, information representing a state related to a disease of the respiratory function system, information representing a state related to a disease of the biocontrol system, information representing subjective symptoms related to a physical state, or information representing a physical state corresponding to the results of a biopsy. The mental state information includes information representing a state related to a mental illness or information representing a state related to mental stress. The specific component intake information includes information regarding the intake of medicines or information regarding the intake of luxury goods. The mental and physical environment information M210 includes information regarding atmospheric abnormalities or information regarding a mentally stressful environment.

[0276] In the explanation of Figure 27, for example, by replacing learning subject, feature information G1, mental and physical recovery ability information D1, vital information H1, behavioral information J1, environmental information R1, self-reported information V1, attribute information Q1, correct answer label B1, organism-related state information M1, mental and physical state information M100, and mental and physical environment information M110 with subject, input information K2, mental and physical recovery ability information D2, vital information H2, behavioral information J2, environmental information R2, self-reported information V2, attribute information Q2, output information L2, organism-related state information M2, mental and physical state information M200, and mental and physical environment information M210, respectively, the explanation can be substituted for the explanation of input information K2 and output information L2.

[0277] Typically, the physical and mental recovery ability information D2, vital information H2, and behavior information J2 are information created based on biological condition raw data A2 acquired from a wearable device (e.g., the biological condition detection device 103 described below) and / or a mobile terminal (e.g., the first terminal 102 described below). However, these pieces of information may also be acquired from the wearable device and / or the mobile terminal.

[0278] Next, the utilization stage of the learning model TM1 will be described with reference to Figs. 29 to 35. Fig. 29 is a block diagram showing an example configuration of an estimation system 40 according to the fourth embodiment. As shown in Fig. 29, the estimation system 40 is connected to a network NW. The network NW includes, for example, the Internet, a closed network, a public telephone network, a LAN (Local Area Network), and a short-range wireless network. The estimation system 40 is part of the information processing system 1 of Fig. 1.

[0279] At least one client system 100 is connected to the network NW. The client system 100 includes a cloud server 101, a plurality of first terminals 102, and a plurality of biological condition detection devices 103. The cloud server 101, the first terminals 102, and the biological condition detection devices 103 are connected to the network NW. In addition, a plurality of second terminals 200 are connected to the network NW.

[0280] The biological condition detection device 103 detects the biological condition of the subject. The biological condition detection device 103 includes, for example, a sensor that detects the biological condition with or without contact. The biological condition detection device 103 outputs biological condition raw data A2, which is raw data indicating the biological condition of the subject. The biological condition raw data A2 includes raw vital data and raw behavioral data of the subject. The raw vital data is raw data indicating the vital signs of the subject. The raw behavioral data is raw data indicating the behavior of the subject. The sensor that detects the raw vital data includes, for example, an optical sensor (light-emitting element and light-receiving element) that performs PPG. In this case, for example, the subject's pulse waveform is detected. Therefore, the biological condition detection device 103 calculates the pulse rate (bpm) based on the pulse waveform. The optical sensor may measure arterial blood oxygen saturation. The sensor that detects the raw vital data may include, for example, a sensor that detects the subject's body temperature or skin temperature. The sensor that detects the raw behavioral data includes, for example, an acceleration sensor and / or a gyro sensor. In this case, for example, the number of steps and exercise intensity (METs) of the subject are detected. That is, the biological state detection device 103 calculates the number of steps and exercise intensity (METs) based on the output of an acceleration sensor and / or a gyro sensor. The sensor that detects the behavioral raw data may include, for example, an acceleration sensor and / or a gyro sensor, and a microphone. In this case, for example, the state of the subject's sleep is detected.

[0281] The biological condition detection device 103 is, for example, a wearable device. The wearable device is worn by the subject. The wearable device is, for example, a wristwatch type, a ring type, or a sticker type. The wearable device may be, for example, the wearable terminal WT of FIG. 1. The biological condition detection device 103 is synchronized with the first terminal 102 and transmits the biological condition raw data A2 to the first terminal 102. The first terminal 102 is, for example, a mobile terminal such as a smartphone. The mobile terminal is carried by the subject. The biological condition detection device 103 may transmit the biological condition raw data A2 to the cloud server 101 or the estimation system 40 via the network NW.

[0282] The biological condition detection device 103 may be a wearable type, a portable type, a tabletop type, or a dedicated detector (measuring device) for detecting a biological condition. The first terminal 102 may have some or all of the functions of the biological condition detection device 103. For example, the first terminal 102 may be a personal computer (PC).

[0283] The first terminal 102 transmits the subject's biological condition raw data A2 to the cloud server 101 via the network NW. The cloud server 101 transmits the biological condition raw data A2 to the estimation system 40 via the network NW. The estimation system 40 processes the biological condition raw data A2. Note that the first terminal 102 may transmit the biological condition raw data A2 to the estimation system 40 via the network NW without providing the cloud server 101. The biological condition raw data A2 may be transmitted directly from the biological condition detection device 103 to the estimation system 40. One or more pieces of information among the environmental information R2, the self-reported information V2, and the attribute information Q2 may be transmitted from the first terminal 102 via the cloud server 101 or directly to the estimation system 40. A part of this information may be transmitted from the second terminal 200 via the cloud server 101 or directly to the estimation system 40.

[0284] More specifically, the estimation system 40 includes a relay server 44, a first database 45, an estimation device 4, a second database 46, and an information providing server 47. The relay server 44, the first database 45, the estimation device 4, the second database 46, and the information providing server 47 are connected to a network NW. The estimation device 4 is, for example, a server. The server is a computer.

[0285] Each of the relay server 44 and the information providing server 47 may include a processing unit, a communication unit, and a memory unit, as well as an input unit and an output unit. The hardware configurations of the processing unit, communication unit, memory unit, input unit, and output unit are similar to the hardware configurations of the processing unit 400, communication unit 401, memory unit 402, input unit 403, and output unit 404 of the estimation device 4 shown in FIG. 31 (described later), or the hardware configuration of the mind-body estimation device SS shown in FIG. 24 or 25 . The first database 45 and the second database 46 include at least a storage device such as a hard disk drive. The first database 45 and the second database 46 may have the same hardware configuration as the relay server 44 or the information providing server 47. The estimation system 40 may not include all or some of the relay server 44, the first database 45, the second database 46, and the information providing server 47.

[0286] 30 is a flowchart showing an example of an estimation method executed by the estimation system 40. The estimation method estimates the subject's biological-related state information MX. The estimation method includes steps S51 to S55.

[0287] 29 and 30 , first, in step S51, the relay server 44 receives the subject's biological condition raw data A2 from the cloud server 101. Furthermore, for example, the relay server 44 may receive one or more of the subject's environmental information R2, self-reported information V2, and attribute information Q2. The relay server 44 implements, for example, an API (Application Programming Interface).

[0288] Next, in step S52, the first database 45 stores the biological condition raw data A2 received by the relay server 44. For example, the first database 45 may store one or more pieces of information selected from the environmental information R2, the self-reported information V2, and the attribute information Q2. Specifically, a record 451 is assigned to each subject in the first database 45. The record 451 is associated with the subject's identification information (personal identification information). The biological condition raw data A2 is recorded in the record 451 of the subject. The record 451 may store one or more pieces of information selected from the environmental information R2, the self-reported information V2, and the attribute information Q2.

[0289] Next, in step S53, the estimation device 4 estimates the subject's biological-related state information MX using the learning model TM1 and the mind-body recovery ability information D2 based on the biological state raw data A2 stored in the first database 45. Details of the estimation process will be described later. Note that the estimation process may utilize one or more of the environmental information R2, the self-report information V2, and the attribute information Q2.

[0290] Next, in step S54, the second database 46 stores the subject's biological-related state information MX (output information L2) estimated by the estimation device 4. Specifically, a record 461 is assigned to each subject in the second database 46. The record 461 is associated with the subject's identification information (personal identification information). The biological-related state information MX is then recorded in the subject's record 461. The second database 46 also stores input information K2 including physical and mental recovery ability information D2. Specifically, the input information K2 is recorded in the subject's record 461.

[0291] Next, in step S55, the information providing server 47 transmits the subject's biological-related condition information MX stored in the second database 46 to the second terminal 200 and / or the first terminal 102 via the network NW. Specifically, the information providing server 47 displays the subject's biological-related condition information MX on the second terminal 200 and / or the first terminal 102. The second terminal 200 is, for example, a terminal of a medical institution, a health management institution, or a research institution. The first terminal 102 is, for example, the subject's terminal. When step S5 is completed, the estimation method ends.

[0292] In the example of FIG. 29, the biological-related state information MX is transmitted to the first terminal 102 via the cloud server 101, but it may also be transmitted directly to the first terminal 102.

[0293] Fig. 31 is a block diagram showing an example configuration of the estimation device 4 in Fig. 29. As shown in Fig. 31, the estimation device 4 includes a processing unit 400, a communication unit 401, and a storage unit 402. The estimation device 4 may further include an input unit 403 and an output unit 404.

[0294] The input unit 403 is an input device for inputting various information to the processing unit 400. For example, the processing unit 400 is a keyboard and pointing device, or a touch panel.

[0295] The output unit 404 outputs various types of information. The output unit 404 includes, for example, a display unit that displays the various types of information. The display unit is, for example, a liquid crystal display or an organic electroluminescence display.

[0296] The communication unit 401 is connected to the network NW. The communication unit 401 communicates with external devices connected to the network NW. The communication unit 401 is a communication device that performs communication according to a predetermined communication protocol, and includes, for example, a network interface controller. The predetermined communication protocol is, for example, a protocol compliant with Ethernet (registered trademark), the Internet Protocol Suite, and a protocol compliant with a short-range wireless communication standard. The external devices are, for example, the first terminal 102, the second terminal 200, and the biological condition detection device 103.

[0297] The storage unit 402 includes one or more storage devices and stores data and computer programs. The storage unit 402 includes a main storage device such as a semiconductor memory, and an auxiliary storage device such as a semiconductor memory and a hard disk drive. The storage unit 402 may also include removable media such as an optical disk. The storage unit 402 may be, for example, a non-transitory computer-readable storage medium.

[0298] The memory unit 402 stores a learning model TM1. The learning model TM1 is a trained model. The learning model TM1 is a computer program. The learning model TM1 causes a computer to function so as to estimate the subject's biological-related state information M2. Specifically, the learning model TM1 causes the computer to function so as to input input information K2 and output output information L2.

[0299] The processing unit 400 executes various calculations. The processing unit 400 includes one or more processors. The processors may be a central processing unit (CPU), a graphics processing unit (GPU), a field programmable gate array (FPGA), a digital signal processor (DSP), or an application specific integrated circuit (ASIC). The processors may be operated by a computer program or by hardwired logic.

[0300] Specifically, the processing unit 400 includes a preprocessing unit 41 and an estimation unit 42. The processing unit 400 may further include a postprocessing unit 43. For example, the processor of the processing unit 400 functions as the preprocessing unit 41, the estimation unit 42, and the postprocessing unit 43 by executing a computer program stored in the storage device of the storage unit 402. The preprocessing unit 41 includes a mind-body estimation unit 411. The preprocessing unit 41 may further include at least one of a statistical processing unit 412, a condition processing unit 413, and an exercise reactivity estimation unit 414. The postprocessing unit 43 preferably includes an interpretation unit 431.

[0301] The pre-processing unit 41 acquires the biological condition raw data A2 from the first database 45. The storage unit 402 stores the biological condition raw data A2. The vital sign raw data of the biological condition raw data A2 includes, for example, heart rate data. The behavior raw data of the biological condition raw data A2 includes, for example, at least one of step count data and exercise intensity data. The pre-processing unit 41 may acquire, from the first database 45, one or more pieces of information from the environmental information R2, the self-reported information V2, and the attribute information Q2.

[0302] The heart rate data includes, for example, information on the heart rate (bpm). The heart rate data may also include, for example, information on the R-R interval. The heart rate is measured or aggregated, for example, at a sampling rate of at least once per minute. The heart rate data is accompanied by the corresponding date, time, minute, and second of measurement and identification information of the subject. The heart rate data includes information on the heart rate arranged in chronological order. The heart rate data may also include, for example, information on the R-R interval arranged in chronological order.

[0303] The step count data includes information on the number of steps (spm). The step count is measured or aggregated, for example, at a sampling rate of at least once per minute. The step count data includes the corresponding date, time, minute, and second of the measurement and the subject's identification information. The step count data includes information on the number of steps arranged in chronological order.

[0304] The exercise intensity data includes MET information per fixed time period (1 minute). That is, exercise intensity is typically expressed in METs. The exercise intensity data is measured or aggregated, for example, at a sampling rate of at least once per minute. The exercise intensity data is accompanied by the corresponding date, time, minute, and second of measurement and subject identification information. The exercise intensity data includes MET information arranged in chronological order.

[0305] The preprocessing unit 41 performs preprocessing on the biological state raw data A2 and generates input information K2, which is the result of the preprocessing. The storage unit 402 stores the input information K2.

[0306] As an example, the mind-body estimation unit 411 of the pre-processing unit 41 calculates the mind-body recovery ability information D2 for each unit period UT based on the heart rate data of the raw vital sign data and the raw behavioral data (e.g., exercise intensity data or step count data). In this case, the raw behavioral data is used to determine whether the person is exercising or at rest. The unit period UT is, for example, one day. Specifically, the mind-body estimation unit 411 calculates the mind-body recovery ability information D2 in the same manner as the mind-body estimation device SS (processing unit SY(SS)) of embodiments 1 to 3.

[0307] As an example, the statistical processing unit 412 of the preprocessing unit 41 performs statistical processing on each of the heart rate data, step count data, and exercise intensity data for each unit period UT. As a result, the statistical processing unit 412 outputs multiple statistical indices resulting from the statistical processing for each of the heart rate data, step count data, and exercise intensity data. The statistical processing performed by the statistical processing unit 412 represents mathematical processing for quantitatively expressing the characteristics and trends of the data being processed. Therefore, by performing statistical processing, the characteristics and trends of the data being processed (heart rate data, step count data, and exercise intensity data) can be accurately extracted, thereby further improving the estimation accuracy of the biological-related condition information M2 using the learning model TM1. The heart rate data, step count data, and exercise intensity data are, in detail, extracted heart rate data, extracted step count data, and extracted exercise intensity data, which will be described later. The statistical processing unit 412 may also perform differential processing and ratio processing for each unit period UT. This will be described in more detail later.

[0308] Examples of statistical indices include the mean value, sum value, standard deviation, x % tile value, median value, minimum value, maximum value, RMSSD, coefficient of variation CV (= standard deviation / mean value), inverse of the coefficient of variation CV, average change amount, percentage of changes equal to or greater than a specified value, statistically processed value of moving median over a time window, and heart rate variability index. Furthermore, RMSSD is calculated not only for heart rate (bpm), but also for steps (SPM) and exercise intensity (METs / min). In this case, RMSSD is the square root of the average of the squares of the differences between consecutively adjacent values. Examples of consecutively adjacent values ​​include consecutively adjacent "inverse of heart rate (bpm)," consecutively adjacent "steps (SPM)," or consecutively adjacent "exercise intensity (METs / min)."

[0309] The statistical processing unit 412 may generate all or some of the multiple types of statistical indices.

[0310] As an example, the condition processing unit 413 of the pre-processing unit 41 executes processing (condition processing) according to specific conditions for each unit period UT on each of the vital raw data and behavior raw data of the biological state raw data A2.

[0311] Specifically, the condition processing unit 413 executes a process of extracting raw vital data for a period indicated by a specific condition from the raw vital data of the subject, and / or a process of extracting raw behavioral data for a period indicated by a specific condition from the raw behavioral data of the subject. This will be described in detail below.

[0312] First, the processing of heart rate data from the raw vital sign data will be described. The condition processing unit 413 extracts heart rate data for a period indicated by specific conditions from the heart rate data for the unit period UT. The specific conditions include at least one of a condition related to sleep, a condition related to wakefulness, and a condition related to the time of day caused by the sun.

[0313] For example, the condition processing unit 413 extracts, from the heart rate data in the unit period UT, heart rate data while awake, heart rate data while asleep, heart rate data during the day, heart rate data at night, heart rate data at a predetermined time before falling asleep, heart rate data at a predetermined time after falling asleep, heart rate data at a predetermined time before waking up, and heart rate data at a predetermined time after waking up. Hereinafter, the heart rate data extracted according to specific conditions may be referred to as "extracted heart rate data."

[0314] The condition processing unit 413 passes multiple types of extracted heartbeat data for the unit period UT to the statistical processing unit 412. The statistical processing unit 412 performs multiple types of statistical processing on each piece of extracted heartbeat data and outputs multiple statistical indices for each piece of extracted heartbeat data. For example, the statistical processing unit 412 performs multiple types of statistical processing on the same extracted heartbeat data to calculate multiple different statistical indices from the same extracted heartbeat data. Therefore, it is possible to extract features of the extracted heartbeat data from different perspectives for the same extracted heartbeat data. This contributes to improving the estimation accuracy of the biological-related state information M2 by the learning model TM1.

[0315] In addition, the statistical processing unit 412 may calculate the difference (hereinafter, day-night difference) between the statistical index obtained by statistically processing the ``daytime heart rate data'' and the statistical index obtained by statistically processing the ``nighttime heart rate data'' for each different statistical index (difference processing).

[0316] In this way, multiple types of statistical indices and multiple types of day-night differences for each of the multiple types of extracted heart rate data are calculated for each unit period UT, and constitute part of the input information K2. In this case, the statistical indices and day-night differences may be collectively referred to as heart rate statistical indices. The storage unit 402 stores the multiple heart rate statistical indices as part of the input information K2.

[0317] The heartbeat statistical index is an example of vital information H2 of the input information K2. The extracted heartbeat data is an example of "extracted raw vital data." Therefore, the statistical processing unit 412 generates the vital information H2 by performing statistical processing on the extracted raw vital data.

[0318] The statistical processing unit 412 and the condition processing unit 413 may generate all or some of the multiple types of heartbeat statistical indices.

[0319] Next, processing of step count data of behavior raw data will be described. The condition processing unit 413 extracts step count data for a period indicated by a specific condition from the step count data for the unit period UT. The specific condition in this case is the same as the specific condition for heart rate data.

[0320] For example, the condition processing unit 413 extracts, from the step count data in the unit period UT, step count data while awake, step count data while asleep, step count data during the day, step count data at night, step count data for a predetermined time before falling asleep, step count data for a predetermined time after falling asleep, step count data for a predetermined time before waking up, and step count data for a predetermined time after waking up. Hereinafter, the step count data extracted according to specific conditions may be referred to as "extracted step count data."

[0321] The condition processing unit 413 passes multiple types of extracted step count data for the unit period UT to the statistical processing unit 412. The statistical processing unit 412 performs multiple types of statistical processing on each piece of extracted step count data and outputs multiple statistical indices for each piece of extracted step count data. For example, the statistical processing unit 412 performs multiple types of statistical processing on the same extracted step count data, thereby calculating multiple different statistical indices from the same extracted step count data. Therefore, it is possible to extract features of the extracted step count data from different perspectives for the same extracted step count data. This contributes to improving the estimation accuracy of the biological-related state information M2 by the learning model TM1.

[0322] In addition, the statistical processing unit 412 may calculate the difference (hereinafter, day-night difference) between the statistical index obtained by statistically processing the "daytime step count data" and the statistical index obtained by statistically processing the "nighttime step count data" for each different statistical index (difference processing).

[0323] Furthermore, the statistical processing unit 412 may calculate the ratio of periods in which the number of steps is zero (hereinafter referred to as occurrence ratio) to periods indicated by specific conditions (ratio processing).

[0324] For example, the statistical processing unit 412 calculates the occurrence rate during wakefulness, the occurrence rate during sleep, the occurrence rate during the day, the occurrence rate at night, the occurrence rate at a specified time before falling asleep, the occurrence rate at a specified time after falling asleep, the occurrence rate at a specified time before waking up, and the occurrence rate at a specified time after waking up.

[0325] In this way, multiple types of statistical indicators, multiple types of day-night differences, and multiple types of occurrence rates for each of the multiple types of extracted step count data are calculated for each unit period UT, and form part of the input information K2. In this case, the statistical indicators, day-night differences, and occurrence rates may be collectively referred to as step count statistical indicators. The storage unit 402 stores the multiple step count statistical indicators as part of the input information K2.

[0326] The step count statistical index is an example of behavior information J2 of the input information K2. The extracted step count data is an example of “extracted behavior raw data.” Therefore, the statistical processing unit 412 generates the behavior information J2 by ​​performing statistical processing on the extracted behavior raw data.

[0327] The statistical processing unit 412 and the condition processing unit 413 may generate all or some of the multiple types of step count statistical indicators.

[0328] Next, we will explain how to process exercise intensity data from the behavioral raw data. The condition processing unit 413 extracts exercise intensity data for a period specified by a specific condition from the exercise intensity data for the unit period UT. The specific condition in this case is the same as the specific condition for heart rate data.

[0329] For example, the condition processing unit 413 extracts, from the exercise intensity data for the unit period UT, exercise intensity data while awake, exercise intensity data while asleep, exercise intensity data during the day, exercise intensity data at night, exercise intensity data for a predetermined time before falling asleep, exercise intensity data for a predetermined time after falling asleep, exercise intensity data for a predetermined time before waking up, and exercise intensity data for a predetermined time after waking up. Hereinafter, the exercise intensity data extracted according to specific conditions may be referred to as "extracted exercise intensity data."

[0330] The condition processing unit 413 passes multiple types of extracted exercise intensity data for the unit period UT to the statistical processing unit 412. The statistical processing unit 412 performs multiple types of statistical processing on each extracted exercise intensity data and outputs multiple statistical indices for each extracted exercise intensity data. For example, the statistical processing unit 412 performs multiple types of statistical processing on the same extracted exercise intensity data to calculate multiple different statistical indices from the same extracted exercise intensity data. Therefore, it is possible to extract features of the extracted exercise intensity data from different perspectives for the same extracted exercise intensity data. This contributes to improving the estimation accuracy of the biological-related condition information M2 by the learning model TM1.

[0331] In addition, the statistical processing unit 412 may calculate the difference (hereinafter, day-night difference) between the statistical index obtained by statistically processing the "daytime exercise intensity data" and the statistical index obtained by statistically processing the "nighttime exercise intensity data" for each different statistical index (difference processing).

[0332] Furthermore, the statistical processing unit 412 may calculate the ratio of the period during which METs are equal to or less than a first specified value to the period specified by the specific condition (hereinafter referred to as the first occurrence ratio) (ratio processing).The condition processing unit 413 may calculate the ratio of the period during which METs are equal to or greater than a second specified value to the period specified by the specific condition (hereinafter referred to as the second occurrence ratio) (ratio processing).

[0333] For example, the statistical processing unit 412 calculates the first occurrence rate and the second occurrence rate during each of the following periods: awake, asleep, daytime, nighttime, a predetermined time before falling asleep, a predetermined time after falling asleep, a predetermined time before waking up, and a predetermined time after waking up.

[0334] In this way, multiple types of statistical indices, multiple types of day-night differences, and multiple types of occurrence rates for each of multiple types of extracted exercise intensity data are calculated for each unit period UT and constitute part of the input information K2. In this case, the statistical indices, day-night differences, and occurrence rates may be collectively referred to as exercise intensity statistical indices. The storage unit 402 stores the multiple exercise intensity statistical indices as part of the input information K2.

[0335] The exercise intensity statistical index is an example of the behavior information J2 of the input information K2. The extracted exercise intensity data is an example of the "extracted behavior raw data." Therefore, the statistical processing unit 412 generates the behavior information J2 by ​​performing statistical processing on the extracted behavior raw data.

[0336] The statistical processing unit 412 and the condition processing unit 413 may generate all or some of the multiple types of exercise intensity statistical indices.

[0337] As described above, when calculating step count statistical indicators and exercise intensity statistical indicators, the statistical processing unit 412 performs statistical processing after processing by the condition processing unit 413. Therefore, according to the fourth embodiment, the statistical processing can be performed for a period that accurately reflects the mental and physical state and the living body's environment. As a result, the estimation accuracy of the living body-related state information M2 by the learning model TM1 can be further improved.

[0338] As described above, the statistical processing unit 412 generates vital information H2 by performing statistical processing on the raw vital data extracted by the condition processing unit 413, and / or generates behavioral information J2 by ​​performing statistical processing on the raw behavioral data extracted by the condition processing unit 413.

[0339] In the fourth embodiment, the mind and body estimation unit 411 receives, for example, extracted heart rate data and extracted step count data, or extracted heart rate data and extracted exercise intensity data, from the condition processing unit 413. Then, the mind and body estimation unit 411 calculates mind and body recovery ability information D2 based on the extracted heart rate data. The mind and body estimation unit 411 uses the extracted step count data or the extracted exercise intensity data to determine whether the user is exercising or at rest.

[0340] For example, for each unit period UT, the mind-body estimation unit 411 calculates a parameter value PV (see embodiments 1 to 3) based on extracted heart rate data for a first period indicated by a specific condition, and calculates a parameter value PV (see embodiments 1 to 3) based on extracted heart rate data for a second period indicated by the specific condition. The first period and the second period are different periods within the unit period UT. The first period and the second period are, for example, a predetermined time after waking up and a predetermined time before falling asleep, a morning period and an afternoon period, or a daytime period and a nighttime period. Note that the condition processing unit 413 may extract heart rate data for three or more different "periods indicated by specific conditions" within the unit period UT. In other words, in this case, three or more extracted heart rate data are obtained within the unit period UT. The mind-body estimation unit 411 may then calculate a parameter value PV for each of the three or more extracted heart rate data in the unit period UT.

[0341] Furthermore, for example, the mind-body estimation unit 411 may calculate the difference (hereinafter, the difference within a unit period) between the parameter value PV based on the extracted heart rate data for a first period indicated by a specific condition and the parameter value PV based on the extracted heart rate data for a second period indicated by a specific condition (difference processing).

[0342] The parameter value PV and the intra-unit period difference are each the mental and physical recovery capacity information D2. That is, in this example, a plurality of pieces of mental and physical recovery capacity information D2 are calculated for each unit period UT.

[0343] In this way, multiple pieces of physical and mental recovery capacity information D2 (multiple parameter values ​​PV and intra-unit period differences) are calculated for each unit period UT and constitute part of the input information K2. The storage unit 402 stores multiple pieces of physical and mental recovery capacity information D2 for each unit period UT as part of the input information K2. The physical and mental recovery capacity information D2 may be referred to as a physical and mental recovery index.

[0344] The mind and body estimation section 411 may calculate one piece of mind and body recovery capacity information D2 or two or more pieces of mind and body recovery capacity information D2 within the unit period UT.

[0345] As an example, the exercise reactivity estimation unit 414 of the preprocessing unit 41 generates exercise reactivity information for each unit period UT based on the heart rate data of the raw vital sign data and the raw behavioral data (e.g., exercise intensity data or step count data). For example, the exercise reactivity estimation unit 414 generates an exercise reactivity estimation index. The exercise reactivity estimation index is an index that indicates the degree of increase in heart rate (e.g., heart rate) after starting exercise.

[0346] In the fourth embodiment, the motor reactivity estimation unit 414 calculates a motor reactivity estimation index for each unit period UT during nighttime and daytime (periods other than nighttime). In addition, the motor reactivity estimation unit 414 calculates the difference between the motor reactivity estimation index for nighttime and daytime (hereinafter referred to as the day-night difference).

[0347] In this way, the motor reactivity estimation indexes (two types in total) for two different time periods and the day-night difference are calculated for each unit period UT and form part of the input information K2. In this case, the motor reactivity estimation indexes and the day-night difference may be collectively referred to as motor reactivity indexes. The memory unit 402 stores multiple motor reactivity indexes as part of the input information K2. The motor reactivity indexes are an example of vital information H2 of the input information K2.

[0348] The exercise responsiveness estimation unit 414 may generate all or some of the multiple types of exercise responsiveness indices.

[0349] As described above with reference to Figures 27 to 31, the heart rate statistical index, step count statistical index, exercise intensity statistical index, mind-body recovery index, and exercise reactivity index constitute input information K2. Hereinafter, the heart rate statistical index, step count statistical index, exercise intensity statistical index, mind-body recovery index, exercise reactivity index, environmental information R2, self-report information V2, and attribute information Q2 may be collectively referred to as "features." The number of features is not particularly limited. Note that "features" may also be referred to as "features FT."

[0350] 31 , the estimation unit 42 estimates the subject's biological-related state information M2 using the learning model TM1. Specifically, the estimation unit 42 inputs input information K2 to the learning model TM1 and obtains output information L2 from the learning model TM1. The storage unit 402 stores the output information L2.

[0351] The learning model TM1 receives input information K2 and outputs output information L2 according to a machine learning algorithm. The machine learning algorithm is not particularly limited, and may be, for example, linear regression, Naive Bayes, Support Vector Machine, neural network, deep neural network (hereinafter referred to as DNN), decision tree, random forest, gradient boosting, or regularized regression. The regularized regression may be, for example, L1 regularized regression or L2 regularized regression. Below, as an example, a case where the machine learning algorithm of the learning model TM1 is DNN will be described.

[0352] FIG. 32 is a diagram schematically illustrating an example of a DNN 50. As shown in FIG. 32, the DNN 50 includes an input layer 51, multiple intermediate layers 52, and an output layer 53. FIG. 32 shows an example of a fully connected DNN. The input layer 51 includes at least one node 511. In this case, the mind-body recovery ability information D2 is input to the node 511. Each of the intermediate layers 52 includes multiple nodes 521. The output layer 53 includes at least one node 531. In the example of FIG. 32, the input layer 51 includes multiple nodes 511, and the output layer 53 includes multiple nodes 531.

[0353] A plurality of feature quantities FT constituting the input information K2 are input to each of the plurality of nodes 511 in the input layer 51. Each node 521 in the intermediate layer 52 converts the output of the previous layer into input to the next layer using trained weights and biases and an activation function. Each node 531 in the output layer 53 outputs a final result (estimated result) based on the output of the previous layer using trained weights and biases and an activation function according to the final output format. In other words, the output layer 53 outputs output information L2. The output information L2 includes a plurality of different types of biological-related state information M2. Note that when the output layer 53 includes one node 531, the output information L2 includes one piece of biological-related state information M2.

[0354] 32, the physical, mental, or environmental state is directly indicated by a numerical value by the biological-related state information M2 (direct numerical value output). The number of nodes 531 is the same as the number of biological-related state information M2 to be output, and may be one, two, or more.

[0355] On the other hand, the post-processing unit 43 in Fig. 31 may perform post-processing on the biological-related condition information M2 to output the biological-related condition information M3. This point will be described with reference to Figs. 33 and 34.

[0356] Fig. 33 is a diagram schematically showing another example of the DNN 50. In the example of Fig. 33, the mental and physical or environmental state is indicated by binary classification using the organism-related state information M3 (binary classification output).

[0357] A first example of binary classification will be described. In the first example, each node 531 directly outputs a numerical value as the biological-related state information M2. In this case, the post-processing unit 43 performs threshold processing on the numerical value indicated by the biological-related state information M2 to perform binary classification. The result of the binary classification is then output as biological-related state information M3.

[0358] A second example of binary classification will be described. In the second example, the biological-related state information M2 output by each node 531 is, for example, a real value between 0 and 1 (hereinafter referred to as a "score value"). For example, if "0" indicates a first state and "1" indicates a second state, the score value indicates a probability value that the mental / physical or environmental state will be classified as the second state. In this case, the post-processing unit 43 performs probability calibration on the score value to calibrate it to a more reliable probability value. The post-processing unit 43 then performs threshold processing on the score value after the probability calibration (hereinafter referred to as a "calibrated score value"), thereby performing binary classification. The result of the binary classification is then output as biological-related state information M3.

[0359] The number of nodes 531 is the same as the number of pieces of biological-related state information M2 to be output, and may be one or more. The number of nodes 511 may be one or more.

[0360] Fig. 34 is a diagram schematically illustrating yet another example of the DNN 50. In the example of Fig. 34, the mental and physical or environmental states are indicated by multi-class classification using the biological-related state information M3 (multi-class classification output). The mental and physical or environmental states are classified into U classes, where U is an integer equal to or greater than 3. The number of classifications (number of classes) U is not particularly limited and can be set arbitrarily.

[0361] As shown in Fig. 34, the DNN 50 includes an input layer 51, multiple intermediate layers 52, and at least one output layer 53 A. In the example of Fig. 34, the DNN 50 includes multiple output layers 53 A.

[0362] The plurality of output layers 53A are provided corresponding to the plurality of different types of biological-related state information M2, respectively. Each output layer 53A outputs output information L2. Specifically, each output layer 53A includes the same number of nodes 531A as the number of classifications (number of classes) U.

[0363] Focus on one output layer 53A. U nodes 531A correspond to U classes, respectively. Each node 531A outputs an output value VL. The output value VL is, for example, a real number between 0 and 1. The sum of the output values ​​VL of the U nodes 531A is "1." Therefore, the output value VL indicates a probability value for classification into the corresponding class. In other words, the class corresponding to the node 531A that outputs the largest output value VL has the highest probability of being correct. However, the post-processing unit 43 performs probability calibration on each output value VL to calibrate each output value VL to a more reliable probability value. The post-processing unit 43 then obtains the largest output value VL among the U output values ​​VL after calibration by probability calibration. The largest output value VL corresponds to the biological-related state information M2. In this way, the U output values ​​VL (output information L2) from the output layer 53A essentially include the biological-related state information M2. Then, the post-processing unit 43 sets the class corresponding to the node 531A that has output the largest output value VL after calibration to the biological-related state information M3. In this manner, the post-processing unit 43 performs multi-class classification.

[0364] The post-processing unit 43 performs multi-class classification for each of the multiple output layers 53A. The number of output layers 53A is the same as the number of pieces of biological-related state information M2 to be output, and may be one, two, or more. When there is one output layer 53A, one piece of biological-related state information M2 and one piece of biological-related state information M3 are output. The number of nodes 511 may be one, two, or more.

[0365] Returning to FIG. 31 , the interpretation unit 431 of the post-processing unit 43 calculates contribution information indicating the degree of contribution of the multiple feature values ​​FT constituting the input information K2 when the learning model TM1 estimates the biological-related state information M2. The contribution information includes the contribution of each feature value when estimating the biological-related state information M2. Specifically, the interpretation unit 431 calculates the contribution information based on the multiple feature values ​​FT, the biological-related state information M2 that is the estimation result, and information on the learning model TM1. In this case, the interpretation unit 431 calculates the contribution information according to a model interpretation method. The model interpretation method may be, for example, Shapley Additive Explanations (SHAP), Individual Conditional Expectation (ICE), Local Interpretable Model-agnostic Explanations (LIME), or Approximate Inverse Model Explanations (AIME). The model interpretation method is not limited to these, as long as it outputs information on the indices that contributed to the estimation result and the breakdown of the contribution. The storage unit 402 stores the contribution information. The second database 46 also stores contribution information. The information providing server 47 transmits the contribution information stored in the second database 46 together with the biological-related state information MX to the second terminal 200 and / or the first terminal 102 via the network NW. By checking the contribution information, the user can recognize what factors have led to the biological-related state information MX.

[0366] 26 to 34, according to the fourth embodiment, the estimation device 4 can obtain highly reliable organism-related state information MX by inputting at least the mind-body recovery capacity information D2 to the learning model TM1. This is because the learning model TM1 effectively uses the mind-body recovery capacity information D2, which is more reliable than conventional information, to estimate the subject's mind-body recovery.

[0367] In particular, in embodiment 4, the input information K2 does not need to include the results of diagnosis and evaluation by a medical professional, the results of tests using medical equipment such as a medical image diagnostic device, or the results of sampling and testing of bodily fluids such as blood. Therefore, in the stage of using the learning model TM1, the estimation device 4 can output the organism-related state information MX with high estimation accuracy without using the results of diagnosis and evaluation by a medical professional, the results of tests using medical equipment such as a medical image diagnostic device, or the results of sampling and testing of bodily fluids such as blood. Therefore, the organism-related state information MX can be obtained while reducing the burden on the subject. In this way, the organism-related state information MX can be obtained without reducing the subject's QOL.

[0368] In addition, the estimation device 4 can continuously acquire the biological state raw data A2 from the wearable device (biological state detection device 103) and / or the mobile terminal (first terminal 102). Therefore, the calculation and monitoring of the biological-related state information MX can be continuously performed. In particular, the biological state raw data A2 is automatically transmitted from the wearable device (biological state detection device 103) and / or the mobile terminal (first terminal 102) to the estimation system 40 (estimation device 4). As a result, the burden on the subject is further reduced. In other words, the biological-related state information MX can be obtained while reducing the burden on the subject.

[0369] Next, the estimation process of the biological-related state information M2 executed in step S53 of Fig. 30 will be described with reference to Fig. 31 and Fig. 35. Fig. 35 is a flowchart showing an example of the estimation process. The estimation process is executed by the estimation device 4 of Fig. 31. As shown in Fig. 35, the estimation process includes steps S71 to S79.

[0370] First, in step S71, the condition processing unit 413 acquires the subject's biological condition raw data A2 from the first database 45 (FIG. 29). The biological condition raw data A2 is stored in the storage unit 402.

[0371] Next, in step S72, the condition processing unit 413 performs processing (condition processing) on ​​the heart rate data, step count data, and exercise intensity data of the subject's biological condition raw data A2 in accordance with specific conditions. As a result, the condition processing unit 413 outputs extracted heart rate data, extracted step count data, and extracted exercise intensity data. The specific conditions are, for example, conditions related to sleep, conditions related to wakefulness, and conditions related to the time of day caused by the sun.

[0372] Next, in step S73, the statistical processing unit 412 performs statistical processing and difference processing on each of the extracted heart rate data, extracted step count data, and extracted exercise intensity data. Additionally, the statistical processing unit 412 performs ratio processing on each of the extracted step count data and extracted exercise intensity data. As a result, the statistical processing unit 412 outputs a heart rate statistical index based on the extracted heart rate data, a step count statistical index based on the extracted step count data, and an exercise intensity statistical index based on the extracted exercise intensity data. These indexes are stored in the storage unit 402 as input information K2. Note that the difference processing and ratio processing may be broadly considered to be statistical processing.

[0373] Next, in step S74, the mind-body estimation unit 411 calculates mind-body recovery capacity information D2 (a mind-body recovery index) based on the extracted heart rate data and the extracted step count data, or the extracted heart rate data and the extracted exercise intensity data (an example of a mind-body recovery capacity estimation process). The mind-body recovery capacity information D2 is stored in the storage unit 402 as input information K2.

[0374] Next, in step S75, the exercise responsiveness estimation unit 414 calculates an exercise responsiveness index based on the heart rate data and behavioral raw data (e.g., exercise intensity data) of the subject's biological condition raw data A2 (an example of an exercise responsiveness estimation process). The exercise responsiveness index is stored in the storage unit 402 as input information K2.

[0375] Next, in step S76, the estimation unit 42 inputs input information K2 including the features generated in steps S71 to S75 to the learning model TM1, and as a result, the learning model TM1 outputs output information L2.

[0376] Next, in step S77, the estimation unit 42 acquires output information L2 from the learning model TM1. The output information L2 includes the subject's biological-related state information M2. The output information L2 is stored in the storage unit 402.

[0377] Next, in step S78, the estimation unit 42 determines whether or not post-processing is required for the organism-related state information M2.

[0378] If it is determined in step S78 that post-processing is not required (NO), the estimation process is completed and the process returns to the main routine of Fig. 30. Cases where post-processing is not required include when the biological-related state information M2 is indicated by direct numerical output (e.g., Fig. 32), or when contribution information is not calculated.

[0379] On the other hand, if it is determined in step S78 that post-processing is necessary (YES), the process proceeds to step S79. Post-processing is necessary, for example, when performing binary classification output or multi-class classification output (e.g., FIGS. 33 and 34), or when calculating contribution information.

[0380] Next, in step S79, the post-processing unit 43 performs post-processing on the biological-related state information M2 and outputs biological-related state information M3. The biological-related state information M3 is stored in the memory unit 402 as output information N2. Furthermore, it is preferable that the interpretation unit 431 calculates contribution information used when the learning model TM1 estimates the biological-related state information M2. The contribution information is stored in the memory unit 402, for example, as output information L2 or output information N2. Then, the estimation process is completed, and the process returns to the main routine of FIG. 30.

[0381] 35 , according to the fourth embodiment, the estimation device 4 can obtain output information L2 (biological-related state information M2) with high estimation accuracy from the learning model TM1 by inputting input information K2 to the learning model TM1. Furthermore, the estimation device 4 can obtain output information N2 (biological-related state information M3) by performing post-processing on the output information L2.

[0382] In this case, the input information K2 only needs to include at least the subject's physical and mental recovery ability information D2. Therefore, the input information K2 does not need to include the results of diagnosis and evaluation by a medical professional, the results of tests using medical equipment such as a medical image diagnostic device, or the results of sampling and testing of bodily fluids such as blood. Therefore, according to the fourth embodiment, by using the learning model TM1, it is possible to estimate the organism-related state information MX with high estimation accuracy while improving the subject's QOL.

[0383] Next, the generation stage of the learning model TM1 will be described with reference to Figures 26, 36, and 37. Figure 36 is a block diagram showing an example configuration of a learning device 3 according to embodiment 4. As shown in Figure 36, the learning device 3 includes a processing unit 31, a communication unit 34, and a storage unit 35. The learning device 3 may also include an input unit 32 and an output unit 33. The hardware configurations of the processing unit 31, the communication unit 34, the storage unit 35, the input unit 32, and the output unit 33 are similar to the hardware configurations of the processing unit 400, the communication unit 401, the storage unit 402, the input unit 403, and the output unit 404 of the estimation device 4 in Figure 31, or the hardware configuration of the mind-body estimation device SS in Figure 24 or 25.

[0384] The storage unit 35 stores data and computer programs. The processing unit 31 includes a learning data acquisition unit 310 and a learning unit 311. For example, the processor of the processing unit 31 functions as the learning data acquisition unit 310 and the learning unit 311 by executing a computer program stored in the storage device of the storage unit 35.

[0385] 37 is a flowchart showing an example of a learning method by the learning device 3. The learning method is an example of a "learning model generation method" of the present disclosure. As shown in FIG. 37, the learning method includes steps S101 to S108.

[0386] First, in step S101, the training data acquisition unit 310 acquires multiple training data sets F1 from the training data creation device 2 ( FIG. 1 ). Specifically, the training data acquisition unit 310 acquires multiple training data sets F1 from the training database DBT ( FIG. 38 ). The storage unit 35 stores the multiple training data sets F1. Some of the multiple training data sets F1 are training data, another part is evaluation data, and still another part is test data.

[0387] Next, in step S102, the learning unit 311 prepares a learning model TM1 before learning. In the learning model TM1 before learning, various parameters are set to initial values.

[0388] Next, in step S103, the learning unit 311 acquires one learning data set F1 from the multiple learning data sets F1 stored in the storage unit 35. In this case, the learning data set F1 is training data.

[0389] Next, in step S104, the learning unit 311 inputs the feature information G1 included in the learning dataset F1 to the learning model TM1 before (or during) learning. As a result, output information is output from the learning model TM1 as an estimation result according to the machine learning algorithm.

[0390] Next, in step S105, the learning unit 311 performs machine learning by comparing the correct label B1 included in the learning dataset F1 with the output information output as the estimation result in step S104 and adjusting various parameters based on a machine learning algorithm and a predetermined adjustment method. In this way, the learning unit 311 causes the learning model TM1 to learn the correlation between the feature information G1 and the correct label B1. The predetermined adjustment method for the various parameters is not particularly limited, and may be, for example, the least squares method, maximum likelihood estimation, EM algorithm, gradient descent, backpropagation, or Bayesian estimation.

[0391] Next, in step S106, the learning unit 311 determines whether a learning termination condition is satisfied. The learning termination condition may be, for example, when the evaluation value of a loss function based on the correct label B1 and the output information output as the estimation result reaches a target value. Alternatively, the learning termination condition may be, for example, when the number of learning iterations (number of epochs) reaches a target number.

[0392] If it is determined in step S106 that the learning termination condition is not satisfied (NO), the process proceeds to step S103. Steps S103 to S105 are repeated until the learning termination condition is satisfied.

[0393] On the other hand, if it is determined in step S106 that the learning termination condition is satisfied (YES), the process proceeds to step S107.

[0394] Next, in step S107, the learning unit 311 adjusts the hyperparameters of the learning model TM1 based on the learning dataset F1 as evaluation data and the input values ​​input by the machine learning engineer via the input unit 32. The adjustment method in this case is not particularly limited, but may be, for example, grid search, random search, or Bayesian optimization.

[0395] Next, in step S108, the learning unit 311 evaluates the estimation accuracy of the learning model TM1 using the learning data set F1 as test data, and the learning method then ends.

[0396] As described above with reference to Figure 37, according to the fourth embodiment, the learning device 3 performs learning using the learning dataset F1 to generate a learning model TM1 that outputs output information L2 when input information K2 is input. That is, the learning device 3 repeats learning using multiple learning datasets F1 to generate a trained learning model TM1 having various trained parameters. The storage unit 35 stores the trained learning model TM1.

[0397] Next, the generation stage of the training data set F1 will be described with reference to FIGS. 26, 38, and 39. FIG. 38 is a block diagram showing an example configuration of a training data creation device 2 according to embodiment 4. As shown in FIG. 38, the training data creation device 2 includes a processing unit 21, a communication unit 24, and a storage unit 25. The training data creation device 2 may also include an input unit 22 and an output unit 23. The hardware configurations of the processing unit 21, the communication unit 24, the storage unit 25, the input unit 22, and the output unit 23 are similar to the hardware configurations of the processing unit 400, the communication unit 401, the storage unit 402, the input unit 403, and the output unit 404 of the estimation device 4 in FIG. 6, or the hardware configuration of the mind-body estimation device SS in FIG. 24 or 25.

[0398] The storage unit 25 stores data and computer programs. The storage unit 25 includes a training database DBT. In Fig. 38, the database is abbreviated as DB. The training database DBT stores a plurality of training data sets F1.

[0399] The processing unit 21 includes a mind-body estimation unit 211 and a correct label creation unit 216. The processing unit 21 may include one or more of a statistical processing unit 212, a condition processing unit 213, and an exercise reactivity estimation unit 214. For example, the processor of the processing unit 400 functions as the mind-body estimation unit 211, the correct label creation unit 216, the statistical processing unit 212, the condition processing unit 213, and the exercise reactivity estimation unit 214 by executing a computer program stored in the storage device of the storage unit 402.

[0400] The processing of the mind-body estimation unit 211, statistical processing unit 212, condition processing unit 213, and exercise reactivity estimation unit 214 is similar to the processing of the mind-body estimation unit 411, statistical processing unit 412, condition processing unit 413, and exercise reactivity estimation unit 414 of the estimation device 4 (Figure 31), respectively.

[0401] For example, in the description of the mind-body estimation unit 411, statistical processing unit 412, condition processing unit 413, and motor reactivity estimation unit 414 of the estimation device 4, the subject can be replaced with the learning subject, biological state raw data A2 with biological state raw data A1, input information K2 with feature information G1, output information L2 with correct answer label B1, mental and physical recovery ability information D2 with mental and physical recovery ability information D1, vital information H2 with vital information H1, behavioral information J2 with behavioral information J1, environmental information R2 with environmental information R1, self-reported information V2 with report information V1, and biological-related state information M2 with biological-related state information M1, thereby replacing the description of the mind-body estimation unit 211, statistical processing unit 212, condition processing unit 213, and motor reactivity estimation unit 214 of the learning data creation device 2.

[0402] Fig. 39 is a flowchart showing an example of a learning data creation method by the learning data creation device 2. As shown in Fig. 39, the learning data creation method includes steps S201 to S206.

[0403] First, in step S201, the processing unit 21 acquires the biological state raw data A1 and the correct answer information Z1 of the learning subject. The biological state raw data A1 and the correct answer information Z1 are stored in the storage unit 25.

[0404] Next, in step S202, the condition processing unit 213 performs processing (condition processing) on ​​the heart rate data, step count data, and exercise intensity data of the biological condition raw data A1 of the learning subject in accordance with specific conditions. As a result, the condition processing unit 213 outputs extracted heart rate data, extracted step count data, and extracted exercise intensity data. The specific conditions are, for example, conditions related to sleep, conditions related to wakefulness, and conditions related to the time of day caused by the sun.

[0405] Next, in step S203, the statistical processing unit 210 performs statistical processing and difference processing on each of the extracted heart rate data, extracted step count data, and extracted exercise intensity data. Additionally, the statistical processing unit 210 performs ratio processing on each of the extracted step count data and extracted exercise intensity data. As a result, the statistical processing unit 210 outputs a heart rate statistical index based on the extracted heart rate data, a step count statistical index based on the extracted step count data, and an exercise intensity statistical index based on the extracted exercise intensity data. These indexes are stored in the storage unit 25 as feature amount information G1. Note that the difference processing and ratio processing may be broadly considered to be statistical processing.

[0406] Next, in step S204, the mind-body estimation unit 211 calculates mind-body recovery capacity information D1 (a mind-body recovery index) based on the extracted heart rate data and the extracted step count data, or the extracted heart rate data and the extracted exercise intensity data (an example of a mind-body recovery capacity estimation process). The mind-body recovery capacity information D1 is stored in the storage unit 25 as feature amount information G1.

[0407] Next, in step S205, the exercise responsiveness estimation unit 214 calculates an exercise responsiveness index based on the heart rate data and behavioral raw data (e.g., exercise intensity data) of the biological state raw data A1 of the learning subject (an example of an exercise responsiveness estimation process). The exercise responsiveness index is stored in the storage unit 25 as feature amount information G1.

[0408] Next, in step S206, the correct label creation unit 216 creates a correct label B1 based on the correct information Z1 and associates it with the feature amount information G1 (correct label creation process). The correct label B1 includes the biological-related state information M1. The correct label B1 is stored in the storage unit 25. As an example, the correct information Z1 is a numerical value that directly indicates the mental, physical, or environmental state. Based on this example, first to third examples will be described.

[0409] As a first example, when the estimation device 4 directly outputs numerical values ​​(for example, in the case of Figure 32), the correct answer label creation unit 216 associates the correct answer information Z1 with the feature information G1 as the correct answer label B1 (biologically related state information M1).

[0410] As a second example, when the estimation device 4 performs binary classification output (e.g., the case of FIG. 33 ), the correct label creation unit 216 compares the correct information Z1 with a threshold and, based on the comparison result, classifies the correct information Z1 (mental, physical, or environmental state) into a first state or a second state. Then, the correct label creation unit 216 sets information indicating the first state (e.g., 0) or information indicating the second state (e.g., 1) to the correct label B1 (biological-related state information M1) in accordance with the classification result.

[0411] As a third example, when the estimation device 4 performs multi-class classification (for example, in the case of Figure 34), the correct label creation unit 216 sets an evaluation corresponding to a class for the correct label B1 (bio-related state information M1) depending on whether the correct information Z1 falls within the numerical range of any of multiple classes.

[0412] In the second and third examples, the correct answer information Z1 after binary classification or multi-class classification may be input via the input unit 22. In this case, the correct answer label creation unit 216 associates the correct answer information Z1 with the feature amount information G1 as the correct answer label B1, similarly to the first example.

[0413] When step S206 is completed, the training data creation method ends. Steps S201 to S206 are repeatedly executed to create multiple training data sets F1. The training data creation device 2 stores the training data sets F1 in the training database DBT. The training data creation device 2 may include one or more of environmental information R1, report information V1, and attribute information Q1 in the training data set F1.

[0414] As described above with reference to Figure 39, according to embodiment 4, the learning data creation device 2 can create a learning data set F1 that is suitable for the learning model TM1 that can estimate the biological-related state information M2 while improving the subject's QOL.

[0415] Here, the multiple learning data sets F1 used to train the learning model TM1 are preferably composed of "information representing the physical and mental states of multiple learning subjects belonging to the same group" or "information representing the environmental state that affects the physical and mental states of multiple learning subjects belonging to the same group." A group is a group of learning subjects that satisfy a predetermined condition related to their physical or mental state or their environmental state. The predetermined condition is, for example, a past occurrence of an abnormality in their physical or mental state, such as the onset of a disease or symptom, or a past stay in an environment with an abnormal state, such as staying in an environment with an abnormal atmosphere.

[0416] As such, it is preferable that the learning subjects when acquiring the learning dataset F1 belong to the same group. Furthermore, it is preferable that the subject from whom the input information K2 to be input into the learning model TM1 is acquired belongs to the same group as the learning subject. In other words, it is preferable that the subject, like the learning subject, also satisfies predetermined conditions regarding mental, physical, or environmental conditions. According to this preferable example, for example, it is possible to explain to the subject the validity of estimating the organism-related state information M2 using the learning model TM1, to explain the reliability of the estimation results, and to improve the acceptability of the estimation results. Furthermore, by acquiring the learning dataset F1 from a learning subject who belongs to the same group as the subject, it is possible to improve the estimation accuracy of the organism-related state information M2.

[0417] The fourth embodiment of the present disclosure has been described above. In the fourth embodiment, the rheumatoid arthritis index information, blood index information, systemic connective tissue disorder index information, and non-organ-specific systemic autoimmune disease index information may be excluded from the organism-related condition information M1, MX. In other words, the organism-related condition information M1, MX may not include the rheumatoid arthritis index information, blood index information, systemic connective tissue disorder index information, and non-organ-specific systemic autoimmune disease index information. The rheumatoid arthritis index information is an index representing the symptoms or activity of rheumatoid arthritis, or information regarding an index representing the symptoms or activity of rheumatoid arthritis. The blood index information is information that directly or indirectly indicates a blood index. The blood index is an index that quantitatively indicates a blood component, the state of a blood component, a substance in the blood, or the state of a substance in the blood. The systemic connective tissue disorder indicator information is information on an indicator that directly or indirectly indicates the symptoms or activity of a systemic connective tissue disorder, or information on an indicator that directly or indirectly indicates an effect caused by the symptoms or activity of a systemic connective tissue disorder. The non-organ-specific systemic autoimmune disease indicator information is information on an indicator that directly or indirectly indicates the symptoms or activity of a non-organ-specific systemic autoimmune disease, or information on an indicator that directly or indirectly indicates an effect caused by the symptoms or activity of a non-organ-specific systemic autoimmune disease. Systemic connective tissue disorders and non-organ-specific systemic autoimmune diseases include systemic lupus erythematosus.

[0418] Although the preferred embodiments and modifications of the present disclosure have been described in detail above with reference to the accompanying drawings, the technical scope of the present disclosure is not limited to such examples. It is clear that a person skilled in the art of the present disclosure can conceive of various modified or altered examples within the scope of the technical idea described in the claims, and it is understood that these also naturally fall within the technical scope of the present disclosure.

[0419] The devices or systems described herein may be implemented as a single device, or may be implemented as multiple devices (e.g., cloud servers) partially or entirely connected via a network. For example, some or all of the mind-body estimation unit 411, statistical processing unit 412, condition processing unit 413, exercise reactivity estimation unit 414, estimation unit 42, and post-processing unit 43 in FIG. 31 may be implemented by the same computer or server. For example, the mind-body estimation unit 411, statistical processing unit 412, condition processing unit 413, exercise reactivity estimation unit 414, estimation unit 42, and post-processing unit 43 in FIG. 31 may each be implemented by a separate computer or server. This also applies to the components of the learning device 3 in FIG. 36 and the components of the training data creation device 2 in FIG. 38. For example, the relay server 44 or the estimation device 4 in FIG. 29 may have the functionality of the information providing server 47. For example, the first database 45 and the second database 46 may be implemented by a single computer or server. For example, the estimation unit 42 may have the functionality of the relay server 44.

[0420] The series of processes performed by the devices described herein may be implemented using software, hardware, or a combination of software and hardware. A computer program for implementing each function of the processing units 400, 31, and 21 may be created and installed on a PC or the like. A computer-readable storage medium storing such a computer program may also be provided. Examples of the storage medium include a magnetic disk, an optical disk, a magneto-optical disk, and a flash memory. The computer program may also be distributed, for example, via a network without using a storage medium. For example, the training database DBT may be located outside the training data creation device 2. For example, the training model TM1 may be located outside the estimation device 4.

[0421] The processes described herein using flowchart diagrams do not necessarily have to be performed in the order shown, some process steps may be performed in parallel, additional process steps may be employed, and some process steps may be omitted.

[0422] 4, 5, 16, 35, 37, and 39, the processing units SY(SS), 400, 31, and 21 execute the computer programs stored in the storage units KI(SS), 402, 35, and 25 to perform the steps included in the information processing method, estimation processing (estimation method), learning method, and learning data creation method. In other words, the computer programs cause the processing units SY(SS), 400, 31, and 21 to execute the steps included in the information processing method, estimation processing (estimation method), learning method, and learning data creation method. The processing units SY(SS), 400, 31, and 21 correspond to examples of the "computer" in the present disclosure. In other words, the computer program product, when executed by the processing units SY(SS), 400, 31, and 21, realizes the steps included in the information processing method, estimation processing (estimation method), learning method, and learning data creation method.

[0423] The estimation device 4 in FIG. 31 may not include, for example, all or some of the statistical processing unit 412, the condition processing unit 413, and the exercise reactivity estimation unit 414. The estimation process in FIG. 35 may not include, for example, all or some of steps S72, S73, and S75. Furthermore, the estimation process in FIG. 35 may not include, for example, steps S78 and S79. The training data creation device 2 in FIG. 38 may not include, for example, all or some of the statistical processing unit 212, the condition processing unit 213, and the exercise reactivity estimation unit 214. The training data creation method in FIG. 39 may not include, for example, all or some of steps S202, S203, and S205.

[0424] The learning device 3 may generate a new learning model (distilled model) by performing learning using the input information K2 input to the learning model TM1 and the output information L2 or N2 output from the learning model TM1 as a learning data set. The input information K2 and the output information L2 and N2 are subject information for the learning model TM1 in the utilization stage. However, in generating a distilled model, the input information K2 and the output information L2 and N2 constituting the learning data set correspond to information about the learning subject for the learning device 3.

[0425] Furthermore, the effects described herein are merely descriptive or exemplary and are not limiting. That is, the technology according to the present disclosure may achieve other effects in addition to or in place of the above-described effects that would be apparent to a person skilled in the art from the description of this specification.

[0426] The following configurations also fall within the technical scope of the present disclosure.

[0427] (Item 1) An estimation device for estimating bio-related state information of a subject, comprising: an estimation unit that inputs input information to a learning model and acquires output information from the learning model; and a memory unit that stores the output information, wherein the bio-related state information includes information representing the subject's mental and physical state or information representing an environmental state that affects the subject's mental and physical state, the input information includes at least mental and physical recovery ability information that indicates the degree of speed of the subject's mental and physical recovery from the end of exercise, and the output information includes the bio-related state information.

[0428] (Item 2) The estimation device according to item 1, wherein the mind-body recovery capacity information indicates a degree of speed of recovery of the subject's heart rate from the end of exercise.

[0429] (Item 3) The estimation device according to item 1 or 2, wherein the mental and physical recovery capacity information is calculated based on a heart rate change amount and an exercise heart rate of the subject, and the heart rate change amount indicates a difference between the exercise heart rate and the resting heart rate of the subject.

[0430] (Item 4) The mental and physical recovery ability information includes parameter values ​​that realize a model function that approximates the relationship between the subject's heart rate change and their exercise heart rate, and the heart rate change indicates the difference between the exercise heart rate, which is the heart rate during a first unit of time while the subject is exercising, and the resting heart rate, which is the heart rate during a second unit of time following the first unit of time while the subject is at rest. This is the estimation device described in item 1 or 2.

[0431] (Item 5) The estimation device according to Item 4, wherein the model function includes a line equation, and the parameter value indicates a slope of the line represented by the line equation.

[0432] (Item 6) The estimation device according to Item 5, wherein a larger absolute value of the slope indicates a faster recovery rate of the subject's heart rate.

[0433] (Item 7) The estimation device according to Item 4, wherein the model function includes a curve equation that approximates the relationship between the amount of change in heart rate and the heart rate during exercise, the parameter value indicates a slope of a line expressed by a linear equation that constitutes an input variable of the curve equation, the curve equation is a function fitted to a virtual curve, the virtual curve is a virtual curve obtained based on representative values ​​of the amount of change in heart rate obtained for each actual measured value of the heart rate during exercise, and each of the representative values ​​is a representative value obtained based on a distribution law from the distribution of actual measured values ​​of the amount of change in heart rate, corresponding to the actual measured value of the heart rate during exercise.

[0434] (Item 8) The estimation device according to any one of Items 1 to 7, wherein the biological-related state information includes information representing the subject's mental and physical state, and the information representing the subject's mental and physical state includes information representing a physical state, information representing a mental state, information regarding the intake of a specific component, or information regarding quality of life, and the information regarding quality of life indicates the physical state, the mental state, or quality of life related to the intake of a specific component.

[0435] (Item 9) The estimation device according to Item 8, wherein the information representing the physical condition includes information representing a condition related to a disease of the circulatory system, information representing a condition related to a disease of the respiratory function system, information representing a condition related to a disease of the biocontrol system, information representing subjective symptoms related to the physical condition, or information representing a physical condition that can be indicated by the results of a biopsy.

[0436] (Item 10) The estimation device according to Item 8, wherein the information representing the mental state includes information representing a state related to a mental illness or information representing a state related to a mental stress.

[0437] (Item 11) The estimation device according to Item 8, wherein the information on the intake of the specific component includes information on the intake of a medicine or information on the intake of a luxury item.

[0438] (Item 12) The estimation device according to any one of Items 1 to 11, wherein the biological-related state information includes information representing an environmental state that affects the mental and physical state of the subject, and the information representing the environmental state that affects the mental and physical state of the subject includes information regarding atmospheric abnormalities or information regarding a mentally stressful environment.

[0439] (Item 13) The estimation device according to any one of Items 1 to 12, wherein the input information further includes one or more of vital sign information of the subject, behavioral information of the subject, information about the subject's environment, self-reported information about the subject's physical and mental state, and attribute information of the subject.

[0440] (Item 14) The estimation device according to any one of items 1 to 13, wherein the mental and physical recovery capacity information includes information on mental and physical recovery capacity for a period indicated by specific conditions, and the specific conditions include at least one of a condition related to sleep, a condition related to wakefulness, and a condition related to a time of day caused by the sun.

[0441] (Item 15) The estimation device according to any one of Items 1 to 14, wherein the learning model is constructed by learning using a learning dataset, and the learning dataset includes at least mental and physical recovery ability information indicating the degree of speed of the learning subject's mental and physical recovery from the end of exercise, and biologically-related state information of the learning subject, and the biologically-related state information of the learning subject includes information representing the mental and physical state of the learning subject, or information representing the state of the environment that affects the mental and physical state of the learning subject.

[0442] (Item 16) A learning model that causes a computer to function to estimate bio-related state information of a subject, the learning model causing the computer to function to input input information and output output information, the bio-related state information including information representing the subject's mental and physical state or information representing an environmental state that affects the subject's mental and physical state, the input information including at least mental and physical recovery ability information that indicates the degree of speed of the subject's mental and physical recovery from the end of exercise, and the output information including the bio-related state information.

[0443] (Item 17) An estimation method for estimating bio-related state information of a subject, comprising the steps of: inputting input information into a learning model; and acquiring output information from the learning model to which the input information has been input, wherein the bio-related state information includes information representing the subject's mental and physical state, or information representing an environmental state that affects the subject's mental and physical state, the input information includes at least mental and physical recovery ability information indicating the degree of speed of the subject's mental and physical recovery from the end of exercise, and the output information includes the bio-related state information.

[0444] (Item 18) A computer program that causes a computer to execute the estimation method according to Item 17.

[0445] (Item 19) A learning model generation method comprising the steps of: acquiring a learning dataset; and performing learning using the learning dataset to generate a learning model that outputs output information when input information is input, wherein the learning dataset includes at least mind-body recovery ability information indicating the degree of speed of the learning subject's mental and physical recovery from the end of exercise, and bio-related state information of the learning subject, wherein the bio-related state information of the learning subject includes information representing the learning subject's mental and physical state, or information representing an environmental state that affects the learning subject's mental and physical state, wherein the input information includes at least mind-body recovery ability information indicating the degree of speed of the subject's mental and physical recovery from the end of exercise, and the output information includes the bio-related state information of the subject, and the bio-related state information of the subject includes information representing the subject's mental and physical state, or information representing an environmental state that affects the subject's mental and physical state.

[0446] (Item 20) A computer program that causes a computer to execute the learning model generation method according to Item 19.

[0447] (Item 21) An information processing device comprising: a processing unit that calculates mental and physical recovery ability information indicating the speed of the subject's mental and physical recovery from the end of exercise based on the subject's heart rate change and exercise heart rate; and a memory unit that stores the mental and physical recovery ability information, wherein the heart rate change indicates the difference between the subject's exercise heart rate and resting heart rate.

[0448] (Item 22) The information processing device according to item 21, wherein the processing unit calculates the amount of change in heart rate, which is the difference between the exercise heart rate, which is the heart rate within a first unit time while the subject is exercising, and the resting heart rate, which is the heart rate within a second unit time following the first unit time while the subject is at rest; approximates the relationship between the amount of change in heart rate and the exercise heart rate using a model function; and acquires parameter values ​​that realize the model function as the mind and body recovery ability information.

[0449] (Item 23) The information processing device according to item 22, wherein the model function includes a linear equation, and the parameter value indicates a slope of a straight line represented by the linear equation.

[0450] (Item 24) The information processing device according to Item 23, wherein a larger absolute value of the slope of the linear equation indicates a faster recovery of the subject's heart rate.

[0451] (Item 25) The information processing device described in Item 22, wherein the model function includes a curve equation that approximates the relationship between the amount of change in heart rate and the heart rate during exercise, the parameter value indicates the slope of a line represented by a linear equation that constitutes an input variable of the curve equation, the curve equation is a function fitted to a virtual curve, the virtual curve is a virtual curve obtained based on a representative value of the amount of change in heart rate obtained for each actual measured value of the heart rate during exercise, and each of the representative values ​​is a representative value obtained based on a distribution law from the distribution of the actual measured values ​​of the amount of change in heart rate, corresponding to the actual measured value of the heart rate during exercise.

[0452] (Item 26) The information processing device according to any one of items 21 to 25, wherein the processing unit estimates a degree of mental and physical recovery of the subject based on the mental and physical recovery capacity information.

[0453] (Item 27) An information processing method for calculating mental and physical recovery ability information indicating the speed of a subject's mental and physical recovery after the end of exercise, comprising: (1) a step of acquiring information on the subject's heart rate; and (2) a step of calculating the mental and physical recovery ability information based on the subject's heart rate change amount and exercise heart rate, wherein the heart rate change amount indicates the difference between the subject's exercise heart rate and resting heart rate.

[0454] (Item 28) A computer program that causes a computer to execute the information processing method according to Item 27.

[0455] (Item 29) A mind-body estimation index is a parameter value that realizes a model function that approximates the relationship between the heart rate of a subject for each unit time, the amount of change in heart rate being the difference between the heart rate in a first unit time when the subject is exercising and the heart rate in a second unit time following the first unit time when the subject is at rest, and the heart rate change amount.

[0456] (Item 30) A mind-body estimation index is a heart rate change amount, which is the difference between the heart rate during a first unit time during which the subject is exercising and the heart rate during a second unit time following the first unit time during which the subject is at rest, for the heart rate of the subject for each unit time, and a gradient of a linear equation of the distribution law, which is obtained based on a curve equation of the distribution law, which is approximated to a virtual curve created based on expected values ​​corresponding to each of the heart rates, obtained from a distribution of multiple heart rate changes corresponding to each of the heart rates, which is created according to a distribution law in a two-dimensional coordinate space to which the relationship between the heart rates is assigned.

[0457] (Item 31) A method for calculating indices for mind-body estimation, comprising: (1) a step of calculating the heart rate of a subject for each unit time; (2) a step of calculating a heart rate change amount, which is the difference between the heart rate within a first unit time in which the subject is exercising and the heart rate within a second unit time subsequent to the first unit time in which the subject is at rest; (3) a step of approximating the relationship between the heart rate and the heart rate change amount using a model function; and (4) a step of acquiring parameter values ​​that realize the model function.

[0458] (Item 32) A method for calculating an index for mind-body estimation, comprising: (1) a step of calculating a heart rate of a subject for each unit time; (2) a step of calculating a heart rate change amount, which is the difference between the heart rate within a first unit time during which the subject is exercising and the heart rate within a second unit time subsequent to the first unit time during which the subject is at rest; (3) a step of assigning the relationship between the heart rate and the heart rate change amount to a two-dimensional coordinate space; (4) a step of creating a distribution of a plurality of heart rate change amounts corresponding to each heart rate in the two-dimensional coordinate space according to a distribution law; (5) a step of obtaining an expected value from the distribution; (6) a step of creating a virtual curve based on the expected value corresponding to each heart rate; (7) a step of approximating the virtual curve to a curve equation of the distribution law; and (8) a step of obtaining a slope of a linear equation of the distribution law based on the curve equation.

[0459] (Item 33) A mind-body estimation method, comprising a step of estimating a degree of mental and physical recovery of the subject using the mind-body estimation index according to item 29 or 30.

[0460] (Item 34) A mind-body estimation device including an estimation unit that estimates a degree of recovery of the mind and body of the subject using the indices for mind-body estimation according to item 29 or 30.

[0461] The present disclosure provides an estimation device, a learning model, an estimation method, a learning model generation method, an information processing device, an information processing method, and a computer program, and has industrial applicability.

[0462] SSS...mind-body estimation system, WT...wearable terminal, SS...mind-body estimation device, NW...network, SY(SS)...processing unit, KI(SS)...storage unit, 2...learning data creation device, 3...learning device, 4...estimation device, 42...estimation unit, 402...storage unit, 411...mind-body estimation unit, 412...statistical processing unit, 413...condition processing unit, 414...motor reactivity estimation unit, 431...interpretation unit, TM1...learning model

Claims

1. An estimation device for estimating bio-related state information of a subject, comprising: an estimation unit that inputs input information to a learning model and obtains output information from the learning model; and a memory unit that stores the output information, wherein the bio-related state information includes information representing the subject's mental and physical state or information representing the state of the environment that affects the subject's mental and physical state, the input information includes at least mental and physical recovery ability information that indicates the degree of speed of the subject's mental and physical recovery from the end of exercise, and the output information includes the bio-related state information.

2. The estimation device according to claim 1, wherein the mental and physical recovery capacity information indicates the degree of speed of recovery of the subject's heart rate after the end of exercise.

3. The estimation device of claim 1 or claim 2, wherein the mental and physical recovery ability information is calculated based on the subject's heart rate change and exercise heart rate, and the heart rate change indicates the difference between the subject's exercise heart rate and resting heart rate.

4. The estimation device of claim 1 or claim 2, wherein the mental and physical recovery ability information includes parameter values ​​that realize a model function that approximates the relationship between the subject's heart rate change and exercise heart rate, and the heart rate change indicates the difference between the exercise heart rate, which is the heart rate within a first unit of time while the subject is exercising, and the resting heart rate, which is the heart rate within a second unit of time following the first unit of time while the subject is at rest.

5. The estimation device according to claim 4, wherein the model function includes a line equation, and the parameter value indicates a slope of the line represented by the line equation.

6. The estimation device according to claim 5, wherein a larger absolute value of the slope indicates a faster recovery of the subject's heart rate.

7. The estimation device according to claim 4, wherein the model function includes a curve equation that approximates the relationship between the amount of change in heart rate and the heart rate during exercise, the parameter value indicates the slope of a line expressed by a linear equation that constitutes an input variable of the curve equation, the curve equation is a function fitted to a virtual curve, the virtual curve is a virtual curve obtained based on a representative value of the amount of change in heart rate obtained for each actual measured value of the heart rate during exercise, and each of the representative values ​​is a representative value obtained based on a distribution law from the distribution of the actual measured values ​​of the amount of change in heart rate, corresponding to the actual measured value of the heart rate during exercise.

8. The estimation device described in claim 1 or claim 2, wherein the biological-related state information includes information representing the subject's physical and mental state, and the information representing the subject's physical and mental state includes information representing the physical state, information representing the mental state, information regarding the intake of specific components, or information regarding quality of life, and the information regarding quality of life indicates the physical state, mental state, or quality of life related to the intake of specific components.

9. The estimation device described in claim 8, wherein the information representing the physical condition includes information representing a condition related to a disease of the circulatory system, information representing a condition related to a disease of the respiratory function system, information representing a condition related to a disease of the biocontrol system, information representing subjective symptoms related to the physical condition, or information representing a physical condition that can be indicated by the results of a biopsy.

10. The estimation device according to claim 8, wherein the information representing the mental state includes information representing a state related to a mental illness or information representing a state related to mental stress.

11. The estimation device according to claim 8, wherein the information regarding the ingestion of the specific component includes information regarding the ingestion of a medicine or information regarding the ingestion of a luxury item.

12. An estimation device as described in claim 1 or claim 2, wherein the biological-related state information includes information representing an environmental state that affects the subject's mental and physical state, and the information representing an environmental state that affects the subject's mental and physical state includes information regarding atmospheric abnormalities or information regarding a mentally stressful environment.

13. An estimation device as described in claim 1 or claim 2, wherein the input information further includes one or more of vital information of the subject, behavioral information of the subject, information about the subject's environment, self-reported information about the subject's physical and mental state, and attribute information of the subject.

14. An estimation device as described in claim 1 or claim 2, wherein the mental and physical recovery ability information includes information on mental and physical recovery ability for a period indicated by specific conditions, and the specific conditions include at least one of conditions related to sleep, conditions related to wakefulness, and conditions related to the time of day caused by the sun.

15. The estimation device described in claim 1 or claim 2, wherein the learning model is constructed by learning using a learning dataset, and the learning dataset includes at least mental and physical recovery ability information indicating the degree of speed of the learning subject's mental and physical recovery from the end of exercise, and biologically-related state information of the learning subject, and the biologically-related state information of the learning subject includes information representing the mental and physical state of the learning subject, or information representing the state of the environment that affects the mental and physical state of the learning subject.

16. A learning model that causes a computer to function to estimate bio-related state information of a subject, the learning model causing the computer to function to input input information and output output information, the bio-related state information including information representing the subject's mental and physical state or information representing the state of the environment that affects the subject's mental and physical state, the input information including at least mental and physical recovery ability information indicating the degree of speed of the subject's mental and physical recovery from the end of exercise, and the output information including the bio-related state information.

17. An estimation method for estimating bio-related state information of a subject, comprising the steps of: inputting input information into a learning model; and acquiring output information from the learning model to which the input information has been input, wherein the bio-related state information includes information representing the subject's mental and physical state, or information representing the state of the environment that affects the subject's mental and physical state, the input information includes at least mental and physical recovery ability information that indicates the degree of speed of the subject's mental and physical recovery from the end of exercise, and the output information includes the bio-related state information.

18. A computer program that causes a computer to execute the estimation method according to claim 17.

19. A learning model generation method comprising the steps of: acquiring a learning dataset; and performing learning using the learning dataset to generate a learning model that outputs output information when input information is input, wherein the learning dataset includes at least mental and physical recovery ability information indicating the degree of speed of the learning subject's mental and physical recovery from the end of exercise, and biological-related state information of the learning subject, wherein the biological-related state information of the learning subject includes information representing the learning subject's mental and physical state, or information representing an environmental state that affects the learning subject's mental and physical state, wherein the input information includes at least mental and physical recovery ability information indicating the degree of speed of the subject's mental and physical recovery from the end of exercise, and the output information includes the biological-related state information of the subject, and the biological-related state information of the subject includes information representing the subject's mental and physical state, or information representing an environmental state that affects the subject's mental and physical state.

20. A computer program that causes a computer to execute the learning model generation method according to claim 19.

21. An information processing device comprising: a processing unit that calculates mental and physical recovery ability information indicating the speed of the subject's mental and physical recovery since the end of exercise based on the subject's heart rate change and heart rate during exercise; and a memory unit that stores the mental and physical recovery ability information, wherein the heart rate change indicates the difference between the subject's heart rate during exercise and resting heart rate.

22. The information processing device of claim 21, wherein the processing unit calculates the amount of change in heart rate, which is the difference between the exercise heart rate, which is the heart rate within a first unit time while the subject is exercising, and the resting heart rate, which is the heart rate within a second unit time following the first unit time while the subject is at rest; approximates the relationship between the amount of change in heart rate and the exercise heart rate using a model function; and obtains parameter values ​​that realize the model function as the mind and body recovery ability information.

23. The information processing device according to claim 22, wherein the model function includes a linear equation, and the parameter value indicates a slope of the line represented by the linear equation.

24. The information processing device according to claim 23, wherein a larger absolute value of the slope of the linear equation indicates a faster recovery of the subject's heart rate.

25. The information processing device of claim 22, wherein the model function includes a curve equation that approximates the relationship between the amount of change in heart rate and the heart rate during exercise, the parameter value indicates the slope of a line represented by a linear equation that constitutes an input variable of the curve equation, the curve equation is a function fitted to a virtual curve, the virtual curve is a virtual curve obtained based on a representative value of the amount of change in heart rate obtained for each actual measured value of the heart rate during exercise, and each of the representative values ​​is a representative value obtained based on a distribution law from the distribution of the actual measured values ​​of the amount of change in heart rate, corresponding to the actual measured value of the heart rate during exercise.

26. An information processing device according to claim 21 or 22, wherein the processing unit estimates the degree of mental and physical recovery of the subject based on the mental and physical recovery capacity information.

27. An information processing method for calculating mental and physical recovery ability information indicating the speed of a subject's mental and physical recovery after the end of exercise, comprising: (1) a step of acquiring information on the subject's heart rate; and (2) a step of calculating the mental and physical recovery ability information based on the subject's heart rate change amount and exercise heart rate, wherein the heart rate change amount indicates the difference between the subject's exercise heart rate and resting heart rate.

28. A computer program that causes a computer to execute the information processing method according to claim 27.

Images