Computer program and fatigue determination method

Abstract

This invention provides a method for generating a learning model, a computer program, and a fatigue determination method for obtaining information that serves as the basis for determining a person's fatigue. [Solution] Training data is acquired that includes motion data representing the movements of a person exercising and heart rate data based on the person's heart rate. Based on the training data, a learning model is generated that outputs heart rate data when motion data is input.

Description

Technical Field

[0001] The present invention relates to a learning model generation method, a computer program, and a fatigue determination method.

Background Art

[0002] Conventionally, in sports such as soccer, it has been practiced to measure the state of a person during exercise using a sensor worn by the person. For example, a person's heart rate or position information is measured using a sensor. Using the measured information, information representing a person's behavior, such as the distance traveled by the person during a match, can be calculated to visualize the person's behavior. Behavior can similarly be visualized for people performing exercise other than sports, such as rehabilitation or physically active games. Patent Document 1 discloses an example of a technique for measuring the state of a person during exercise using a sensor.

Prior Art Literature

Patent Literature

[0003]

Patent Document 1

Summary of the Invention

Problem to be Solved by the Invention

[0004] Although it is possible to visualize a person's behavior using a sensor, there are problems with the use of the visualized information. On the other hand, injuries to people who exercise, such as athletes, are a major problem. The occurrence of injuries is related to a person's fatigue. In order to prevent injuries, it is required to determine a person's fatigue. To determine fatigue, it is necessary to obtain information that serves as a basis for determining fatigue.

[0005] The present invention has been made in view of these circumstances, and its purpose is to provide a method for generating a learning model, a computer program, and a fatigue determination method for obtaining information that serves as the basis for determining a person's fatigue from information obtained using sensors attached to a person. [Means for solving the problem]

[0006] The method for generating a learning model according to the present invention is characterized by acquiring training data that includes motion data representing the movements of a person performing exercise and heart rate data based on the person's heart rate, and generating a learning model that outputs heart rate data when motion data is input, based on the training data.

[0007] The computer program according to the present invention is characterized by causing a computer to perform the following processes: acquire motion data representing the movements of a person exercising, and first heart rate data based on the heart rate of the person; input the acquired motion data into a learning model that outputs heart rate data when motion data is input; acquire second heart rate data output by the learning model; and output the difference between the first heart rate data and the second heart rate data.

[0008] The fatigue determination method according to the present invention is characterized by using a sensor attached to a person to measure the position and heart rate of the person exercising, acquiring motion data representing the movements of the person exercising and first heart rate data based on the heart rate based on the position and heart rate measured using the sensor, inputting the acquired motion data to a learning model that outputs heart rate data when motion data is input, acquiring second heart rate data output by the learning model, and outputting information regarding the person's fatigue according to the first heart rate data and the second heart rate data.

[0009] In one embodiment of the present invention, a learning model is generated that outputs heart rate data when motion data is input, by learning using training data that includes motion data representing the movements of a person exercising and heart rate data based on heart rate. By measuring a person's movements, acquiring motion data, inputting the motion data into the learning model, and acquiring the heart rate data output by the learning model, heart rate data can be estimated. Since a person's heart rate changes according to fatigue, the estimated heart rate data can serve as a basis for determining a person's fatigue.

[0010] In one embodiment of the present invention, first heart rate data based on actually measured heart rate and second heart rate data estimated using a learning model are acquired. The difference between the first and second heart rate data indicates that the burden on the body is different from normal. Therefore, by outputting the difference between the first and second heart rate data, information that visualizes a person's fatigue can be output. [Effects of the Invention]

[0011] In this invention, it becomes possible to estimate a person's heart rate data from their motion data using a learning model. The invention offers excellent effects, such as being able to determine a person's fatigue level based on the estimated heart rate data. [Brief explanation of the drawing]

[0012] [Figure 1] This is a schematic diagram showing an example configuration of a learning model generation system for generating learning models. [Figure 2] This is a block diagram showing the first example of the configuration of a sensor device. [Figure 3] This is a block diagram showing a second example of the configuration of a sensor device. [Figure 4] This is a block diagram showing an example of the internal functional configuration of a learning model generation device. [Figure 5] This is a conceptual diagram showing an example of the content of measurement data. [Figure 6]It is a flowchart illustrating an example of a processing procedure for generating a learning model. [Figure 7] It is a conceptual diagram showing an example of content of motion data. [Figure 8] It is a conceptual diagram illustrating functions of a learning model. [Figure 9] It is a schematic diagram showing a configuration example of a fatigue determination system that determines fatigue of an athlete. [Figure 10] It is a block diagram showing an example of an internal functional configuration of a fatigue determination apparatus. [Figure 11] It is a flowchart illustrating an example of a processing procedure for determining fatigue of an athlete. [Figure 12] It is a conceptual diagram showing an example of content of input data. [Figure 13] It is a schematic diagram showing an example of an image displayed by the fatigue determination apparatus. [Figure 14] It is a schematic diagram showing an example of an image displayed by the fatigue determination apparatus for outputting differences and cumulative sums of differences relating to a plurality of athletes. [Figure 15] It is a flowchart illustrating an example of a processing procedure for determining a physical condition of an athlete. [Figure 16] It is a schematic diagram showing an example of an image including exercise intensity and fatigue degree for a plurality of athletes. [Figure 17] It is a schematic diagram showing an example of an image including a list of exercise intensity, fatigue degree and physical condition for a plurality of athletes. [Figure 18] It is a schematic diagram showing an example of an image including exercise intensity, fatigue degree and physical condition for a specific athlete. DETAILED DESCRIPTION OF THE INVENTION

[0013] Hereinafter, the present invention will be specifically described with reference to the drawings showing embodiments thereof. <Embodiment 1> In this embodiment, a learning model is used to determine the fatigue level of an athlete performing sports. First, the method for generating the learning model will be described. Figure 1 is a schematic diagram showing an example configuration of a learning model generation system 200 for generating a learning model. The learning model generation system 200 includes a sensor device 1 attached to an athlete 10 performing sports, and a learning model generation device 2 that performs information processing for generating a learning model. The sensor device 1 is a wearable sensor attached to the athlete 10. For example, the sensor device 1 is part of clothing or equipment. The sensor device 1 may also be attached to the athlete 10's body. Using the sensor device 1, the athlete 10's position, heart rate, and step count are measured, and the learning model generation device 2 acquires positional information indicating the athlete 10's position, heart rate, and step count, and performs information processing for generating a learning model.

[0014] Figure 2 is a block diagram showing a first example of the configuration of the sensor device 1. The sensor device 1 comprises a control unit 11, a position sensor 12, a heart rate sensor 13, a step counter 18, a storage unit 14, and an interface unit 15. The control unit 11 controls each part of the sensor device 1. For example, the control unit 11 is configured using a processor. The control unit 11 may also perform time measurement processing. The storage unit 14 stores programs or data necessary for the operation of the sensor device 1. For example, the storage unit 14 is a non-volatile memory. For example, the storage unit 14 stores sensor identification information specific to the sensor device 1 for identifying the sensor device 1.

[0015] The position sensor 12 is a sensor that measures the position of the sensor device 1. The position sensor 12 measures its position using a positioning system such as GPS (Global Positioning System). For example, the position sensor 12 measures its position by receiving signals transmitted from positioning satellites such as GPS satellites. The control unit 11 may perform part of the processing for measuring the position. The position sensor 12 measures the position of the player 10 by measuring the position of the sensor device 1.

[0016] The heart rate sensor 13 measures the heart rate of the athlete 10 over a predetermined period of time. For example, the heart rate sensor 13 measures the heart rate over a 20-second period. For example, the heart rate sensor 13 is an optical sensor. The step counter 18 measures the number of steps. The number of steps is the number of steps taken within a predetermined period of time. For example, the step counter 18 has an acceleration sensor and performs a process to determine the number of steps according to the measurement result of the acceleration sensor. The sensor device 1 may also be configured such that the control unit 11 performs the process of determining the number of steps according to the measurement result of the acceleration sensor.

[0017] The interface unit 15 is equipped with a removable portable memory 16 and writes data to the portable memory 16. The control unit 11 stores data including position information indicating the position measured by the position sensor 12, heart rate measured by the heart rate sensor 13, and step count measured by the step counter 18 in the portable memory 16 via the interface unit 15. The data including position information, heart rate, and step count also includes information indicating the time when the position was measured, the time when the heart rate was measured, and the time when the step count was measured. Furthermore, the data including position information, heart rate, and step count also includes sensor identification information. The control unit 11 stores data including position information, heart rate, and step count obtained over multiple periods in the portable memory 16 via the interface unit 15.

[0018] The portable memory 16 detaches from the interface unit 15 and is mounted on a memory reader 31 located outside the sensor device 1. The memory reader 31 reads the data stored in the portable memory 16, including location information, heart rate, and step count. Alternatively, the sensor device 1 may be configured such that the interface unit 15 is directly mounted on the memory reader 31 without using the portable memory 16. In this configuration, the control unit 11 stores the location information, heart rate, and step count in the storage unit 14, and the memory reader 31 reads the data, including location information, heart rate, and step count, through the interface unit 15.

[0019] Figure 3 is a block diagram showing a second example of the configuration of the sensor device 1. The sensor device 1 comprises a control unit 11, a position sensor 12, a heart rate sensor 13, a storage unit 14, and a transmission unit 17. The control unit 11, storage unit 14, position sensor 12, and heart rate sensor 13 are the same as in the first example. The transmission unit 17 transmits data to a receiving device 32 located outside the sensor device 1 using wired or wireless communication. The control unit 11 transmits data including position information, heart rate, and step count obtained over multiple periods from the transmission unit 17 to the receiving device 32.

[0020] The sensor device 1 may be configured to include both an interface unit 15 and a transmission unit 17. The learning model generation system 200 may be configured to measure the position of the sensor device 1 by a method other than using a positioning system. For example, the sensor device 1 may transmit a signal to the outside, a receiving device located outside the sensor device 1 may receive the signal from the sensor device 1, and the position of the sensor device 1 may be measured based on the signal received by the receiving device.

[0021] Figure 4 is a block diagram showing an example of the internal functional configuration of the learning model generation device 2. The learning model generation device 2 executes a learning model generation method. The learning model generation device 2 is a computer such as a server or a personal computer. The learning model generation device 2 comprises a calculation unit 21, a memory 22, a drive unit 23, a storage unit 24, an operation unit 25, a display unit 26, and an interface unit 27. The calculation unit 21 is configured using, for example, a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), or a multi-core CPU. The calculation unit 21 may also be configured using a quantum computer. The memory 22 stores temporary data generated in connection with calculations. The memory 22 is, for example, RAM (Random Access Memory). The drive unit 23 reads information from a recording medium 20 such as an optical disc or portable memory. The storage unit 24 is non-volatile and is, for example, a hard disk or a non-volatile semiconductor memory.

[0022] The arithmetic unit 21 causes the drive unit 23 to read the computer program 241 recorded on the recording medium 20, and stores the read computer program 241 in the storage unit 24. The arithmetic unit 21 executes the necessary processing for the learning model generation device 2 according to the computer program 241. The computer program 241 may be a computer program product. The computer program 241 may be downloaded from outside the learning model generation device 2. Alternatively, the computer program 241 may be pre-stored in the learning model generation device 2. In these cases, the learning model generation device 2 does not need to have a drive unit 23.

[0023] The operation unit 25 accepts input of information such as text by receiving operations from the user. The operation unit 25 is, for example, a touch panel, pen tablet, keyboard, or pointing device. The display unit 26 displays images. The display unit 26 is, for example, a liquid crystal display or an EL display (electroluminescent display). The operation unit 25 and the display unit 26 may be integrated into one unit.

[0024] The interface unit 27 is connected to a memory reader 31 or a receiving device 32, which are not shown in Figure 4. The memory reader 31 inputs data including position information, heart rate, and step count read from the portable memory 16 to the learning model generation device 2 via the interface unit 27. Alternatively, the receiving device 32 inputs data including received position information, heart rate, and step count to the learning model generation device 2 via the interface unit 27. The interface unit 27 may be configured to have the portable memory 16 installed and to read data from the portable memory 16, or it may be configured to receive data transmitted from the sensor device 1. In this way, the learning model generation device 2 acquires data including position information indicating the position of the athlete 10 as measured by the sensor device 1, the athlete 10's heart rate as measured by the sensor device 1, and the athlete 10's step count as measured by the sensor device 1, via the interface unit 27.

[0025] The learning model generation device 2 is composed of multiple computers, and data may be stored in a distributed manner across multiple computers, and processing may be executed in a distributed manner across multiple computers. The learning model generation device 2 may be implemented using cloud computing, or it may be implemented using multiple virtual machines installed within a single computer.

[0026] The memory unit 24 stores player data containing personal information of multiple players 10. The player data includes sensor identification information of the sensor device 1 worn by the player 10, the player 10's birthday, and the player 10's position in their sport.

[0027] The learning model generation system 200 includes multiple sensor devices 1 attached to multiple athletes 10. The learning model generation device 2 acquires data including location information, heart rate, and step count from each of the multiple sensor devices 1. Specifically, the learning model generation device 2 acquires location information indicating the position of the athlete 10 as measured by the multiple sensor devices 1, the heart rate of the athlete 10 as measured by the multiple sensor devices 1, and the step count of the athlete 10 as measured by the multiple sensor devices 1. The learning model generation device 2 stores the measurement data 242, including the location information, heart rate, and step count related to the multiple athletes 10, in the storage unit 24. For example, when sports practice is conducted, the location information, heart rate, and step count related to each athlete 10 are acquired using the sensor devices 1, and the measurement data 242, including the acquired location information, heart rate, and step count, is stored.

[0028] Figure 5 is a conceptual diagram showing an example of the contents of measurement data 242. Measurement data 242 includes location information, heart rate, and step count for each athlete 10. Measurement data 242 also includes athlete identification information to identify athlete 10 and position information indicating athlete 10's position in the sport. Athlete identification information is information unique to athlete 10. Athlete identification information corresponds to personal identification information. Athlete identification information may be the same as sensor identification information. Alternatively, athlete identification information may be recorded in the athlete data in association with sensor identification information. The learning model generation device 2 refers to the athlete data, identifies the athlete identification information corresponding to the sensor identification information contained in the data acquired from the sensor device 1, and records it in measurement data 242.

[0029] Position information is different depending on the position of player 10. For example, if the sport is soccer, the position information will be different depending on whether player 10 is a forward or a goalkeeper. For example, the learning model generator 2 refers to the player data, identifies the position information corresponding to the player identification information, and records it in the measurement data 242. Note that the measurement data 242 does not necessarily contain player identification information or position information.

[0030] Associated with player identification information and position information for player 10, location information indicating player 10's position, player 10's heart rate, and player 10's step count are recorded. The location information is associated with the time when the position was measured. Similarly, the heart rate and step count are associated with the time when the heart rate and step count were measured. For example, the time associated with the heart rate or step count is the time when the predetermined time for measuring the heart rate or step count began or ended.

[0031] The learning model generation device 2 generates training data for generating a learning model based on the measurement data 242, and performs the process of generating a learning model that has been trained using the training data. Figure 6 is a flowchart of an example of the procedure for generating a learning model. Hereinafter, steps will be abbreviated as S. The calculation unit 21 of the learning model generation device 2 executes the process according to the computer program 241.

[0032] The learning model generation device 2 generates motion data representing the actions of each player 10 from the position information and step count of each player 10 included in the measurement data 242 (S11). Figure 7 is a conceptual diagram showing an example of the contents of the motion data. The motion data includes the player's velocity, acceleration, absolute value of acceleration, step count, metabolic power, and MSF (muscular fatigue) as feature quantities that represent the characteristics of the player's actions. By using multiple types of feature quantities as motion data, the characteristics of the player's actions can be represented in detail. Note that the motion data does not necessarily have to include any of the following: velocity, acceleration, absolute value of acceleration, step count, metabolic power, and MSF.

[0033] In S11, the calculation unit 21 calculates velocity, acceleration, and the absolute value of acceleration based on the change in position information and the change in time associated with the position information. The calculation unit 21 also includes the number of steps in the motion data based on the number of steps contained in the measurement data 242. For example, the calculation unit 21 includes the number of steps contained in the measurement data 242 itself, or a calculated value such as the number of steps calculated from the number of steps contained in the measurement data 242 with a changed length of time to be counted, as the number of steps in the motion data. The calculation unit 21 calculates the metabolic power as the value obtained by multiplying acceleration and velocity. Based on the change in distance traveled and direction of travel, the calculation unit 21 calculates the number of times the angle changes of 100 degrees or more occur when moving 4m, which is the MSF.

[0034] In S11, the calculation unit 21 calculates the average or maximum value of each feature included in the motion data within a predetermined time. For example, the predetermined time is 20 seconds. The motion data also includes the average or maximum value of each feature in each of multiple periods. That is, the calculation unit 21 calculates the average or maximum value of velocity, acceleration, absolute value of acceleration, number of steps, metabolic power, and MSF in each of multiple periods. Figure 7 shows an example in which the motion data includes the average or maximum value of each feature in the period from 0 to 20 seconds before, the period from 20 to 40 seconds before, and the period from 40 to 60 seconds before, based on an arbitrary time. By including features from multiple periods in the motion data, the temporal changes in the athlete's 10 movements are represented. The calculation unit 21 generates motion data relating to multiple time points. Note that the learning model generation system 200 may also be configured such that the sensor device 1 does not measure the number of steps, and the learning model generation device 2 calculates the number of steps in S11 based on the distance traveled, change in direction of movement, change in velocity, or change in acceleration obtained from the change in position information.

[0035] The learning model generation device 2 then calculates the heart rate percentage (heart rate data) for each athlete 10 from the heart rate of each athlete 10 included in the measurement data 242 (S12). The heart rate percentage is the value obtained by dividing the heart rate by the maximum heart rate. The maximum heart rate is the highest possible heart rate for a person. The maximum heart rate depends on the person's age. The heart rate percentage is a value obtained by correcting the heart rate according to the person's age and corresponds to the heart rate data. By correcting the heart rate according to the person's age, the influence of age can be eliminated when generating the learning model and determining fatigue. For example, the maximum heart rate per minute for an athlete is expressed by the following equation (1). Maximum heart rate = 210 - 0.5 × age …(1)

[0036] In S12, the calculation unit 21 reads the birthday of player 10 from the player data based on the player identification information, calculates the age, and calculates the maximum heart rate using equation (1). The calculation unit 21 also calculates the heart rate ratio by dividing the heart rate included in the measurement data 242 by the maximum heart rate. For example, if the measurement data 242 includes heart rate data for 20 seconds, the calculation unit 21 converts the maximum heart rate per minute calculated using equation (1) to the maximum heart rate over 20 seconds and calculates the heart rate ratio. Alternatively, the calculation unit 21 may convert the heart rate included in the measurement data 242 to heart rate per minute and calculate the heart rate ratio by dividing the heart rate by the maximum heart rate per minute. Note that the maximum heart rate may be determined by a method other than using equation (1). For example, various formulas other than equation (1) have been conventionally proposed for calculating the maximum heart rate according to age, so the maximum heart rate may be calculated using a formula other than equation (1). For example, a table recording the correspondence between age and maximum heart rate may be pre-stored in the storage unit 24, and the maximum heart rate corresponding to age may be identified based on the table. The calculation unit 21 calculates the heart rate for multiple time points. The processes S11 and S12 may be executed in reverse order.

[0037] The learning model generation device 2 then generates training data including motion data and heart rate (S13). In S13, the calculation unit 21 generates training data by associating motion data and heart rate related to the same time. For example, the calculation unit 21 generates a dataset associating motion data and heart rate when the reference time for calculating motion data and the time associated with the heart rate that formed the basis of the heart rate coincide within a predetermined error. The calculation unit 21 generates multiple datasets and generates training data including multiple datasets.

[0038] The training data further includes a pre-training heart rate, which is the heart rate at a point in time prior to the point in time when the heart rate that formed the basis of the heart rate data was obtained. The pre-training heart rate corresponds to the pre-training heart rate data. The pre-training heart rate is the heart rate at a point in time that is a predetermined retrospective period. For example, the pre-training heart rate is the heart rate obtained from the heart rate 60 to 80 seconds prior. The calculation unit 21 selects the pre-training heart rate from the multiple calculated heart rates and associates it with the operation data and heart rate.

[0039] The training data further includes player identification information and position information. The calculation unit 21 associates the player identification information and position information with motion data and heart rate. The training data includes multiple datasets, each dataset containing associated motion data, heart rate, pre-race heart rate, player identification information, and position information.

[0040] By executing processes S11 to S13, the learning model generation device 2 acquires training data 243. The calculation unit 21 stores the generated training data 243 in the storage unit 24. The learning model generation device 2 executes processes S11 to S13 for multiple players 10. One or more datasets are generated for each player 10, and the training data 243 includes multiple datasets relating to multiple players 10.

[0041] The processes S11 to S13 may be executed whenever position information, heart rate, and step count are obtained from the sensor device 1, or they may be executed when a certain amount of information has been recorded in the measurement data 242. Alternatively, the processes S11 to S13 may be performed outside the learning model generation device 2, and the generated training data 243 may be input to the learning model generation device 2 from an external source.

[0042] The learning model generation device 2 then uses the training data 243 to generate a learning model used to predict heart rate (S14). The learning model is realized by the arithmetic unit 21 performing information processing according to the computer program 241. The storage unit 24 stores the data necessary to realize the learning model. The learning model may be constructed using hardware. The learning model may also be realized using a quantum computer.

[0043] Figure 8 is a conceptual diagram illustrating the functionality of the learning model. The learning model receives motion data, pre-game heart rate, athlete identification information, and position information as input. The learning model is trained to output heart rate when motion data, pre-game heart rate, athlete identification information, and position information are input. The learning model is constructed using LightGBM (Light Gradient Boosting Machine). The learning model may also be constructed using a neural network, transformer, or LSTM (Long Short-Term Memory).

[0044] In S14, the calculation unit 21 inputs motion data, pre-training heart rate, player identification information, and position information contained in the training data 243 into the learning model and trains the learning model. The learning model outputs a heart rate in response to the input motion data, pre-training heart rate, player identification information, and position information. The calculation unit 21 acquires the heart rate output by the learning model and adjusts the calculation parameters of the learning model so that the error between the heart rate associated with the motion data, pre-training heart rate, player identification information, and position information input into the learning model in the training data 243 and the heart rate output by the learning model is reduced. That is, the parameters are adjusted so that a heart rate that is almost identical to the heart rate associated with the motion data, heart rate, pre-training heart rate, player identification information, and position information is output. The calculation unit 21 performs machine learning of the learning model by repeatedly processing using multiple datasets contained in the training data 243 and adjusting the parameters of the learning model.

[0045] For example, the calculation unit 21 adjusts the parameters of a learning model using Optuna®. For example, the calculation unit 21 divides the training data 243 into multiple parts, generates multiple learning models by cross-validation, and generates a single learning model by averaging and fusing the multiple learning models. The calculation unit 21 may also adjust the parameters of the learning model using other algorithms. The calculation unit 21 stores the trained data with the final adjusted parameters in the storage unit 24. In this way, a trained learning model is generated. After S14 is completed, the learning model generation device 2 terminates the process for generating the learning model.

[0046] Generally, the higher the intensity of a person's exercise, the higher their heart rate. Therefore, there is a correlation between athlete 10's movements and heart rate, and it is possible to generate a learning model that generates athlete 10's heart rate from movement data representing athlete 10's movements. Also, a person's heart rate increases when they are highly fatigued. The heart rate obtained from movement data using the learning model represents athlete 10's heart rate in their normal state. If the actual heart rate is higher than the heart rate obtained from movement data using the learning model, it can be determined that athlete 10 is more fatigued than usual. In this way, the heart rate obtained from movement data using the learning model can be used as evidence to determine athlete 10's fatigue.

[0047] Heart rate is influenced by the previous heart rate. For example, even if the exercise intensity is the same, a higher previous heart rate will result in a higher heart rate. By inputting the pre-exercise heart rate in addition to the movement data into the learning model, the learning model can output a more accurate heart rate. There may be differences in the relationship between athlete 10's movements and heart rate depending on the individual athlete 10. By inputting athlete identification information in addition to the movement data into the learning model, the learning model can output a more accurate heart rate according to the individuality of athlete 10. If athlete 10's position is different, the content of the exercise performed by athlete 10 will differ, and there may be differences in the relationship between movement data and heart rate. By inputting position information in addition to movement data into the learning model, the learning model can output a more accurate heart rate according to athlete 10's position. Note that the learning model may not have any of the pre-exercise heart rate, athlete identification information, or position information input.

[0048] Next, a method for determining athlete 10 fatigue using a learning model will be explained. Figure 9 is a schematic diagram showing an example configuration of a fatigue determination system 400 for determining athlete 10 fatigue. The fatigue determination system 400 performs a fatigue determination method. The fatigue determination system 400 includes a sensor device 1 attached to athlete 10 performing sports, and a fatigue determination device 4 that performs information processing for determining fatigue. Sensor device 1 is the same as the sensor device 1 used in the learning model generation system 200. Sensor device 1 measures athlete 10's position, heart rate, and step count. Fatigue determination device 4 acquires position information indicating athlete 10's position, heart rate, and step count from sensor device 1.

[0049] Figure 10 is a block diagram showing an example of the internal functional configuration of the fatigue determination device 4. The fatigue determination device 4 is a computer such as a personal computer, tablet computer, or smartphone. The fatigue determination device 4 comprises a calculation unit 41, a memory 42, a drive unit 43, a storage unit 44, an operation unit 45, a display unit 46, and an interface unit 47. The calculation unit 41 is configured using, for example, a CPU, GPU, or multi-core CPU. The calculation unit 41 may also be configured using a quantum computer. The memory 42 stores temporary data generated in connection with calculations. The memory 42 is, for example, RAM. The drive unit 43 reads information from a recording medium 40 such as an optical disc or portable memory. The storage unit 44 is non-volatile and is, for example, a hard disk or non-volatile semiconductor memory.

[0050] The calculation unit 41 causes the drive unit 43 to read the computer program 441 recorded on the recording medium 40, and stores the read computer program 441 in the storage unit 44. The calculation unit 41 executes the necessary processing for the fatigue determination device 4 according to the computer program 441. The computer program 441 may be a computer program product. The computer program 441 may be downloaded from outside the fatigue determination device 4. Alternatively, the computer program 441 may be pre-stored in the fatigue determination device 4. In these cases, the fatigue determination device 4 does not need to have a drive unit 43.

[0051] The operation unit 45 accepts input of information such as text by receiving operations from the user. The operation unit 45 is, for example, a touch panel, pen tablet, keyboard, or pointing device. The display unit 46 displays images. The display unit 46 is, for example, a liquid crystal display or an EL display. The operation unit 45 and the display unit 46 may be integrated into one unit.

[0052] The interface unit 47 is connected to a memory reader 31 or a receiving device 32, which is not shown in Figure 10. The memory reader 31 or receiving device 32 inputs data including location information, heart rate, and step count to the fatigue determination device 4 through the interface unit 47. The interface unit 47 may be equipped with a portable memory 16 and read data from the portable memory 16, or it may be configured to receive data transmitted from the sensor device 1. In this way, the fatigue determination device 4 acquires data including location information indicating the position of the athlete 10 as measured by the sensor device 1, the athlete 10's heart rate as measured by the sensor device 1, and the athlete 10's step count as measured by the sensor device 1, through the interface unit 47. The fatigue determination device 4 may be composed of multiple computers. The fatigue determination device 4 may be implemented using cloud computing, or it may be implemented by multiple virtual machines installed in a single computer.

[0053] The fatigue determination device 4 includes a learning model 442. The learning model 442 is realized by the arithmetic unit 41 performing information processing according to a computer program 441. The learning model 442 is a learning model learned by the learning model generation device 2. The fatigue determination device 4 includes the learning model 442 by storing learned data, which records the parameters of the learning model learned by the learning model generation device 2, in the storage unit 44. For example, the learned data is read from the recording medium 40 by the drive unit 43 or downloaded. The learning model 442 may be configured by hardware. The learning model 442 may be realized using a quantum computer. Alternatively, the learning model 442 may be located outside the fatigue determination device 4, and the fatigue determination device 4 may perform processing using the external learning model 442. For example, the learning model 442 may be configured in the cloud.

[0054] The memory unit 44 stores athlete data. The fatigue determination device 4 acquires data including location information, heart rate, and step count from the sensor device 1. Specifically, the fatigue determination device 4 acquires location information indicating the position of athlete 10 as measured by the sensor device 1, athlete 10's heart rate, and athlete 10's step count. The fatigue determination device 4 stores the measurement data, including location information, heart rate, and step count, in the memory unit 24. For example, when a sports match is held, location information, heart rate, and step count related to athlete 10 are acquired using the sensor device 1, and the measurement data, including the acquired location information, heart rate, and step count, is stored.

[0055] The fatigue determination device 4 generates input data for input to the learning model and uses the learning model 442 to perform a process to determine the fatigue of player 10. Figure 11 is a flowchart showing an example of the procedure for determining the fatigue of player 10. The calculation unit 41 of the fatigue determination device 4 executes the process according to the computer program 441.

[0056] The fatigue determination device 4 generates motion data representing the movements of athlete 10 from the position information and step count of athlete 10 included in the measurement data (S201). In S201, the calculation unit 41 generates motion data by calculating velocity, acceleration, absolute value of acceleration, metabolic power, and MSF based on the change in position information and the change in time associated with the position information. The calculation unit 41 also includes the step count based on the step count included in the measurement data in the motion data. As shown in Figure 7, the motion data includes the average or maximum value of velocity, acceleration, absolute value of acceleration, step count, metabolic power, and MSF for each of multiple periods. Through the processing in S201, the fatigue determination device 4 acquires motion data. Note that the fatigue determination system 400 may also be configured such that the sensor device 1 does not measure the step count, and the fatigue determination device 4 calculates the step count based on the distance traveled, change in direction of movement, change in velocity, or change in acceleration obtained from the change in position information in S201. The motion data does not necessarily have to include any of the following: velocity, acceleration, absolute value of acceleration, number of steps, metabolic power, and MSF.

[0057] The fatigue determination device 4 then calculates the first heart rate by calculating the heart rate percentage of athlete 10 from the heart rate of athlete 10 included in the measurement data (S202). The first heart rate is the actual heart rate percentage calculated from the heart rate of athlete 10 and corresponds to the first heart rate data. In S202, the calculation unit 41 calculates the age of athlete 10 based on the athlete data, calculates the maximum heart rate, and calculates the first heart rate. The calculation unit 41 calculates the first heart rate for multiple time points. The processes of S201 and S202 may be executed in the reverse order.

[0058] The fatigue determination device 4 then generates input data for input to the learning model 442 (S203). Figure 12 is a conceptual diagram showing an example of the contents of the input data. The input data includes motion data, pre-operation heart rate, athlete identification information, and position information. In S203, the calculation unit 41 identifies the first heart rate at a predetermined time point in the past from the reference time used to calculate the motion data as the pre-operation heart rate. The calculation unit 41 generates the input data by associating the motion data, pre-operation heart rate, athlete identification information, and position information, and stores it in the storage unit 44. The calculation unit 41 also associates the input data and the first heart rate related to the same time. For example, the calculation unit 41 associates the motion data and the first heart rate when the reference time used to calculate the motion data and the time associated with the heart rate on which the first heart rate was derived coincide within a predetermined error.

[0059] The processing in S201 to S203 may be performed each time position information, heart rate, and step count are obtained from the sensor device 1, or it may be performed when a certain amount of information has been recorded in the measurement data. Alternatively, the processing in S201 to S203 may be performed outside the fatigue determination device 4, and the generated input data and the first heart rate may be input to the fatigue determination device 4 from an external source.

[0060] The fatigue determination device 4 inputs the input data to the learning model 442 (S204). In S204, the calculation unit 41 inputs the input data to the learning model 442 and causes the learning model 442 to execute processing. The learning model 442 outputs a second heart rate in response to the input data. The second heart rate is a value estimated by using the learning model 442 to determine the heart rate of the athlete 10, and corresponds to the second heart rate data. The fatigue determination device 4 acquires the second heart rate output by the learning model 442 (S205).

[0061] The fatigue determination device 4 then calculates the difference obtained by subtracting the second heart rate from the first heart rate (S206). In S206, the calculation unit 41 subtracts the second heart rate, which was obtained by inputting the input data into the learning model 442, from the first heart rate associated with the input data. The calculated difference is the difference between the actual heart rate of the athlete 10 and the value estimated for the athlete 10's heart rate using the learning model 442. If the first heart rate exceeds the second heart rate, the difference becomes a positive value, indicating that the athlete 10's heart rate is higher than usual. Since the heart rate is higher than usual, it is possible that the athlete 10's body is under greater strain than usual, and the athlete 10 is fatigued. The larger the difference, the greater the strain on the athlete 10's body, and the greater the fatigue level, which represents the intensity of fatigue.

[0062] The fatigue determination device 4 then calculates the cumulative sum of the differences (S207). For example, the fatigue determination device 4 repeats the processes S201 to S206 multiple times, and performs the S207 process each time a difference is obtained. For example, the fatigue determination device 4 calculates multiple time-series differences by performing the S201 to S206 processes based on measurement data relating to a certain length of time, and then performs the S207 process. In S207, the calculation unit 41 calculates the cumulative sum of the differences by sequentially adding up the multiple differences. When calculating the cumulative sum, the calculation unit 41 simply adds up the differences even if they are negative values.

[0063] The sum of the differences is a value calculated based on information continuously obtained from player 10, and therefore reflects player 10's condition. A positive sum of the differences indicates that player 10's heart rate remains higher than normal, and that player 10's body is under greater strain than usual. Therefore, a positive sum of the differences clearly indicates that player 10 is fatigued. The larger the sum of the differences, the greater the fatigue level of player 10.

[0064] The fatigue determination device 4 then outputs the difference and the sum of the differences (S208). In S208, the calculation unit 41 generates an image including the difference and the sum of the differences and displays it on the display unit 46. The larger the difference and the sum of the differences, the greater the fatigue level of player 10, so the difference and the sum of the differences can be used as information representing the fatigue level. The difference and the sum of the differences output in S208 represent the fatigue level of player 10. In an image including the difference and the sum of the differences, it is not necessary to explicitly output that it is fatigue level or information related to fatigue, as long as the difference and the sum of the differences are output. This is because it is sufficient for the user to recognize that it is fatigue level or information related to fatigue from the difference and the sum of the differences. The calculation unit 41 also stores the history of the difference and the history of the sum of the differences in the storage unit 44. The fatigue determination device 4 may output only the difference or only the sum of the differences.

[0065] The fatigue determination device 4 then determines whether the calculated difference or sum of differences is greater than or equal to a predetermined threshold (S209). In S209, the calculation unit 41 compares the difference or sum of differences with a predetermined threshold and makes a determination. The threshold is either stored in the storage unit 44 in advance or included in the computer program 441. The threshold for the difference and the threshold for the sum of differences are different.

[0066] If the difference or sum of differences is less than the threshold (S209: NO), the fatigue determination device 4 terminates the process of determining the fatigue of player 10. If the difference or sum of differences is greater than or equal to the threshold (S209: YES), the fatigue determination device 4 outputs a warning indicating that player 10 is fatigued (S210). In S210, the calculation unit 41 generates an image that includes the difference, the sum of differences, and the warning, and displays it on the display unit 46. Note that in S209 and S210, the fatigue determination device 4 may terminate processing if the difference or sum of differences is less than the threshold, and output a warning if the difference or sum of differences exceeds the threshold. The output warning is a notification corresponding to the difference or sum of differences. After executing S210, the fatigue determination device 4 terminates the process of determining the fatigue of player 10. The fatigue determination device 4 repeats the processes of S201 to S210, updating the image displayed on the display unit 46.

[0067] Figure 13 is a schematic diagram showing an example of an image displayed by the fatigue determination device 4. The image includes the name and age of player 10. The calculation unit 41 generates an image including the name and age of player 10 based on the personal information recorded in the player data. The image includes the date and time, the current time, and the time when fatigue determination was started. The image also displays player 10's maximum heart rate, the first heart rate which is the actual heart rate, the second heart rate estimated by the learning model 442, the difference obtained by subtracting the second heart rate from the first heart rate, and the sum of the differences. In this way, information regarding player 10's fatigue is output. The user can find out player 10's current state by checking player 10's first heart rate. The user can also find out player 10's fatigue level by checking the difference and the sum of the differences which represent the fatigue level.

[0068] Furthermore, the image includes a graph showing the change in fatigue level over time. The calculation unit 41 uses the sum of differences as information representing the fatigue level, generates a graph showing the change in the sum of differences over time as the change in fatigue level over time, and displays the image including the generated graph on the display unit 46. By checking the change in fatigue level over time, the user can find out how fatigued the athlete 10 is and how the fatigue is accumulating. For example, by adjusting the athlete 10's exercise according to the change in fatigue level over time, the fatigue of the athlete 10 can be suppressed.

[0069] Figure 13 shows an example where a warning is issued indicating that player 10 is fatigued. In the example shown in Figure 13, a text warning is displayed indicating that the fatigue level exceeds an acceptable value. By checking the warning, the user can understand that player 10's fatigue level is dangerously high. For example, stopping player 10's exercise when the warning is issued can help prevent injuries.

[0070] The processing in S201 to S210 can be performed for multiple players 10. For example, the fatigue determination device 4 executes S201 to S210 sequentially or in parallel for multiple players 10 and outputs the difference and the sum of the differences for multiple players 10. The fatigue determination device 4 also stores the difference history and the sum of the differences history for each of the multiple players 10 in the storage unit 44. In S208, the fatigue determination device 4 may output the difference and the sum of the differences for multiple players 10. The calculation unit 41 generates an image including the difference and the sum of the differences for multiple players 10 and displays it on the display unit 46.

[0071] Figure 14 is a schematic diagram showing an example of an image displayed by the fatigue determination device 4 to output differences and sums of differences for multiple athletes 10. The image contains information about each of the multiple athletes 10. The image includes the name of each athlete 10, the first heart rate, the second heart rate, the difference obtained by subtracting the second heart rate from the first heart rate, and the sums of the differences. Furthermore, for each athlete 10, a graph showing the change in fatigue level over time is included. For example, the change in the sum of differences over time is used as the change in fatigue level over time. In this way, information about the fatigue of multiple athletes 10 is output.

[0072] The user can check information about the fatigue of multiple players 10 and know the current status of multiple players 10. The user can also know the fatigue level of multiple players 10. Figure 14 shows an example where a warning is output with the text "Caution!" indicating that one player 10 is fatigued. By checking the warning, the user can know that there is a player 10 whose fatigue level is dangerously high. When the user operates the control unit 45 and inputs an instruction to select one player 10 from multiple players 10, the fatigue determination device 4 may display an image containing information about the one player 10, as shown in Figure 13.

[0073] The fatigue determination system 400 may include a plurality of fatigue determination devices 4. Each of the plurality of fatigue determination devices 4 may individually perform S201 to S210 for any of the players 10.

[0074] As detailed above, in this embodiment, the position, heart rate, and step count of athlete 10 are measured using a sensor device 1 attached to athlete 10, motion data and heart rate related to athlete 10 are acquired, and a learning model is generated by learning using training data including motion data and heart rate. The learning model outputs the heart rate when motion data is input. By using the learning model, the heart rate can be estimated from the motion data. Since a person's heart rate changes according to fatigue, the heart rate estimated using the learning model can serve as a basis for determining athlete 10's fatigue. If the first heart rate, which is the actual heart rate, exceeds the second heart rate, which is the heart rate estimated using the learning model, it can be determined that athlete 10 is under greater strain than usual and is fatigued. By calculating the difference obtained by subtracting the second heart rate from the first heart rate, or the sum of the cumulative differences, the degree of fatigue of athlete 10 can be visualized. By taking care of athlete 10 according to the visualized degree of fatigue, such as stopping a fatigued athlete 10 from exercising, it is possible to suppress the occurrence of injuries.

[0075] In this embodiment, the fatigue detection device 4 is shown to output a warning according to the degree of fatigue. However, the fatigue detection device 4 may output information other than warnings according to the degree of fatigue. For example, the fatigue detection device 4 may output a warning not according to the degree of fatigue, but according to the difference obtained by subtracting the second heart rate from the first heart rate, the progression of the cumulative sum of these differences, or the progression of the degree of fatigue. It may also notify the timing of rest or player substitution according to these factors. In other words, the fatigue detection device 4 only needs to output notifications according to the difference or the cumulative sum of differences, including notifications according to the progression of fatigue or the degree of fatigue.

[0076] <Embodiment 2> Generally, the higher the intensity of a person's exercise, the higher their heart rate. Furthermore, the second heart rate is obtained as a value corresponding to the movement data representing athlete 10's actions, using the learning model 442. Therefore, the second heart rate can be used as an indicator of the intensity of the exercise performed by athlete 10. The second heart rate is expressed as a numerical value in the range of 0 to 1.0 (0% to 100%). Since the learning model 442 is trained using movement data and training data including heart rate obtained over a predetermined period, such as the past month, the exercise intensity indicated by the second heart rate represents a percentage of the exercise intensity performed by athlete 10 in the past. Exercise intensity includes both quantitative and qualitative aspects. Even low-intensity exercise can become high in intensity if performed for a long period, and high-intensity exercise can also become high in intensity even if performed for a short period. Embodiment 2 presents a model that clarifies the relationship between exercise intensity using the second heart rate as an indicator, fatigue level, and athlete 10's physical condition.

[0077] The configuration of the fatigue determination system 400 is the same as in Embodiment 1. The fatigue determination device 4 performs processing to manage the physical condition of the player 10. Figure 15 is a flowchart showing an example of the procedure for determining the physical condition of the player 10. The fatigue determination device 4 performs processing S301 to S307, which is the same as S201 to S207.

[0078] The fatigue determination device 4 then identifies the physical condition of the athlete 10 (S308). Similar to Embodiment 1, the fatigue determination device 4 uses the difference obtained by subtracting the second heart rate from the first heart rate, or the sum of these differences, as information representing the degree of fatigue. For example, the value of the sum of the differences is taken as the fatigue level value. The greater the degree of fatigue, the greater the fatigue level value. A positive fatigue level value indicates a high degree of fatigue, and a negative fatigue level value indicates a low degree of fatigue. The fatigue determination device 4 also uses the second heart rate as an indicator representing the intensity of the exercise performed by the athlete 10. Specifically, the value of the intensity of the exercise performed by the athlete 10 is the value of the second heart rate, which can take a value in the range of 0% to 100%.

[0079] The fatigue assessment device 4 determines the physical condition of athlete 10 by comparing the intensity of the exercise performed by athlete 10 with athlete 10's fatigue level. Generally, the higher the exercise intensity, the greater the fatigue level tends to be. If the fatigue level is low despite high exercise intensity, athlete 10 is considered to be in good physical condition. If the fatigue level is high despite low exercise intensity, athlete 10 is considered to be in poor physical condition. If the fatigue level is high despite high exercise intensity, or if the fatigue level is low despite low exercise intensity, athlete 10 is considered to be in normal physical condition. In this way, the physical condition of athlete 10 can be determined by comparing exercise intensity and fatigue level.

[0080] In S308, the calculation unit 41 identifies the physical condition of athlete 10 based on the exercise intensity and fatigue level. The calculation unit 41 expresses the physical condition of athlete 10 as a numerical value. The calculation unit 41 calculates the physical condition such that the higher the exercise intensity and the lower the fatigue level, the larger the numerical value of the physical condition (good physical condition), and the lower the exercise intensity and the higher the fatigue level, the smaller the numerical value of the physical condition (poor physical condition). For example, the calculation unit 41 calculates the numerical value of the physical condition by performing calculations according to a predetermined algorithm included in the computer program 441 based on the exercise intensity and fatigue level. For example, a table that records the relationship between exercise intensity, fatigue level, and physical condition is stored in the storage unit 44 in advance, and the calculation unit 41 identifies the physical condition of athlete 10 by extracting the physical condition value associated with exercise intensity and fatigue level from the table.

[0081] The fatigue determination device 4 performs the processes S301 to S308 for each of the multiple athletes 10. Through the processes S301 to S308, the fatigue determination device 4 stores in the memory unit 44 the difference obtained by subtracting the second heart rate from the first heart rate, the cumulative sum of the differences, and the physical condition for each of the multiple athletes 10. The fatigue determination device 4 also repeatedly performs the processes S301 to S308. That is, the fatigue determination device 4 generates operational data at multiple points in time and obtains the difference obtained by subtracting the second heart rate from the first heart rate, the cumulative sum of the differences, and the physical condition at each point in time. For example, the fatigue determination device 4 performs the processes S301 to S308 periodically, such as once a day. As a result, the fatigue determination device 4 stores in the memory unit 44 the history of the difference obtained by subtracting the second heart rate from the first heart rate, the history of the cumulative sum of the differences, and the history of the physical condition for each of the multiple athletes 10.

[0082] The fatigue determination device 4 then outputs the intensity of the exercise performed by the multiple athletes 10 and the fatigue level of the multiple athletes 10 (S309). In S309, the calculation unit 41 outputs the exercise intensity and fatigue level by displaying an image containing the exercise intensity and fatigue level of the multiple athletes 10 on the display unit 46. Figure 16 is a schematic diagram showing an example of an image containing the exercise intensity and fatigue level of the multiple athletes 10. Based on the difference obtained by subtracting the second heart rate from the first heart rate, or the sum of the differences, stored in the storage unit 44, the calculation unit 41 calculates the total fatigue level of the multiple athletes 10 at a specific point in time, and the average intensity of the exercise performed by the multiple athletes 10.

[0083] The image displayed on the display unit 46 includes a calendar indicating a specific point in time (date). For example, by operating the control unit 45, the user can specify a specific point in time using the calendar, and the calculation unit 41 calculates the total fatigue level and the average exercise intensity at the specified point in time. The specific point in time can also be changed by the user operating the control unit 45. The calculation unit 41 displays an image on the display unit 46 that includes the calculated total fatigue level and average exercise intensity. By viewing the image, the user can check the total fatigue level of multiple athletes 10 and the exercise intensity performed by all multiple athletes 10. For example, the user can manage the fatigue level and exercise intensity for an entire sports team to which multiple athletes 10 belong.

[0084] Furthermore, in S309, the calculation unit 41 displays an image on the display unit 46 that includes a graph showing the time-dependent changes in the total fatigue level and the average exercise intensity for multiple athletes 10, based on the history of the difference obtained by subtracting the second heart rate from the first heart rate and the history of the cumulative sum of these differences. In the graph included in Figure 16, the horizontal axis represents time, and the vertical axis represents the total fatigue level and the average exercise intensity. The solid line graph shows the average exercise intensity, and the dashed line graph shows the total fatigue level. The user can check the time-dependent changes in the overall fatigue level of multiple athletes 10, and the time-dependent changes in the exercise intensity performed by all multiple athletes 10. For example, the user can check how the overall fatigue level and exercise intensity of the team changed over time.

[0085] Furthermore, in S309, the calculation unit 41 displays images on the display unit 46 that include the names and fatigue levels of players 10 with high fatigue levels and players 10 with low fatigue levels at a specific point in time. The user can then identify players 10 with particularly high fatigue levels and players 10 with particularly low fatigue levels. For example, the user can manage the players 10 according to their fatigue levels, such as having players 10 with particularly high fatigue levels rest or encouraging players 10 with particularly low fatigue levels to exercise.

[0086] The fatigue determination device 4 outputs a list of exercise intensity, fatigue level, and physical condition for multiple athletes 10 at a specific point in time (S310). In S310, the calculation unit 41 outputs the list of exercise intensity, fatigue level, and physical condition for multiple athletes 10 at a specific point in time by displaying an image containing the list on the display unit 46. The user can specify a specific point in time using a calendar by operating the operation unit 45.

[0087] Figure 17 is a schematic diagram showing an example of an image containing a list of exercise intensity, fatigue level, and physical condition for multiple athletes 10. In the example shown in Figure 17, physical condition is indicated by an arrow. An upward arrow indicates good physical condition, and a downward arrow indicates poor physical condition. The calculation unit 41 displays the image containing the arrow corresponding to the physical condition value on the display unit 46. By displaying the list, the user can easily check the status of multiple athletes 10. For example, the user can manage multiple athletes 10 according to each athlete's exercise intensity, fatigue level, or physical condition. For example, among multiple athletes 10 belonging to a sports team, the overall condition of the team can be managed by having athletes 10 who are unwell rest and athletes 10 who are in good condition participate in the game.

[0088] In addition, in S310, the calculation unit 41 displays an image on the display unit 46 that includes the average intensity of the exercises performed by multiple athletes 10 at a specific point in time, as well as the total and average fatigue levels of the multiple athletes 10. The user can then check the overall fatigue level of the multiple athletes 10 and the overall intensity of the exercises performed by all of the multiple athletes 10.

[0089] The fatigue determination device 4 outputs the exercise intensity, fatigue level, and physical condition of a specific athlete 10 (S311). In S311, the calculation unit 41 outputs the exercise intensity, fatigue level, and physical condition of a specific athlete 10 by displaying an image containing the exercise intensity, fatigue level, and physical condition of the specific athlete 10 on the display unit 46. For example, when a user operates the operation unit 45, a specific athlete 10 is specified using an image as shown in Figure 17, and the calculation unit 41 displays an image containing the exercise intensity, fatigue level, and physical condition of the specified athlete 10 on the display unit 46.

[0090] Figure 18 is a schematic diagram showing an example of an image containing exercise intensity, fatigue level, and physical condition for a specific athlete 10. The image includes the name of the specific athlete 10. In S311, the calculation unit 41 displays an image containing the fatigue level at a specific point in time on the display unit 46. The user can specify a specific point in time using a calendar by operating the operation unit 45. Also in S311, the calculation unit 41 calculates the average fatigue level over a predetermined period of time prior to the specific point in time, and displays an image containing the calculated average fatigue level as the most recent average fatigue level on the display unit 46. The predetermined period of time is, for example, one week. The user can specify a specific athlete 10 and check the fatigue level of that athlete 10.

[0091] In S311, the calculation unit 41 displays an image on the display unit 46 that includes the exercise intensity, fatigue level, and physical condition at multiple points in time. The exercise intensity, fatigue level, and physical condition included in the image are associated with the date indicating the respective point in time and information indicating whether the exercise at that time was practice or a match. The calculation unit 41 may also display an image that includes a graph showing the changes in exercise intensity, fatigue level, and physical condition over time. By viewing the image, the user can check the exercise intensity, fatigue level, and physical condition at each point in time. The user can also check the changes in exercise intensity, fatigue level, and physical condition over time for a specific athlete 10.

[0092] The calculation unit 41 may receive an instruction from the user to change a specific player 10 by operating the operation unit 45, change the specific player 10 according to the instruction, and output the exercise intensity, fatigue level, and physical condition of the changed specific player 10. The processes of S309, S310, and S311 may be executed in any other order, or they may be executed alternately and repeatedly. Alternatively, any of the processes of S309, S310, and S311 may be omitted. After the processes of S309, S310, and S311 are completed, the fatigue determination device 4 terminates the process for determining the physical condition of the player 10.

[0093] As detailed above, in Embodiment 2, the fatigue determination device 4 outputs the second heart rate as the exercise intensity. By outputting the exercise intensity, the user can check the intensity of the exercise performed by the athlete 10. The fatigue determination device 4 also identifies the athlete 10's physical condition based on the fatigue level and exercise intensity, and outputs the athlete 10's physical condition. By outputting the athlete 10's physical condition, the user can check the athlete 10's physical condition and manage the athlete 10 according to their physical condition.

[0094] In Embodiments 1 and 2 described above, a form was shown in which the difference or the sum of the differences itself is used as information representing the degree of fatigue. However, the fatigue determination device 4 may also output other information corresponding to the difference or the sum of the differences as information representing the degree of fatigue. The information representing the degree of fatigue indicates that the greater the difference or the sum of the differences, the greater the degree of fatigue. In Embodiments 1 and 2, a form was shown in which the difference obtained by subtracting the second heart rate from the first heart rate is used. However, the value obtained by subtracting the first heart rate from the second heart rate may be used as the difference, and if the difference is a negative value, it may be shown that the greater the absolute value of the difference, the greater the degree of fatigue.

[0095] In Embodiments 1 and 2, a configuration was shown in which the heart rate is used as a value corrected according to the person's age, but information other than the heart rate may be used as a value corrected according to the person's age. For example, a value obtained by converting the heart rate to the heart rate for a specific age may be used. In Embodiments 1 and 2, a configuration was shown in which a single learning model is used, but the learning model generation system 200 may be configured to generate learning models for each player or position, and the fatigue determination device 4 may be configured to acquire a second heart rate using the learning model for each player or position.

[0096] In Embodiments 1 and 2, the sensor device 1 is shown to measure positional information, heart rate, and step count related to the athlete 10. However, the sensor device 1 may also measure other information. For example, the sensor device 1 may have an acceleration sensor and measure the acceleration generated in the athlete 10. In Embodiments 1 and 2, the sensor device 1 is shown to be attached to an athlete 10 performing sports. However, the learning model generation system 200 may also generate a learning model by attaching the sensor device 1 to a general person other than the athlete 10. Furthermore, the fatigue determination system 400 may acquire motion data and a second heart rate from a general person and output information regarding fatigue level, exercise intensity, or physical condition. For example, a learning model may be generated and information regarding fatigue level, exercise intensity, or physical condition may be output for a person undergoing rehabilitation, a person playing a physical game, a person training, or a person performing exercises related to daily life.

[0097] The present invention is not limited to the embodiments described above, and various modifications are possible within the scope of the claims. That is, embodiments obtained by combining technical means that have been appropriately modified within the scope of the claims are also included in the technical scope of the present invention. [Explanation of symbols]

[0098] 10 players 1. Sensor device 12 Position Sensors 13 Heart rate sensor 200 Learning Model Generation Systems 2. Learning Model Generation Device 243 training data 400 Fatigue Assessment System 4. Fatigue detection device 40 Recording media 441 Computer Programs 442 Learning Models

Claims

1. Motion data representing the movements of a person exercising, and first heart rate data based on the person's heart rate are acquired. The acquired motion data is input to a learning model that outputs heart rate data when motion data is input, and the second heart rate data output by the learning model is acquired. Output the difference between the first heart rate data and the second heart rate data. A computer program characterized by causing a computer to perform a process.

2. The difference is the value obtained by subtracting the second heart rate data from the first heart rate data. The information regarding the fatigue of the person is output in such a way that the greater the difference, the greater the intensity of fatigue. The computer program according to claim 1, characterized in that it causes a computer to perform a process.

3. Calculate the cumulative sum of the above differences, The information regarding the fatigue of the person is output in such a way that a larger sum of values ​​indicates a greater intensity of fatigue. The computer program according to claim 2, characterized in that it causes a computer to perform the processing.

4. Output a notification corresponding to the difference or the sum of the differences. A computer program according to claim 2 or 3, characterized in that it causes a computer to perform a process.

5. The first heart rate data and the second heart rate data for multiple people are acquired, Output information regarding the fatigue levels of the aforementioned multiple individuals. The computer program according to claim 1, characterized in that it causes a computer to perform a process.

6. The first heart rate data and the second heart rate data are values ​​obtained by correcting a person's heart rate according to the person's age. The computer program according to feature 1.

7. The aforementioned motion data includes speed, acceleration, step count, metabolic power, or MSF (muscular fatigue). The computer program according to feature 1.

8. The aforementioned motion data includes the average or maximum value of data representing a person's movements over multiple periods. The computer program according to feature 1.

9. Prior heart rate data is obtained based on the heart rate at a predetermined time before the time when the heart rate that forms the basis of the first heart rate data is obtained. The learning model, which outputs heart rate data when additional motion data is input, receives the acquired pre-heart rate data as further input, and the second heart rate data output by the learning model is acquired. The computer program according to claim 1, characterized in that it causes a computer to perform a process.

10. By acquiring personal identification information to identify a person, or position information indicating a person's position in a given sport, The learning model, which outputs heart rate data when personal identification information or position information is further input, is then given the acquired personal identification information or position information, and the second heart rate data output by the learning model is retrieved. The computer program according to claim 1, characterized in that it causes a computer to perform a process.

11. The second heart rate data is output as the intensity of the exercise performed by the person. The computer program according to claim 1, characterized in that it causes a computer to perform a process.

12. Output the intensity of the exercise and the degree of fatigue of the person corresponding to the difference or the sum of the differences. The computer program according to claim 11, characterized in that it causes a computer to perform a process.

13. Based on the intensity of the exercise and the degree of fatigue, the physical condition of the person is determined. The computer program according to claim 12, characterized in that it causes a computer to perform a process.

14. Output a list of the exercise intensity, fatigue level, and physical condition for multiple individuals. The computer program according to claim 13, characterized in that it causes a computer to perform a process.

15. Using sensors attached to a person, the position of the person exercising and the person's heart rate are measured. Based on the position and heart rate measured using the aforementioned sensor, motion data representing the movements of the person performing the exercise, and first heart rate data based on the heart rate are acquired. The acquired motion data is input to a learning model that outputs heart rate data when motion data is input, and the second heart rate data output by the learning model is acquired. Based on the first heart rate data and the second heart rate data, information regarding the person's fatigue is output. A fatigue determination method characterized by the following.

Images