Practice support device, control method for practice support device, and control program for practice support device

Abstract

A practice support device according to the present invention comprises: an acquisition unit that acquires practice content and environment information pertaining to a practice environment or user state information pertaining to a state of a user during past running practice by the user together with a corresponding time series; a processing unit that processes information pertaining to the practice continuation rate for each time-series section obtained by dividing the time series into prescribed time periods; and an inference unit that uses a mathematical model using the information pertaining to the practice continuation rate, the practice content and the environment information or the user state information during the past practice to infer the practice content, the environment information, or the state of the user with a relatively high degree of contribution to the practice continuation rate.

Description

Practice support device, control method for practice support device, and control program for practice support device

[0001] The present disclosure relates to a practice support device, a control method for a practice support device, and a control program for a practice support device.

[0002] Conventionally, there are training support systems that support training for users who run. For example, Patent Document 1 discloses a training support system that presents an appropriate training menu using low-speed jogging and / or high-speed jogging in accordance with changes in the subject's biological information.

[0003] Japanese Unexamined Patent Application Publication No. 2013-244053

[0004] In order to obtain training effects, it is necessary to maintain motivation to accumulate daily training. However, conventional techniques do not take into account training tendencies resulting from the individual characteristics of the user. Therefore, the presented training menu may not arouse the user's interest or motivation, resulting in the user not continuing training. That is, there has been a demand for proposing training that matches the user's individual characteristics and is easy for the user to implement.

[0005] A practice support device according to an aspect of the present disclosure includes: an acquisition unit that acquires practice content, environmental information related to the practice environment, or user status information related to the user's status during a user's past running practice, together with a corresponding time series; a processing unit that processes information related to a practice continuation rate for each time series section obtained by dividing the time series by a predetermined time period; and an estimation unit that estimates practice content, a practice environment, or a user status that has a relatively high degree of contribution to the practice continuation rate, using a mathematical model that uses the information related to the practice continuation rate, the practice content in past practice, the environmental information, or the user status information.

[0006] A control method for a training support device according to one aspect of this disclosure includes: an acquisition step of acquiring training content and environmental information relating to the training environment or user state information relating to the user's state during past driving practice sessions by the user, along with a corresponding time series; a processing step of processing information regarding the training continuation rate for each time series interval obtained by dividing the time series into predetermined time intervals; and an estimation step of estimating training content, training environment, or user state that contribute relatively highly to the training continuation rate using a mathematical model that utilizes the information regarding the training continuation rate, training content during past training sessions, environmental information, or user state information.

[0007] A control program for a training support device according to one aspect of this disclosure includes a command to cause the training support device to perform an acquisition step of acquiring training content and environmental information relating to the training environment or user state information relating to the user's state during past driving practice sessions by the user, along with the corresponding time series; a processing step of processing information regarding the training continuation rate for each time series interval obtained by dividing the time series into predetermined time intervals; and an estimation step of estimating training content, training environment or user state that contribute relatively highly to the training continuation rate using a mathematical model that utilizes the information regarding the training continuation rate, training content during past practice sessions, environmental information or user state information.

[0008] According to one aspect of this disclosure, it is possible to provide a training support device, etc., that proposes training tailored to the user's individuality and that is easy for the user to implement.

[0009] Figure 1 is a schematic diagram showing the configuration of a practice support system according to one aspect of this disclosure. Figure 2 is an example of a functional block diagram of a server (practice support device) and a user terminal (communication terminal) according to one aspect of this disclosure. Figure 3 is a diagram showing an example of the display screen of a user terminal according to one aspect of this disclosure. Figure 4(a) is an example of a dataset according to one aspect of this disclosure, and (b) is a schematic diagram for explaining one aspect of this disclosure. Figures 5(a) and (b) are schematic diagrams for explaining one aspect of this disclosure. Figure 6 is a schematic diagram for explaining one aspect of this disclosure. Figure 7 is a diagram showing an example of the display screen of a user terminal according to one aspect of this disclosure. Figure 8 is a flowchart showing an example of server operation according to one aspect of this disclosure. Figure 9 is a flowchart showing an example of server operation according to one aspect of this disclosure. Figure 10 is a flowchart showing an example of server operation according to one aspect of this disclosure. Figure 11 is a schematic diagram for explaining one aspect of this disclosure. Figure 12 is a diagram showing an example of the display screen of a user terminal according to one aspect of this disclosure. Figure 13 is a schematic diagram for explaining one aspect of this disclosure. Figure 14 is a schematic diagram for explaining one aspect of this disclosure. Figures 15(a) to 15(c) show examples of user terminal display screens according to one aspect of the present disclosure.

[0010] Hereafter, an embodiment of the invention relating to this disclosure will be described using the figures. Note that the figures are examples, and this disclosure is not limited to what is shown in the figures. For example, the illustrated server (practice support device), user terminal (communication device), database server, number and size ratio of sensor devices, dataset (table), display screen, mathematical model, numerical values ​​in the mathematical model, and flowchart are examples, and this disclosure is not limited to these.

[0011] <System Configuration> Figure 1 shows an example configuration of a training support system according to one aspect of this disclosure. The training support system 600 may be an information processing system that supports the training of a user (runner). According to one aspect of this disclosure, based on data (information) obtained from the user's daily training, the system estimates the user's training trends and proposes training content that is easy for the user to continue. In this specification, "training" includes "workout," "training," "discipline," etc., and may include training that an individual does on their own or training that is done under the guidance of an expert or trainer. Furthermore, in the following description, training will be described as running training related to running.

[0012] The practice support system 600 includes a server 100, a database server 101, a user communication terminal (user terminal) 200, and sensor devices 301 and 302.

[0013] The server 100 can perform various processes related to the practice support services realized by the practice support system 600. The server 100 is also connected to the user terminal 200 via the network 500. The network 500 may include wireless networks and wired networks, and may include, for example, wireless LANs (WLANs), wide area networks (WANs), ISDNs (integrated service digital networks), wireless LANs, CDMA (code division multiple access), LTE (long term evolution), LTE-Advanced, 4th generation communication (4G), 5th generation communication (5G), and 6th generation communication (6G) and later mobile communication systems, or combinations thereof.

[0014] Server 100 further transmits and receives various types of data with the database server 101. The database server 101 stores (stores) various types of data necessary for realizing the functions of the practice support system 600. In Figure 1, Server 100 and the database server 101 are shown connected via the network 500, but Server 100 and the database server 101 may also be connected by a dedicated internal network. In Figure 1, one Server 100 and one Database Server 101 are shown, but this is not the only way. That is, each function described as being provided by Server 100 may be realized by multiple servers, and there may be multiple Database Servers 101. Also, Server 100 may be, for example, a distributed server system that operates cooperatively by communicating via a network, or a so-called cloud server. That is, Server 100 is not limited to a physical server, but may also include a virtual server created by software.

[0015] The user terminal 200 is a communication terminal used by the user, and has an application installed for using the training support service (hereinafter also referred to as the "training support app"). The user terminal 200 may transmit information acquired by its own device and sensor devices 301 and 302 to the server 100 via the training support app. The user terminal 200 may also be a communication device with a location information acquisition function such as GPS (Global Positioning System), carried by the user while they are running, and capable of measuring the user's running position.

[0016] In Figure 1, a smartphone is shown as the user terminal 200, but the user terminal 200 can be any terminal that can realize the functions described in each embodiment described below. For example, the user terminal 200 may be a computer (e.g., a tablet terminal), a handheld computer device (not limited to, but for example, a wearable device (glasses-type device (smart glasses), watch-type device (smartwatch), etc.)), a smart speaker, etc.).

[0017] The user carries or wears at least one of the user terminal 200 and sensor devices 301 and 302 while running. Sensor devices 301 and 302 are wearable devices with communication functions that are attached to the user's body and can acquire the user's running information, biometric information, and location information, and transmit the acquired information to the user terminal 200. Sensor device 301 may be a motion sensor that can acquire information about the user's running form and distance traveled, and transmit the acquired information to the user terminal 200. Sensor device 302 may include a vital sensor that can detect biometric information such as the user's body temperature, blood pressure, heart rate, and respiratory rate per unit time, in addition to measuring distance traveled. However, in one aspect of this disclosure, the acquisition of information about running form and biometric information is not essential. That is, according to one aspect of this disclosure, data obtained by running while holding the user terminal 200 is sufficient, and the attachment of motion sensors and vital sensors is not essential.

[0018] Next, we will explain the hardware and functional configurations of the server 100 and user terminal 200 using Figure 2.

[0019] In terms of hardware configuration, the user terminal 200 includes a control unit 210, a communication unit 220, a display unit 230, an input / output unit 240, and a storage unit 270.

[0020] The control unit 210 controls the communication unit 220, the display unit 230, and the input / output unit 240. The control unit 210 is typically a processor, and is implemented by a central processing unit (CPU), an MPU (Micro Processing Unit), a GPU (Graphics Processing Unit), etc. The control unit 210 reads a program stored in the storage unit 270 and executes the code or instructions contained in the program to perform the functions and methods shown in each embodiment.

[0021] The control unit 210 controls the communication between the user terminal 200, the server 100, and the sensor devices 301 and 302, which is performed by the communication unit 220, and transmits and receives various data between them. For example, the control unit 210 may acquire data such as the distance traveled and the travel time as data measured by the sensor devices 301 and 302 during the user's journey.

[0022] The storage unit 270 stores various programs and data necessary for the operation of the user terminal 200. For example, the storage unit 270 may store the program for the practice support application described above. The storage unit 270 may include, for example, flash memory, and may also include memory (RAM (Random Access Memory), ROM (Read Only Memory), etc.) that provides a working area for the control unit 210. It may also temporarily store lap times, which will be described later.

[0023] The communication unit 220 is implemented as hardware such as a NIC (Network Interface Card), a network adapter, communication software, or a combination thereof. The communication unit 220 transmits and receives various data to and from the server 100 and sensor devices 301 and 302 via the network 500 in accordance with a predetermined protocol.

[0024] The display unit 230 is a monitor that displays data according to the display data written to the frame buffer. The display unit 230 may be, for example, a touch panel, a touch display, etc.

[0025] The input / output unit 240 includes an input device for inputting various operations to the user terminal 200, and an output device for outputting processing results processed by the user terminal 200. The input device may include, for example, a touch panel, a touch display, a camera, and a microphone, and the output device may include, for example, a display, a touch panel, a speaker, and the like.

[0026] Next, the functional configuration will be described. The user terminal 200 includes a display processing unit 211, an input / output processing unit 212, and a sensor data acquisition unit 213, as functions realized by the control unit 210. Note that among the functional units shown in Figure 2, functional units that are not essential in each embodiment may be omitted. Furthermore, the functions or processing of each functional unit may be realized by machine learning or AI (Artificial Intelligence) to the extent that it is feasible. Note that some of the various processes described as being performed by the user terminal 200 may be performed by the server 100.

[0027] The display processing unit 211 controls the display of data on the display unit 230. For example, the display processing unit 211 displays the estimation results and practice content suggestions from the server 100 (described later) on the display unit 230.

[0028] The input / output processing unit 212 controls the transmission of various types of information to external devices via the input / output unit 240. The input / output processing unit 212 may transmit various types of information to each functional unit in response to user input operations received by the input device, or transmit information from each functional unit to output devices such as touch panels, monitors, and speakers.

[0029] The sensor data acquisition unit 213 acquires various data measured by the sensor devices 301 and 302.

[0030] Next, the server will be described. The server 100 has a hardware configuration that includes a control unit 110, a communication unit 120, a display unit 130, and a storage unit 170.

[0031] The storage unit 170 is typically implemented using various recording media such as HDDs (Hard Disk Drives), SSDs (Solid State Drives), and flash memory, and has the function of storing various programs and data necessary for the operation of the server 100. The storage unit 170 also includes memory (RAM, ROM, etc.) that provides a working area for the control unit 110.

[0032] The control unit 110 is typically a processor, and is implemented by a central processing unit (CPU), MPU, GPU, etc. The control unit 110 may perform the functions and methods shown in each embodiment by reading a program stored in the storage unit 170 and executing code or instructions contained in the program.

[0033] The communication unit 120 is implemented as hardware such as a NIC and a network adapter, communication software, or a combination thereof. The communication unit 120 may send and receive various types of data to and from the user terminal 200 via the network 500 using any communication protocol.

[0034] Next, the functional configuration will be described. The server 100 includes an acquisition unit 111, a processing unit 112, an estimation unit 113, a generation unit 114, and an output unit 115, as functions realized by the control unit 110. Note that among the functional units shown in Figure 2, functional units that are not essential in each embodiment may be omitted. Furthermore, the functions or processing of each functional unit may be realized by machine learning or AI to the extent that it is feasible.

[0035] The acquisition unit 111 acquires the content of the user's past driving practice sessions, environmental information related to the practice environment, or user status information related to the user's condition, along with the corresponding time series. "Practice content" refers to information that allows evaluation of the intensity and load of the practice performed by the user. Practice content may include, but is not limited to, the distance driven, speed, type of practice, intensity, etc. "Environmental information related to the practice environment" refers to information about the environment and conditions at the time of practice. Environmental information may include, but is not limited to, the weather, temperature, humidity, road surface condition (wet, dry), air quality, etc. "User status information related to the user's condition" refers to information about the user's physical and psychological state. User status information may include, but is not limited to, the stress level, mood, presence or absence of injury, etc.

[0036] Practice content, environmental information, and user status information may be obtained during practice from sensor devices 301 and 302, from the user via a user terminal, or from existing services. For example, practice content may be data such as running speed and distance measured by sensor devices 301 and 302, which may be transmitted to the server 100 via a practice support application on the user terminal 200. Alternatively, an input screen for practice content and practice environment may be displayed on the display unit 230 of the user terminal 200, and the practice content and practice environment entered by the user may be transmitted to the server 100. Alternatively, information such as weather, humidity, and temperature may be obtained from an external service that provides weather information based on location information obtained by sensor devices 301 and 302 or the user terminal 200.

[0037] Furthermore, user status information can also be displayed on the display unit 230 of the user terminal 200, where an input screen is shown allowing the user to input their stress level and mood during practice. The information entered by the user may then be sent to the server 100. Figure 3 shows an example of a user status input screen displayed on the user terminal 200. In the example of screen D10, stress, mood, and physical condition are shown as being entered in stages, but the stress level and mood may also be entered as a score from 1 to 100, for example. Alternatively, the user may select from pre-set options such as "good," "average," and "poor." However, the input method and indicators for user status information are not limited to these.

[0038] Figure 4(a) shows an example of a data table that stores the content of past running practice sessions by a user, environmental information, and user status information, along with time-series information. Note that the figure is just an example, and the types and formats of data stored are not limited to those shown. In the example in Figure 4(a), table TB10 stores the distance run, speed, and intensity as the content of the practice. Note that the intensity may be calculated from the distance and speed. Alternatively, the type of practice, such as jogging, pace running, interval running, or cross-training, may be recorded at the time of practice, and a pre-set intensity according to the type of practice may be stored in table TB10. Alternatively, the type of practice itself may be stored. In the example in Figure 4(a), table TB10 also stores temperature, humidity, and weather as environmental information. Furthermore, table TB10 stores stress and physical condition levels as scores, and whether or not there is a malfunction as user status information. Note that table TB10 may be stored in the database server 101.

[0039] The processing unit 112 processes information regarding the practice continuation rate for each time series interval, which is a predetermined time interval in which the time series is divided. The processing unit 112 may divide the period during which the above-mentioned practice content information is acquired into predetermined time intervals in the time series. The predetermined time is not limited to this, but may be, for example, one week. The information regarding the practice continuation rate is information that indicates the degree to which the user has continued practicing, and may be determined, for example, based on information regarding the days on which practice was performed and the days on which practice was not performed. For example, Figure 4(b) shows the table TB10 divided into one-week time series intervals, with days on which the user practiced marked as "○" and days on which practice was not marked as "×". The processing unit 112 may designate weeks in which the user practiced on three or more days as "intervals with a high practice continuation rate" and weeks in which the user practiced on fewer than three days as "intervals with a low practice continuation rate". In the example in Figure 4(b), sections W1, W3, W4, and W6 are sections with high training continuation rates, while sections W2 and W5 are sections with low training continuation rates. Note that the definition of high / low training continuation rates is not limited to the above.

[0040] The estimation unit 113 uses a mathematical model that incorporates information on the practice continuation rate, the content of past practice sessions, and environmental information or user state information to estimate the practice content, practice environment, or user state that contributes relatively highly to the practice continuation rate.

[0041] The estimation unit 113 estimates the degree of contribution using a mathematical model that calculates conditional probabilities based on predetermined conditions. The mathematical model that calculates conditional probabilities may be Bayesian estimation, Gaussian process regression, etc. Here, Bayesian estimation will be explained as an example.

[0042] The estimation unit 113 estimates, using Bayesian estimation methods, how much practice content and practice environment contribute to the practice continuation rate, based on information obtained during past practice sessions, for example, from table TB10. That is, when information on practice content and practice environment is provided, the estimation unit 113 estimates, using Bayesian estimation methods, the extent to which they contribute to the practice continuation rate. Here, Bayesian estimation may be performed using a Bayesian network with nodes representing the practice continuation rate and each item included in the practice content, environment information, and user state information. Figure 5(a) shows an example of a Bayesian network. In the Bayesian network BN10, items included in various information from the user's past practice sessions may be represented as nodes N11 to N16, and each node N11 to N16 may show the influence it has on node N17 of the practice continuation rate. In other words, the estimation unit 113 performs Bayesian estimation assuming that the information in table TB10 relating to intervals W1, W3, W4, and W6, which have high practice continuation rates in Figure 4(b), contributes to high practice continuation rates, and the information in table TB10 relating to intervals W2 and W5, which have low practice continuation rates in Figure 4(b), contributes to low practice continuation rates.

[0043] The dependencies between nodes, the values ​​of conditional probabilities, and the types of nodes in the Bayesian network described below are examples only, and the present invention is not limited thereto. Furthermore, the dependencies between nodes in the Bayesian network may be defined and described by an analyst with specialized knowledge. Alternatively, the dependencies between nodes in the Bayesian network may be analyzed and described by machine learning.

[0044] Furthermore, when describing the Bayesian network, the processing unit 112 converts continuous variables among the respective items into discrete variables. For example, humidity in environment information may be converted into "high", "low" or "normal" according to the range of the difference value from the average humidity in the corresponding period. Also, for stress in user status information, if the stress is represented by a score for example, it may be converted into "present", "absent" or "normal" according to the magnitude of the score. Further, when the practice intensity is represented by a score, it may be converted into "high", "low" or "medium" according to the magnitude of the score.

[0045] The estimating unit 113 uses the Bayesian network BN10 described based on the user's information from past practice to estimate practice contents, a practice environment, or a user status that have a relatively high degree of contribution to the case of "high practice continuation rate" as a predetermined condition. FIG. 5(b) shows the Bayesian network BN20 when the practice continuation rate is set to "high", that is, when the probability of "high" practice continuation rate at node N27 is set to 100%. In the Bayesian network BN20, the conditional probability of each node may represent the degree of contribution to the practice continuation rate. In the example of FIG. 5(b), from the conditional probability values of nodes N21, N22, N23, and N25, it may be determined that the practice continuation rate tends to be high when the user has "low air temperature", "good mood", "low stress", and "moderate practice intensity".

[0046] Furthermore, the estimating unit 113 estimates the degree of contribution using a mathematical model that classifies information obtained from the acquiring unit 111 and the processing unit 112 based on the practice continuation rate. The mathematical model for classification may be linear discriminant analysis, binomial discriminant analysis, logistic regression, support vector machine, decision tree, or the like. Here, linear discriminant analysis will be described as an example.

[0047] When performing linear discriminant analysis, the processing unit 112 performs conversion to unify the scale by, for example, subtracting an average value from each continuous variable among the respective items and dividing the result by a standard deviation value. Note that any method other than this may be used to unify the scale.

[0048] The estimation unit 113 estimates, using a linear discriminant analysis method, to what extent practice contents, practice environment and the like contribute to the practice continuation rate based on, for example, information in table TB10 acquired during past practice. FIG. 6 is a schematic diagram illustrating estimation by linear discriminant analysis. The estimation unit 113 labels information in table TB10 related to sections W1, W3, W4, and W6 with high practice continuation rates in FIG. 4(b) as data with high practice continuation rates, and labels information in table TB10 related to sections W2 and W5 with low practice continuation rates in FIG. 4(b) as data with low practice continuation rates. The estimation unit 113 uses these data to generate a discrimination model (discriminant function) 80 that discriminates between high and low practice continuation rates. With the discrimination model 80, data of sections with low practice continuation rates and data of sections with high practice continuation rates can be classified as shown in graph G10.

[0049] Here, the discriminant function z can be expressed by the following formula. Note that each feature value constituting the discriminant function z is not limited to the following.

[0050]

[0051] In the discriminant function z, x represents each item (feature) such as practice content, environmental information, or user state information, and the coefficient w indicates how much each feature influences the classification when classifying the data. In other words, the coefficient w indicates the degree of contribution of those features to the discriminant function. Furthermore, as shown in graph G10, a discriminant function is generated such that a larger value of z classifies the data into an interval with a high practice continuation rate, and a smaller value of z classifies the data into an interval with a low practice continuation rate. Therefore, for example, suppose that in the discriminant function z, the absolute values ​​of the coefficient w are largest in the order of speed, practice intensity, stress, and mood. In this case, the estimation unit 113 may estimate that speed, practice intensity, stress, and mood contribute relatively more to the practice continuation rate in this order. The estimation unit 113 may then determine the influence of the magnitude of these features on the practice continuation rate depending on how the value of the discriminant function z changes when the magnitude of these features is changed. For example, suppose that in the discriminant function z, the velocity feature x has a negative coefficient w, and when the value input to the velocity feature x is decreased, the value of the discriminant function z shifts in the positive direction. In this case, the estimation unit 113 may estimate that slowing down the speed leads to an improvement in the practice continuation rate.

[0052] The output unit 115 may output information to be displayed on the user terminal 200, which is the estimation result from the estimation unit 113 described above. Figure 7 shows an example of a display screen of the estimation result shown on the display unit 230 of the user terminal 200. As shown in screen D11, the user may be provided with information about practice content, practice environment, or the user's condition that has been estimated by the estimation unit 113 to have a relatively high contribution to the practice continuation rate. Note that the figure is just an example, and the information and format displayed are not limited to these.

[0053] The control method for the server 100 described above will be explained using the flowchart in Figure 8. The acquisition unit 111 acquires the content of past running practice sessions by the user, environmental information regarding the practice environment, or user status information regarding the user's state, along with the corresponding time series (step T11). As described above, the acquisition unit 111 may acquire this information from the user terminal 200 or a predetermined external service. The information acquired by the acquisition unit 111 may be stored in, for example, the table TB10 in Figure 4(a) and stored in the database server 101. The processing unit 112 processes information regarding the practice continuation rate for each time series interval, which is a predetermined time interval (step T12). For example, the processing unit 112 may determine whether each time series interval of a week is a high or low practice continuation rate interval. The estimation unit 113 uses a mathematical model that incorporates information on the practice continuation rate, the content of past practice sessions, and environmental information or user state information to estimate the practice content, practice environment, or user state that contributes relatively highly to the practice continuation rate (step T13).

[0054] The estimation unit 113 may use mathematical models based on conditional probability or mathematical models related to classification. For example, Figure 9 shows a flowchart when the mathematical model based on conditional probability is the Bayesian estimation described above. Steps S11 and S12 in Figure 9 correspond to steps T11 and T12 in Figure 8, and their explanation is omitted. Steps S12 to S14 in Figure 9 correspond to step T13 in Figure 8. The estimation unit 113 creates a network diagram (Bayesian network) based on the acquired data (step S12). The Bayesian network may be the Bayesian network BN10 described in Figure 5(a). The estimation unit 113 finds the conditional probability of each node when the practice continuation rate is high (step S13). This may refer to finding the conditional probabilities of the other nodes when the conditional probability of a high practice continuation rate at node N27 of the Bayesian network BN20 in Figure 5(b) is set to 100%. Then, the estimation unit 113 acquires nodes with relatively high conditional probabilities (step S14). This is as explained using Figure 5(b). After step S14, the output unit 115 outputs the estimation results to the application (practice support application) (step S17). At this time, the display unit 230 of the user terminal 200 may display, for example, screen D11 as shown in Figure 7. Note that the processing in steps S15 and S16 is not necessarily required. Steps S15 and S16 will be described later.

[0055] Furthermore, Figure 10 shows a flowchart of the case where the mathematical model for classification used in the estimation by the estimation unit 113 is the linear discriminant analysis described above. Steps S21 and S22 in Figure 10 correspond to steps T11 and T12 in Figure 8, and their explanation is omitted. Steps S22 to S24 in Figure 10 correspond to step T13 in Figure 8. The estimation unit 113 generates a discriminant model (discriminant function) for the continuation rate based on the acquired data (step S22). The discriminant function may be the one described in Figure 6. Next, the estimation unit 113 obtains coefficients that have a high influence on the continuation rate based on the discriminant function (step S23). As described above, these may be coefficients with large absolute values ​​in the discriminant function. After that, the estimation unit 113 obtains the conditions under which the acquired coefficients improve the practice continuation rate (step S24). As described above, this may be determined depending on whether the value of the discriminant function transitions in the positive or negative direction when the value of the coefficient is increased or decreased. After step S24, the output unit 115 outputs the estimation result to the application (practice support application) (step S27). At this time, the display unit 230 of the user terminal 200 may display, for example, screen D11 as shown in Figure 7. Note that the processes in steps S25 and S26 do not necessarily need to be performed. Steps S25 and S26 will be described later.

[0056] Thus, according to one aspect of this disclosure, factors contributing to an improvement in the user's practice retention rate can be estimated. Therefore, it is possible to propose training that is tailored to the user's individuality and easy for the user to implement. Furthermore, by using a linear discriminant function as a mathematical model, if label definitions based on retention rates are prepared in advance, a wide range of label types can be distinguished, and the computational load is low.

[0057] Furthermore, according to one aspect of this disclosure, the above estimation is performed by a mathematical model that calculates conditional probabilities based on predetermined conditions. In this case, Bayesian estimation (Bayesian network) may be used as the mathematical model. This makes it possible to describe even if the relationships between factors contributing to the improvement of practice continuation rates are nonlinear or involve interaction phenomena. Moreover, the above estimation is possible even with only a small amount of data on past practice. Furthermore, the Bayesian network makes it possible to explicitly show the causal relationships of each factor, making it easy to understand the relationships between factors, and also allows for the flexible incorporation of empirical rules into the model. In addition, the model can learn on its own by updating it with data based on past practice and accumulating cause and effect. In other words, the accuracy of the estimation can be improved.

[0058] Furthermore, according to one aspect of this disclosure, the above estimation is performed by a mathematical model related to clustering. In this case, a K-means clustering model may be used as the mathematical model. This makes it possible to handle the variability that occurs day by day within, for example, a data set with a "high retention rate". Also, since there is no need to prepare high / low labels for the practice retention rate in advance, the practice retention rate can be flexibly interpreted as high / medium / low, etc. The clustering method is not limited to K-means, but any previously known method may be adopted. Examples of previously known methods include mixture normal distributions, Ward's method, centroid method, shortest distance method, and group average method.

[0059] Furthermore, according to one aspect of this disclosure, information regarding the practice environment in past user practice sessions is taken into consideration in the mathematical model, making it possible to accurately estimate the factors of the practice environment that affect the user's practice retention rate.

[0060] Furthermore, according to one aspect of this disclosure, since the user's state during past practice sessions is taken into consideration in the mathematical model, it is possible to accurately estimate the user's state that affects the user's practice retention rate.

[0061] Furthermore, according to one aspect of this disclosure, since the content of past user practice is taken into consideration in the mathematical model, it is possible to accurately estimate the content of practice that affects the user's practice retention rate.

[0062] According to one embodiment of this disclosure, practice content, practice environment, or user state that contribute relatively highly to the practice continuation rate is estimated according to the conditions of the scheduled practice day. This will be explained using Figure 11. First, a Bayesian network BN10 shown in Figure 5(a) is generated based on the user's past practice information. Now, suppose that on the scheduled practice day, the weather is "sunny," the user's mood is "good," stress is "normal," humidity is "low," and temperature is "high." The estimation unit 113 adjusts the conditional probabilities of the weather node N11, mood node N12, stress node N13, humidity node N14, and temperature node N15 in the Bayesian network BN10 to match the conditions of the scheduled practice day. That is, each of the above nodes may be set to "100%" for items that match the conditions of the scheduled practice day, as shown in the Bayesian network BN30 in Figure 11. Then, at the practice continuation rate node N37, the "high" practice continuation rate is set to "100%." The estimation unit 113 may estimate that the user's practice continuation rate will be higher when the practice intensity is "medium," based on the conditional probability at node N36 of the practice intensity in this case.

[0063] Thus, according to one aspect of this disclosure, practice content is presented that is easy for the user to continue practicing, depending on the environment on the day the user practices and the user's physical and psychological state. This can improve the user's practice retention rate.

[0064] In the example shown in Figure 11, the state of the scheduled practice day is set in nodes other than the practice intensity node in the Bayesian network BN30. However, setting the state of each node is not mandatory, and nodes for which the state of the scheduled practice day is unknown do not need to be set. For example, if the temperature and humidity of the scheduled practice day are unknown, only the node related to the user's state may be set. In this case, the estimation unit 113 may estimate the levels of practice intensity, humidity, and temperature that will increase the user's practice continuation rate.

[0065] Next, steps S15-S17 and S25-S27, respectively, in the flowcharts of Figures 9 and 10 will be explained. The generation unit 114 generates a new practice content proposal based on the level of practice content estimated by the estimation unit 113. The generation unit 114 may input the prompt generated based on the level of practice content estimated by the estimation unit 113 into a pre-prepared large-scale language model, and the output from the large-scale language model may be used as the new practice content proposal. The prompt may be an instruction to the large-scale language model to update the practice content originally planned for the scheduled practice day according to the estimation result by the estimation unit 113. For example, as in the example in Figure 11, suppose the estimation unit 113 estimates that the user's practice continuation rate will be high when the practice intensity is "medium" on a certain scheduled practice day. In this case, the practice intensity of the originally planned practice content is "strong," and there is a possibility that the user will not practice with the originally planned practice content. At this time, the generation unit 114 may generate prompts instructing the user to change the training intensity from "strong" to "medium" and to propose specific training content. For example, the generation unit 114 may generate a message instructing the user to modify the planned "fast pace run of 5 minutes / km for 10km" to a low-intensity training session. The generation unit 114 may also input the generated prompts into a large-scale language model and use the output content obtained as a proposal for new training content. In the above example, the large-scale language model may output training content of "low-intensity pace run of 6 minutes to 6.5 minutes / km for 10km." The output unit 115 may then output this proposal for new training content to the user terminal 200. Figure 12 shows an example of the screen of the user terminal 200 at this time. As shown in screen D20, information about the proposed new training content may be displayed on the user terminal 200. Note that the prompts and output content are not limited to those described above. For example, a prompt may be generated instructing the user to change the planned "5km jog" to "skip the workout" (take a break from training). Then, in response to this prompt, the text "If you are busy, don't push yourself and take a rest" may be output to the user terminal 200 by the large-scale language model.

[0066] Thus, according to one aspect of this disclosure, specific practice content is proposed to the user. Therefore, it is possible to provide a practice support system that further contributes to the continuation of practice without requiring the user to think of practice content.

[0067] Furthermore, the above-mentioned proposal of a new practice schedule may also be made for multiple future practice dates. This will be explained using Figure 13. In Figure 13, for each of the future dates, November 1st, 2nd, and 3rd, the weather and temperature forecast, and the expected mood and stress levels are input to each node of the Bayesian network BN10. Then, the practice intensity required to keep the user practicing may be estimated from, for example, the conditional probability value of practice intensity in the Bayesian network BN40, based on the result of setting the probability of "high" in the practice continuation rate node to 100%. The generation unit 114 may then propose the output result of inputting the prompt generated based on the estimation result into a large-scale language model as a new practice schedule. Furthermore, the user's requests may be taken into consideration when proposing a new practice schedule. For example, if the generation unit 114 receives a request to take a break from practice via the practice support application, it may generate a prompt suggesting that the user take a break from practice. Table TB20 in Figure 13 shows an example of how the planned practice content may have been modified according to the estimation results by the estimation unit 113 or the user's request.

[0068] In endurance sports such as running, there are obstacles to training, including the fact that repetitive movements increase the load on specific body parts, which can be a cause of injury, and that the long duration of each training session places a significant psychological burden. On the other hand, if training is suspended, cardiopulmonary function declines, and the training effect is not achieved. According to one aspect of this disclosure, an optimal training menu is proposed, taking into account indicators that evaluate these physical and psychological burdens.

[0069] This will be explained using Figure 14. First, the acquisition unit 111 acquires information regarding the user's psychological changes before and after practice. For example, the acquisition unit 111 may acquire a score indicating the user's own psychological state (mood) before practice, which is entered by the user in the practice support application. Similarly, after practice, a score indicating the user's psychological state is acquired, and the change in score before and after practice may be stored in advance. Figure 14(a) is an example of a table that stores the change in the score indicating the user's psychological state before and after practice for each type of practice. In the example of table TB30, it can be seen that the user's score decreases by "15 points" in the case of "interval running" and by "2 points" in the case of "pace running". In addition, the acquisition unit 111 may acquire the total distance run, total time, total number of practice sessions, and the user's subjective fatigue level as indicators of physical load, and use them as cumulative load values.

[0070] The estimation unit 113 may include nodes related to the score of the psychological state for each training session in the Bayesian network. The estimation unit 113 may also include nodes related to physical load and nodes related to injury in the Bayesian network. Figure 14(b) is an example of a Bayesian network that includes nodes N58 and N59 related to the score of the psychological state for each training session, node N60 related to physical load, and node N61 related to injury. The Bayesian network BN50 reflects the training environment, the user's condition, and physical load on the target day of estimation, and the result obtained is that the training intensity with a high continuation rate is "medium". Referring to nodes N58 and N59 related to the score of the psychological state for each training session, it is estimated from the conditional probability value that pace running (node ​​N58) prevents a decrease in the score of the psychological state. Therefore, the generation unit 114 may generate a suggestion to perform "pace running" at "medium intensity".

[0071] Thus, according to one aspect of this disclosure, a training menu is proposed that takes into further consideration the user's physical load and the user's preferences regarding the type of training. Therefore, it is possible to propose a training menu that is easier to implement.

[0072] <Example of User Terminal Screen> Figure 15 shows an example of a display screen of a user terminal 200 in one embodiment of the present disclosure. Note that the figure is just an example and the present disclosure is not limited thereto. Figure 15(a) is an example of a screen for inputting initial settings when using the above-mentioned training support service using a training support application. Screen D30 may include an input area 31 for subjective evaluation of the user's personality and tendencies, an input area 32 for the purpose of training, an selection area 33 for whether or not to link with an external application regarding the schedule, and an selection area 34 for concerns regarding the progress of training. Figure 15(b) is an example of a screen that displays the training menu for a training day. Screen D40 may include the training menu 41 for the training day. Note that the training menu 41 may display a modified training menu based on the estimation results using the mathematical model described above. It may also include an area 42 that displays the user's evaluation according to the user's past training content. Figure 15(c) is an example of a screen that displays suggestions to the user. Screen D50 may include an area 51 that displays running courses, training menus, etc., recommended to the user. It may also include a button 52 that allows users to register for various running-related events. Furthermore, it may include an area 53 that displays advice and other information to reduce the user's physical and psychological burden.

[0073] This disclosure has been described based on various drawings and embodiments, but it should be noted that those skilled in the art will find it easy to make various modifications and alterations based on this disclosure. Therefore, it should be noted that these modifications and alterations are within the scope of this disclosure. For example, the functions included in each means, each step, etc., can be rearranged in a logically consistent manner, and multiple means or steps, etc., can be combined into one or divided. Furthermore, the configurations shown in the above embodiments may be combined as appropriate. For example, each component described as being provided by server 100 may be implemented by multiple servers in a distributed manner. Also, the processing described as a function of server 100 may be performed by user terminal 200. Conversely, processing described as being performed by user terminal 200 may be performed by server 100.

[0074] For example, the above described the degree of contribution to the training continuation rate. However, this disclosure is not limited to the training continuation rate, and may include, for example, the growth rate, injury rate, recovery rate, psychological or physical recovery rate, etc.

[0075] Furthermore, the above described estimation using Bayesian networks (Bayesian estimation) and linear discriminant methods. However, this disclosure is not limited to the methods described above, and existing methods such as expected value, variance, binomial discriminant analysis, regression, and support vector machines may be used.

[0076] In addition to the above, environmental information may also include wind speed, wind direction, type of road surface (asphalt, concrete, soil, tartan, grass, artificial turf, wood chips, etc.), traffic conditions (presence or absence of cars and pedestrians, etc.), safety (public safety), and time of day (morning, noon, night).

[0077] In addition to the above, user status information may also include resting heart rate, exercise heart rate, muscle mass (power, flexibility), fatigue level (tired, normal, energetic), running skill (beginner, intermediate, advanced), diet, concentration level, etc.

[0078] Furthermore, the above describes the manner in which training content (training menus) are proposed. However, this disclosure is not limited to this. For example, suggestions to improve training retention rates may include not only training menus, but also training locations, training times, competitions, communities, etc. Running shoes, clothing, etc. may also be suggested. In addition, information about the user's diet may be used to estimate the user's tendencies, for example, that a high calorie intake leads to long-distance running. Based on this, training menus with a high retention rate may be suggested in conjunction with diet, taking into account meal content and timing, or meal content may be suggested in accordance with the planned training menu. Furthermore, training menus that improve retention rates may be suggested based on periodization.

[0079] Furthermore, if the training intensity estimated by the estimation unit 113 is the same as the intensity of the training content originally planned, the training intensity and training content do not need to be changed.

[0080] Each functional unit of the server 100 or the practice support device 100 may be implemented by logic circuits (hardware) or dedicated circuits formed on an integrated circuit (IC (Integrated Circuit) chip, LSI (Large Scale Integration)), or by software using a CPU (Central Processing Unit). Furthermore, each functional unit may be implemented by one or more integrated circuits, and the functions of multiple functional units may be implemented by a single integrated circuit.

[0081] The programs of each embodiment of this disclosure may be provided stored in a storage medium readable by the practice support device. The storage medium is a “non-temporary tangible medium” capable of storing programs. The programs include, for example, software programs and practice support device programs. When each functional unit of the practice support device 100 is implemented by software, the practice support device 100 functions as an acquisition unit 111, a processing unit 112, an estimation unit 113, a generation unit 114, and an output unit 115, with the processor executing a program loaded into memory.

[0082] The storage medium may, where appropriate, include one or more semiconductor-based or other integrated circuits (ICs) (e.g., field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), hard disk drives (HDDs), hybrid hard drives (HHDs), optical disks, optical disk drives (ODDs), magneto-optical disks, magneto-optical drives, floppy diskettes, floppy disk drives (FDDs), magnetic tapes, solid-state drives (SSDs), RAM drives, secure digital cards or drives, any other suitable storage medium, or two or more suitable combinations thereof. The storage medium may, where appropriate, be volatile, non-volatile, or a combination of volatile and non-volatile.

[0083] Furthermore, the program disclosed herein may be provided to the practice support device 100 via any transmission medium capable of transmitting the program (such as a communication network or broadcast waves).

[0084] Furthermore, each embodiment of this disclosure can also be realized in the form of data signals embedded in a carrier wave, where the program is embodied by electronic transmission. The programs of this disclosure may be implemented using, for example, scripting languages ​​such as JavaScript® and Python, C language, Go language, Swift®, Koltin®, Java®, etc.

[0085] Those skilled in the art will understand that the embodiments described above are specific examples of the following embodiments: [1] The practice support device of the present disclosure includes: an acquisition unit that acquires practice content and environmental information relating to the practice environment or user state information relating to the user's state during past driving practice sessions by a user, along with a corresponding time series; a processing unit that processes information regarding the practice continuation rate for each time series interval obtained by dividing the time series into predetermined time intervals; and an estimation unit that estimates practice content, practice environment or user state that contribute relatively highly to the practice continuation rate using a mathematical model that uses the information regarding the practice continuation rate, the practice content during past practice sessions, the environmental information, or the user state information. [2] In the practice support device of [1] above, the estimation unit may estimate the degree of contribution using a mathematical model that calculates a conditional probability based on predetermined conditions. [3] In the practice support device of [1] above, the estimation unit may estimate the degree of contribution using a mathematical model that classifies the information obtained from the acquisition unit and the processing unit based on the practice continuation rate. [4] In the training support device described in [1] above, the estimation unit may estimate the degree of contribution using a mathematical model that clusters the information obtained from the acquisition unit and the processing unit based on the training continuation rate. [5] In the training support device described in [1] above, the acquisition unit may acquire at least one of the weather, road surface conditions, temperature, humidity, and sunshine duration during training as environmental information relating to the training environment. [6] In the training support device described in [1] above, the acquisition unit may acquire at least one of the user's psychological state, physical condition, presence or absence of injury during training, and information relating to changes in the user's psychological state before and after training as user status information. [7] In the training support device described in [1] above, the acquisition unit may acquire at least one of the training intensity, training location, event, clothing, and shoes as training content. [8] In the practice support device described in any of [1] to [4] above, the estimation unit may use a mathematical model generated by at least one of Bayesian estimation and linear discriminant analysis as the mathematical model.[9] In the practice support device described in [1] above, the acquisition unit may further include an output unit that acquires a scheduled practice date for which practice is proposed to the user, and outputs to the user's communication terminal the practice content, practice environment, or user state that is estimated by the estimation unit to contribute relatively highly to the practice continuation rate, along with the scheduled practice date.

[10] The practice support device described in [9] above further includes a generation unit that generates a proposal for new practice content based on the practice content, practice environment, or user state estimated by the estimation unit, and the generation unit inputs a prompt generated based on the practice content, practice environment, or user state estimated by the estimation unit into a pre-prepared large-scale language model to obtain output content which is the proposal for new practice content, and the output unit may output the proposal for new practice content to the user terminal.

[11] The control method of the present disclosure includes an acquisition step in which a practice support device acquires practice content and environmental information relating to the practice environment or user state information relating to the user's state during past driving practice sessions by a user, along with a corresponding time series; a processing step in which the time series is divided into time intervals of predetermined time and processes information relating to the practice continuation rate; and an estimation step in which a mathematical model using the information relating to the practice continuation rate, the practice content during past practice sessions, the environmental information, or the user state information estimates practice content, practice environment, or user state that contribute relatively highly to the practice continuation rate.

[12] The control program of the present disclosure includes a command to cause a practice support device to execute: an acquisition step of acquiring practice content and environmental information relating to the practice environment or user state information relating to the user's state during past driving practice sessions by the user, along with a corresponding time series; a processing step of processing information relating to the practice continuation rate for each time series interval obtained by dividing the time series into predetermined time intervals; and an estimation step of estimating practice content, practice environment or user state that contribute relatively highly to the practice continuation rate using a mathematical model that uses the information relating to the practice continuation rate, the practice content during past practice sessions, the environmental information or the user state information.

[0086] 100 Server (Practice Support Device) 110 Control Unit 111 Acquisition Unit 112 Processing Unit 113 Estimation Unit 114 Generation Unit 115 Output Unit 120 Communication Unit 130 Display Unit 170 Storage Unit 101 Database Server 200 User Terminal (Communication Terminal) 210 Control Unit 211 Display Processing Unit 212 Input / Output Processing Unit 213 Sensor Data Acquisition Unit 220 Communication Unit 230 Display Unit 240 Input / Output Unit 270 Storage Unit 301 Sensor Device 302 Sensor Device 500 Network 600 Practice Support System

Claims

1. A training support device comprising: an acquisition unit that acquires the content of training and environmental information relating to the training environment or user status information relating to the user's state during past training sessions by the user, along with a corresponding time series; a processing unit that processes information regarding the training continuation rate for each time series interval obtained by dividing the time series into predetermined time intervals; and an estimation unit that estimates training content, training environment, or user status that contribute relatively highly to the training continuation rate using a mathematical model that utilizes the information regarding the training continuation rate, the content of training during past training sessions, the environmental information, or the user status information.

2. The practice support device according to claim 1, wherein the estimation unit estimates the degree of contribution using a mathematical model that calculates a conditional probability based on predetermined conditions.

3. The practice support device according to claim 1, wherein the estimation unit estimates the degree of contribution using a mathematical model that classifies the information obtained from the acquisition unit and the processing unit based on the practice continuation rate.

4. The practice support device according to claim 1, wherein the estimation unit estimates the degree of contribution using a mathematical model that clusters the information obtained from the acquisition unit and the processing unit based on the practice continuation rate.

5. The training support device according to claim 1, wherein the acquisition unit acquires at least one of the weather conditions, road surface conditions, temperature, humidity, and sunshine duration during training as environmental information relating to the training environment.

6. The practice support device according to claim 1, wherein the acquisition unit acquires, as user status information, at least one of the following: the user's psychological state during practice, physical condition, presence or absence of injury, and information regarding changes in the user's psychological state before and after practice.

7. The training support device according to claim 1, wherein the acquisition unit acquires at least one of the training intensity, training location, event, clothing, and shoes as the training content.

8. The training support device according to claims 1 to 4, wherein the estimation unit uses a mathematical model generated by at least one of Bayesian estimation and linear discriminant analysis as the mathematical model.

9. The practice support device according to claim 1, further comprising: an acquisition unit that acquires a practice schedule date for which practice is proposed to the user; and an output unit that outputs to the user's communication terminal, together with the practice schedule date, the practice content, environmental information, or user status information that are estimated by the estimation unit according to the status of the practice schedule date and which have a relatively high contribution to the practice continuation rate.

10. The practice support device according to claim 9, further comprising a generation unit that generates a proposal for new practice content based on the practice content estimated by the estimation unit, the practice environment, or the user's state, wherein the generation unit inputs a prompt generated based on the practice content estimated by the estimation unit, the practice environment, or the user's state to a pre-prepared large-scale language model, and the output content obtained is the proposal for new practice content, and the output unit outputs the proposal for new practice content generated by the generation unit to the user's communication terminal.

11. A control method for a training support device, comprising: an acquisition step of acquiring the content of training and environmental information relating to the training environment or user state information relating to the user's state during past driving practice sessions by a user, along with a corresponding time series; a processing step of processing information regarding the training continuation rate for each time series interval obtained by dividing the time series into predetermined time intervals; and an estimation step of estimating training content, environmental information or user state that contribute relatively highly to the training continuation rate using a mathematical model that utilizes the information regarding the training continuation rate, the content of training during past training sessions, the environmental information or the user state information.

12. A control program for a training support device, which includes a command to cause the training support device to execute: an acquisition step of acquiring the content of training and environmental information relating to the training environment or user state information relating to the user's state during past driving practice sessions by the user, along with a corresponding time series; a processing step of processing information regarding the training continuation rate for each time series interval obtained by dividing the time series into predetermined time intervals; and an estimation step of estimating training content, environmental information or user state that contribute relatively highly to the training continuation rate using a mathematical model that utilizes the information regarding the training continuation rate, the content of training during past training sessions, the environmental information or the user state information.

Images