JP2021171375A - Information processing device, information processing method, and program - Google Patents

Abstract

To provide an information processing device, an information processing method and a program for appropriately determining a pointer for user's walking training.SOLUTION: A server 300 includes an acquisition part 322 for acquiring first data representing a measurement result of the movement of a user during the n-th walking training from a walking assist device, and acquiring second data representing a selection result of a parameter selected by the user from among a plurality of parameters to be a pointer during the n-th walking training from a terminal device, and a determination part 324 for inputting the first data and the second data during the n-th walking training of the user acquired by the acquisition part to a preliminarily learned model, and determining the recommendation parameter to be a point for the n+k-th walking training of the user on the basis of an output result of the model to which the first data and the second data are inputted.SELECTED DRAWING: Figure 7

Description

The present invention relates to an information processing device, an information processing method, and a program.

A walking assist device for assisting a user's walking is known (see, for example, Patent Document 1). On the other hand, there is known a technique of extracting a topic from a text using PLSA (Probabilistic Latent Semantic Analysis) and generating a Bayesian network using the extracted topic as an explanatory variable (see, for example, Patent Document 2).

JP-A-2016-214504 Japanese Unexamined Patent Publication No. 2016-51220

However, in the conventional technique, even if there is no guidance from a doctor or a physiotherapist, when a user wears a walking assist device and performs walking training by himself / herself, it is necessary to appropriately determine what kind of guideline should be used for training. Was not possible in some cases.

The present invention has been made in consideration of such circumstances, and an object of the present invention is to provide an information processing device, an information processing method, and a program capable of appropriately determining a guideline for walking training of a user. Let's try.

The information processing apparatus, information processing method, and program according to the present invention have the following configurations.
(1) A first aspect of the present invention is to measure the movement of the user during the nth walking training from a walking assist device that measures the movement when the user walks while assisting the user's walking. The selection result of at least one parameter selected by the user from a plurality of parameters that serve as a guideline for the nth walking training from the terminal device available to the user by acquiring the first data representing the result. When at least the first data and the second data of the user who has performed the mth walking training and the acquisition unit for acquiring the second data representing the above are input, the m + kth walking training of the user is performed. The first data and the second data at the time of the nth walking training of the user acquired by the acquisition unit are applied to the model trained to output the recommended parameters that should sometimes be used as a guideline. Based on the output result of the model in which the first data and the second data are input, a determination unit for determining the recommended parameters to be used as a guideline for the n + kth walking training of the user. It is an information processing device provided.

(2) The second aspect of the present invention is the information processing device according to the above (1), wherein the acquisition unit further responds to the n + kth walking training questionnaire from the terminal device by the user. The third data representing the above is acquired, and the determination unit determines the first data and the second data of the user at the time of the nth walking training, and the third data of the user at the time of the n + kth walking training. Data is input to the model, and based on the output result of the model in which the first data, the second data, and the third data are input, it is used as a guideline for the n + kth walking training of the user. It determines the recommended parameters to be used.

(3) A third aspect of the present invention is the information processing apparatus according to the above (1) or (2), wherein the model is any one of a deep neural network, a topic model, and a Bayesian network. It is a combination of one model or a plurality of models.

(4) A fourth aspect of the present invention is the information processing apparatus according to any one of the above (1) to (3), wherein the first data of the user and the second aspect of the user at the mkth time. The recommended parameters output by the model during the m-th walking training of the user when data is input to the model, and the parameters selected by the user from the plurality of parameters during the m-th walking training. A learning unit for learning the model is further provided so as to match with.

(5) A fifth aspect of the present invention is from a walking assist device in which a computer measures the movement of the user when the user walks while assisting the user to walk, to the user at the time of the nth walking training. At least one parameter selected by the user from a plurality of parameters that serve as a guideline for the nth walking training from the terminal device that acquires the first data representing the motion measurement result and is available to the user. When at least the first data and the second data of the user who has acquired the second data representing the selection result of and performed the mth walking training are input, the m + kth walking training of the user is performed. The first data and the second data at the time of the nth walking training of the acquired user are input to the model trained to output the recommended parameters that should be used as a guide, and the second data is input. This is an information processing method for determining the recommended parameters to be used as a guideline for the n + kth walking training of the user based on the output result of the model in which the first data and the second data are input.

(6) A sixth aspect of the present invention is from a walking assist device that measures the movement of the user when the user walks while assisting the user to walk on a computer, to the user at the time of the nth walking training. Acquiring the first data representing the motion measurement result, at least one selected by the user from a plurality of parameters that serve as a guideline for the nth walking training from the terminal device available to the user. Acquiring the second data representing the parameter selection result, and when at least the first data and the second data of the user who has performed the m-th walking training are input, the user's m + k-th walking The first data and the second data at the time of the nth walking training of the acquired user are input to the model trained to output the recommended parameters that should be used as a guideline at the time of training. A program for determining the recommended parameters to be used as a guideline for the n + kth walking training of the user based on the output result of the model to which the first data and the second data are input. Is.

According to the above aspect, the guideline for walking training of the user can be appropriately determined.

It is a figure which shows an example of the structure of the walking assist system which concerns on embodiment. It is a figure which shows an example of the structure of the walking assist device which concerns on embodiment. It is a figure which shows an example of the structure of the terminal apparatus which concerns on embodiment. It is a figure which shows an example of the content of a walking training application. It is a figure which shows other example of the content of a walking training application. It is a figure which shows other example of the content of a walking training application. It is a figure which shows an example of the configuration of the server which concerns on embodiment. It is a flowchart which shows an example of the flow of a series of processing of the server which concerns on embodiment. It is a figure which shows an example of a parameter estimation model. This is an example of a database that combines walking training measurement data, parameter selection data, and questionnaire response data. It is a figure for demonstrating a topic model. It is a figure which shows an example of a Bayesian network.

Hereinafter, the information processing apparatus, the information processing method, and the embodiment of the program of the present invention will be described with reference to the drawings.

[System configuration]
FIG. 1 is a diagram showing an example of the configuration of the walking assist system 1 according to the embodiment. The walking assist system 1 according to the embodiment includes, for example, a walking assist device 100, a terminal device 200, and a server 300. The walking assist device 100 and the terminal device 200 are connected by wireless communication such as Wi-Fi or Bluetooth (registered trademark, hereinafter omitted). The terminal device 200 and the server 300 are connected via a network NW such as a LAN (Local Area Network) or a WAN (Wide Area Network), for example. The server 300 is an example of an "information processing device".

[Structure of walking assist device]
FIG. 2 is a diagram showing an example of the configuration of the walking assist device 100 according to the embodiment. The walking assist device 100 according to the embodiment includes, for example, a main frame 102, subframes 103L and 103R, drive sources 104L and 104R, a control device 105, hip joint angle sensors 106L and 106R, and a battery 107.

The main frame 102 is mounted on the user's torso. The main frame 102 is formed by combining a rigid material such as hard resin or metal and a flexible material such as fiber. A belt 111 is attached to the main frame 102, and the belt 111 is attached to the user's waist. A waist supporter 112 made of a flexible material is attached to the front surface of the main frame 102 (position facing the back surface of the waist).

Each of the subframes 103L and 103R is connected to the main frame 102 so that it can be displaced about the hip joint of the user. The subframe 103L is mounted on the user's left leg, and the subframe 103R is mounted on the user's right leg. The subframe 103L includes a leg supporter 113L and an arm portion 114L. The subframe 103R includes a leg supporter 113R and an arm portion 114R. Each of the leg supporters 113L and 113R is configured by combining a rigid material and a flexible material, and is attached to the thigh. Each of the arm portions 114L and 114R is formed of hard resin or metal. Each of the arm portions 114L and 114R extends downward along the thigh portion, and connects the output shaft of the drive source 104 and the leg supporter 113. That is, the subframes 103L and 103R are connected to the main frame 102 via the drive source 104.

The drive source 104L displaces the subframe 103L mounted on the user's left leg with respect to the main frame 102. The drive source 104R displaces the subframe 103R mounted on the user's right leg with respect to the main frame 102. Each of the drive sources 104L and 104R is, for example, an actuator such as a motor, and appropriately includes one or both of a reduction mechanism and a compliance mechanism. Each of the drive sources 104L and 104R is controlled by the control device 105 so as to output the assist torque. Under the control of the control device 105, each of the drive sources 104L and 104R uses the power of the battery 107 to power the arm portion 114. The power applied to the arm portion 114 is transmitted to the user's leg body via the leg supporter 113.

The control device 105 is housed in, for example, the main frame 102. The control device 105 may be attached to or stored in the subframe 103 instead of being housed in the main frame 102, or may be provided separately from the walking assist device 100.

The control device 105 is, for example, an electronic circuit unit including a CPU (Central Processing Unit), a RAM (Random Access Memory), a ROM (Read Only Memory), and the like. Further, the electronic circuit unit includes a wireless communication module such as an antenna, a transmitter, and a receiver.

The control device 105 controls the drive sources 104L and 104L to adjust the assist torque applied to the user. For example, the control device 105 calculates the difference angle between the left and right hip joint angles (the angle between the left and right hip joints) based on the user's hip joint angles detected by the hip joint angle sensors 106L and 106R, respectively. do. The control device 105 calculates the phase of the difference angle and the walking frequency based on the calculated difference angle. Then, the control device 105 determines the assist torque for the left and right legs based on the phase of the difference angle. Specifically, the control device 105 calculates the phase of the virtual vibrator that synchronizes with the user's walking based on the calculated phase of the difference angle and the walking frequency, and the left and right based on the phase of the vibrator. Determine the assist torque for the leg body.

Further, based on the detection results of the hip joint angle sensors 106L and 106R, the control device 105 further increases the number of steps, the average stride length, the average walking speed, the hip joint movable angle, the extension, and the difference angle (pinching angle). The training parameters described later, such as the time elapsed during flexion (extension flexion time) and the degree of left-right symmetry during walking, are also calculated.

The hip joint angle sensor 106L detects the relative angle between the subframe 103L and the main frame 102 as the angle of the hip joint of the user's left leg. The hip joint angle sensor 106R detects the relative angle between the subframe 103R and the main frame 102 as the angle of the hip joint of the user's right leg. Each of the hip joint angle sensors 106L and 106R is, for example, an absolute type angle sensor and is arranged next to the waist of the user. Each of the hip joint angle sensors 106L and 106R outputs a signal indicating the detected hip joint angle to the control device 105.

The battery 107 supplies electric power to the drive sources 104L and 104L and the control device 105. The battery 107 is housed inside the mainframe 102, for example. The battery 107 may be attached to or housed in the subframe 103, and may be provided separately from the walking assist device 100.

In this way, the walking assist device 100 causes the user to act as the walking assist torque by the power of the drive source 104 powered by the battery 107 via the main frame 102 and the subframes 103L and 103R, so that the user can walk. Assist exercise.

[Terminal device configuration]
FIG. 3 is a diagram showing an example of the configuration of the terminal device 200 according to the embodiment. The terminal device 200 according to the embodiment includes, for example, a communication unit 202, a touch panel 204, a control unit 220, and a storage unit 230.

The communication unit 202 includes, for example, a wireless communication module such as an antenna, a transmitter, and a receiver. The communication unit 202 wirelessly communicates with the control device 105 of the walking assist device 100, and communicates with the server 300 via the network NW.

The touch panel 204 is a user interface in which, for example, a display device such as an LCD (Liquid Crystal Display) or an organic EL (Electroluminescence) display and an input device such as a touch pad are combined. The display device and the input device may not be integrally configured as the touch panel 204, but may be provided separately.

The control unit 220 is realized by, for example, a processor such as a CPU or a GPU (Graphics Processing Unit) executing a program stored in the storage unit 230. The control unit 220 may be realized by hardware such as LSI (Large Scale Integration), ASIC (Application Specific Integrated Circuit), or FPGA (Field-Programmable Gate Array), or may be realized by collaboration between software and hardware. May be done.

The storage unit 230 is realized by, for example, an HDD (Hard Disk Drive), a flash memory, an EEPROM (Electrically Erasable Programmable Read Only Memory), a ROM, a RAM, or the like. Various programs such as firmware and application programs are stored in the storage unit 230.

For example, a dedicated application that supports walking training when the walking assist device 100 is attached from among a plurality of applications installed in the terminal device 200 by the user operating the touch panel 204 (hereinafter, referred to as a walking training application). ) Is selected. In this case, the control unit 220 executes the walking training application in response to the user's selection operation on the touch panel 204, and displays various contents on the touch panel 204.

FIG. 4 is a diagram showing an example of the contents of the walking training application. As shown in the figure, when the walking training application is executed, the walking training menu for each user is automatically displayed as content on the touch panel 204. In the walking training menu, some training parameters (also referred to as training items) that are guidelines or targets for walking training are presented.

The training parameters include, for example, a higher training parameter that determines a rough training policy and a lower training parameter that determines a more detailed training policy. The upper training parameters include, for example, three types of parameters such as A, B, and C. Parameter A is, for example, a parameter related to "walking" during walking training. Parameter B is, for example, a parameter related to "speed" during walking training. Parameter C is, for example, a parameter related to "endurance" during walking training.

On the other hand, the lower training parameters include 15 types of parameters such as a to o. Parameter a is, for example, a parameter related to “measurement time”. Parameter b is, for example, a parameter related to “walking time”. Parameter c is, for example, a parameter related to “step count”. The parameter d is, for example, a parameter related to the “average stride length”. The parameter e is, for example, a parameter related to the “average pace”. The parameter f is, for example, a parameter relating to “average walking speed”. The parameter g is, for example, a parameter relating to the “average walking ratio”. The parameter h is, for example, a parameter related to “walking distance”. The parameter i is, for example, a parameter related to the “hip joint movable angle”. The parameter j is, for example, a parameter related to the “sandwich angle”. The parameter k is, for example, a parameter relating to “extension / flexion time”. The parameter l is, for example, a parameter related to “symmetry”. The parameter m is, for example, a parameter related to “waveform”. The parameter n is, for example, a parameter related to “moving image”. Parameter o is, for example, a parameter related to "waveform and moving image".

The training parameters are not limited to the above-mentioned examples, and other parameters may be added, and some parameters may be replaced with other parameters or omitted. Further, the training parameters do not have to have a hierarchical structure such as upper and lower levels.

5 and 6 are diagrams showing other examples of the content of the walking training application. For example, when the user selects the parameter d as the lower training parameter, the touch panel 204 displays the measured value related to the “average pace” as illustrated in FIG. Further, when the user selects the parameters i, j, and k as the lower training parameters, the touch panel 204 has the "hip joint movable angle", "pinching angle", and "extension / flexion time" as illustrated in FIG. The related measurement values are displayed.

The measured values of these training parameters are acquired from the walking assist device 100. For example, the control device 105 of the walking assist device 100 measures all the above training parameters and transmits the measurement results to the terminal device 200. When the control unit 220 of the terminal device 200 receives the measurement results of all the training parameters from the walking assist device 100 via the communication unit 202, the control unit 220 touches the touch panel according to the selection result of the user's parameters from the plurality of measurement results. Select the measurement result to be displayed on 204. Then, the control unit 220 displays the selected measurement result on the touch panel 204.

[Server configuration]
Hereinafter, the configuration of the server 300 will be described. The server 300 may be a single device, or may be a system in which a plurality of devices connected via a network NW operate in cooperation with each other. That is, the server 300 may be implemented by a plurality of computers (processors) included in a distributed computing system or a cloud computing system.

FIG. 7 is a diagram showing an example of the configuration of the server 300 according to the embodiment. As shown in the figure, the server 300 includes, for example, a communication unit 302, a control unit 320, and a storage unit 330.

The communication unit 302 includes, for example, a NIC (Network Interface Card). The communication unit 302 communicates with the terminal device 200 and the like via the network NW.

The control unit 320 includes, for example, an acquisition unit 322, a determination unit 324, and a learning unit 326. These components are realized, for example, by a processor such as a CPU or GPU executing a program stored in the storage unit 330. Some or all of the components of the control unit 320 may be realized by hardware such as LSI, ASIC, or FPGA, or may be realized by the cooperation of software and hardware.

The storage unit 330 is realized by, for example, an HDD, a flash memory, an EEPROM, a ROM, a RAM, or the like. Various programs such as firmware and application programs are stored in the storage unit 330. In addition to the program referred to by the processor, the storage unit 330 stores model data 332 and the like. The model data 332 is data (program or data structure) that defines the parameter estimation model MDL described later.

[Processing flow of information processing device]
Hereinafter, a series of processing flows of the server 300 will be described according to the flowchart. FIG. 8 is a flowchart showing an example of a series of processing flows of the server 300 according to the embodiment. The processing of this flowchart may be performed in conjunction with, for example, the number of walking trainings of the user. Specifically, when the user wears the walking assist device 100 and performs walking training once a day, the processing of this flowchart is performed on a daily basis. Further, when the server 300 is implemented by a plurality of computers included in a distributed computing system or a cloud computing system, a part or all of the processing of this flowchart may be processed in parallel by the plurality of computers. Hereinafter, as an example, the walking training this time will be described assuming that it is the nth time (n is an arbitrary natural number).

First, the acquisition unit 322 waits until the nth walking training is started (step S100), and when the nth walking training is started, the terminal device 200 sends the second walking training via the communication unit 302. Acquire the nth questionnaire response data (step S102).

The questionnaire response data is, for example, data representing the result of an actual response by the user to a questionnaire (also referred to as a survey) asked by the user before starting walking training. For example, suppose a user launches a walking training application on a terminal device 200 to start walking training. In this case, for example, the control unit 220 of the terminal device 200 causes the touch panel 204 to display the questionnaire when the walking training application is activated. The questions in the questionnaire include questions about the walking assist device 100 and questions about the user's own physical condition and health that specialists such as doctors and physiotherapists ask patients. When the user inputs the answer to the questionnaire on the touch panel 204, the control unit 220 transmits the answer result as the questionnaire answer data to the server 300 via the communication unit 202. The questionnaire response data is an example of "third data".

Next, the determination unit 324 receives the nth questionnaire response data acquired by the acquisition unit 322 in the process of S102, and the walking training measurement data and parameter selection data acquired during the previous n-1th walking training. Is input to the trained (trained) parameter estimation model MDL defined by the model data 332 (step S104).

The walking training measurement data is data representing the measurement results of various training parameters measured by the walking assist device 100. The parameter selection data is data representing the result of a parameter actually selected by the user from a plurality of training parameters that serve as a guideline for walking training. The walking training measurement data is an example of "first data", and the parameter selection data is an example of "second data".

The parameter estimation model MDL is the walking training measurement data and parameter selection data of the user who performed the walking training in the n-1st time (that is, the previous time), and the questionnaire of the user who plans to perform the walking training in the nth time (that is, this time). When the answer data is input, the model is trained to estimate the training parameters to be selected by the user when performing the nth walking training, and output the estimation results as recommended parameters. The learning method of the parameter estimation model MDL will be described later.

FIG. 9 is a diagram showing an example of the parameter estimation model MDL. As shown, the parameter estimation model MDL may be implemented by a deep neural network composed of an input layer, a plurality of hidden layers (intermediate layers), and an output layer. The deep neural network may be an RNN (Recurrent Neural Network) in which a part of the hidden layer is replaced with an LSTM (Long short-term memory).

When the parameter estimation model MDL is implemented by a neural network, the model data 332 contains, for example, connection information on how units contained in each of a plurality of layers constituting the neural network are connected to each other, and connection information. It contains various information such as the coupling coefficient given to the data input / output between the units.

The connection information includes, for example, information such as the number of units included in each layer, information for specifying the type of unit to which each unit is connected, the activation function of each unit, and the gate provided between the units of the hidden layer. include. The activation function may be, for example, a rectified linear function (ReLU function), a sigmoid function, a step function, or other function. The gate selectively passes or weights the data transmitted between the units, for example, depending on the value returned by the activation function (eg 1 or 0). The coupling coefficient includes, for example, a weight given to the output data when data is output from a unit of a certain layer to a unit of a deeper layer in a hidden layer of a neural network. Further, the coupling coefficient may include a bias component peculiar to each layer.

Returning to the description of the flowchart of FIG. Next, the determination unit 324 determines a recommended parameter to be selected by the user during the nth (that is, this time) walking training based on the output result of the parameter estimation model MDL (step S106).

Next, the determination unit 324 transmits the determined recommended parameter to the user's terminal device 200 via the nth walking training via the communication unit 302 (step S108). In response to this, the control unit 220 of the terminal device 200 displays the content on the touch panel 204 with the recommended parameters selected in advance before the user selects the content. This allows the user to know the walking training menu to be tackled this time without the guidance of an expert. At this time, the user does not necessarily have to select the recommended parameters, but can select any other training parameter.

Next, the acquisition unit 322 waits until the nth walking training is completed (step S110), and when the nth walking training is completed, the nth time from the terminal device 200 via the communication unit 302. (That is, this time), the walking training measurement data and the parameter selection data of the user who performed the walking training are acquired (step S112). The nth walking training measurement data and parameter selection data acquired in the process of S112 are used in the process of S104 in the next cycle.

Next, the learning unit 326 receives the n-1st walking training measurement data and parameter selection data acquired in the processing of S112 in the previous cycle, and the nth time acquired in the processing of S102 in the current cycle. Teacher data is generated based on the questionnaire response data and the nth parameter selection data acquired in the process of S112 of this cycle (step S114).

The teacher data is a data set in which the correct output data to be output by the machine learning model is associated with a certain input data as a teacher label (also referred to as a target). For example, the learning unit 326 includes input data including the n-1st walking training measurement data, parameter selection data, and nth questionnaire response data, and output data including the nth parameter selection data. A data set that combines the above is generated as teacher data.

Next, the learning unit 326 learns the parameter estimation model MDL based on the generated teacher data (step S116).

FIG. 10 shows an example of a database that combines walking training measurement data, parameter selection data, and questionnaire response data. For example, when n is 3, the learning unit 326 inputs the walking training measurement data and the parameter selection data for the second walking training and the questionnaire response data for the third walking training as the teacher data input data. Then, the parameter selection data for which the number of walking trainings is the third is used as the output data of the teacher data.

Then, the learning unit 326 inputs the combination of the second walking training measurement data and the parameter selection data and the third questionnaire response data into the parameter estimation model MDL as input data of the teacher data. When these data are input, the parameter estimation model MDL outputs the training parameters that the user should select during the third walking training as recommended parameters.

The learning unit 326 learns the parameter estimation model MDL so that the third recommended parameter output by the parameter estimation model MDL and the parameter represented by the parameter selection data obtained during the third walking training match. In other words, the learning unit 326 learns the parameter estimation model MDL so that the third recommended parameter output by the parameter estimation model MDL approaches the parameter actually selected by the user during the third walking training. For example, the learning unit 326 obtains the difference between the parameters, and learns the parameter estimation model MDL by using a stochastic gradient descent method or the like so that the difference becomes small.

When the learning unit 326 learns the parameter estimation model MDL as described above, the learning unit 326 updates the model data 332 stored in the storage unit 330. This completes the processing of this flowchart.

According to the embodiment described above, the server 300 performs the nth walking training from the walking assisting device 100 that measures the movement when the user walks while assisting the walking of the user via the terminal device 200. Acquire the walking training measurement data at the time. The server 300 acquires the parameter selection data at the time of the nth walking training from the terminal device 200. Further, the server 300 acquires the questionnaire response data at the time of the n + 1th walking training from the terminal device 200. When the server 300 acquires the nth walking training measurement data and the parameter selection data and the n + 1th questionnaire response data, the server 300 inputs these data into the parameter estimation model MDL and outputs the parameter estimation model MDL as the output result. Based on this, the recommended parameters that should be used as guidelines for the user's n + 1th walking training are determined. In this way, by using machine learning, it is possible to appropriately determine a guideline for walking training of a user without the guidance of a specialist such as a doctor or a physiotherapist.

<Modified example of the embodiment>
Hereinafter, a modified example of the above-described embodiment will be described. In the above-described embodiment, the parameter estimation model MDL has been described as being implemented by a deep neural network, but the present invention is not limited to this. For example, the parameter estimation model MDL may be implemented by a combination of a topic model and a Bayesian network.

FIG. 11 is a diagram for explaining a topic model, and FIG. 12 is a diagram showing an example of a Bayesian network. As shown in FIG. 11, the topic model regarded a plurality of walking training measurement data, a plurality of parameter selection data, and a plurality of questionnaire response data collected each time the walking training was repeated as one matrix. Occasionally, techniques such as LDA (Latent Dirichlet Allocation) and PLSA are used to classify each element of the matrix into one of a plurality of topics (clusters) such as T1, T2, and T3.

As illustrated in FIG. 12, each topic classified by the topic model is input to the Bayesian network. The Bayesian network outputs estimated parameters by stochastic inference from the input topic. The determination unit 324 may determine recommended parameters by using such a topic model in combination with a Bayesian network.

Further, in the above-described embodiment, the learning unit 326 selects the nth recommended parameter from the input data including the n-1st walking training measurement data and the parameter selection data and the nth questionnaire response data. Although it has been described as learning the parameter estimation model MDL so as to output it, the present invention is not limited to this.

For example, the learning unit 326 is in the nth time from the input data including the nkth walking training measurement data and the parameter selection data before the n-1th time and the nth questionnaire response data. The parameter estimation model MDL may be trained to output the recommended parameters. k is a natural number greater than or equal to 1 and less than n. That is, the nkth time may include the n-1th time.

In addition, the learning unit 326 is the nth time from not only the nkth time but also the nth time from the total of j walking training measurement data and the parameter selection data going back from the nkth time to the nkth time. The parameter estimation model MDL may be trained to output the recommended parameters. j is any natural number less than n. For example, when n is 10, k is 1, and j is 4, the learning unit 326 is the tenth from the walking training measurement data and parameter selection data for a total of five times from the fifth to the ninth walking training. The parameter estimation model MDL is trained to output the recommended parameters for the second time.

The embodiment described above can be expressed as follows.
At least one memory that stores the program and
With at least one processor,
When the processor executes the program,
From the walking assist device that measures the movement when the user walks while assisting the user's walking, the first data representing the measurement result of the user's movement at the time of the nth walking training is acquired.
From the terminal device available to the user, second data representing the selection result of at least one parameter selected by the user from among a plurality of parameters that serve as a guideline for the nth walking training is acquired.
When at least the first data and the second data of the user who has performed the mth walking training are input, the recommended parameters to be used as a guideline at the m + kth walking training of the user are output. The first data and the second data at the time of the nth walking training of the acquired user were input to the model learned in the above, and the first data and the second data were input. Based on the output result of the model, the recommended parameter to be used as a guideline for the n + kth walking training of the user is determined.
An information processing device that is configured as such.

Although the embodiments for carrying out the present invention have been described above using the embodiments, the present invention is not limited to these embodiments, and various modifications and substitutions are made without departing from the gist of the present invention. Can be added.

1 ... Walking assistance system, 100 ... Walking assistance device, 102 ... Mainframe, 103 ... Subframe, 104 ... Drive source, 105 ... Control device, 106 ... Hip joint angle sensor, 107 ... Battery, 200 ... Terminal device, 202 ... Communication Unit, 204 ... Touch panel, 220 ... Control unit, 230 ... Storage unit, 300 ... Server, 302 ... Communication unit, 320 ... Control unit, 322 ... Acquisition unit, 324 ... Decision unit, 326 ... Learning unit, 330 ... Storage unit

Claims (6)

From the walking assist device that measures the movement when the user walks while assisting the user's walking, the first data representing the measurement result of the user's movement at the time of the nth walking training is acquired, and the user Acquires second data representing the selection result of at least one parameter selected by the user from a plurality of parameters that serve as a guideline for the nth walking training from a terminal device that can be used by the user.
When at least the first data and the second data of the user who has performed the mth walking training are input, the recommended parameters to be used as a guideline at the m + kth walking training of the user are output. The first data and the second data at the time of the nth walking training of the user acquired by the acquisition unit are input to the model learned in the above, and the first data and the second data are input. Based on the output result of the model in which is input, a determination unit for determining the recommended parameters to be used as a guideline for the n + kth walking training of the user, and a determination unit.
Information processing device equipped with. The acquisition unit further acquires the third data representing the response result of the user to the n + kth walking training questionnaire from the terminal device.
The determination unit inputs the first data and the second data of the user at the time of the nth walking training and the third data of the user at the time of the n + kth walking training into the model. Based on the output result of the model to which the first data, the second data, and the third data are input, the recommended parameter to be used as a guideline for the n + kth walking training of the user is determined.
The information processing device according to claim 1. The model is one of a deep neural network, a topic model, and a Bayesian network, or a combination of a plurality of models.
The information processing device according to claim 1 or 2. The recommended parameters at the time of the mth walking training of the user output by the model when the first data and the second data of the user at the mkth time are input to the model, and the mth time. A learning unit for learning the model is further provided so that the parameter selected by the user from the plurality of parameters matches during the walking training.
The information processing device according to any one of claims 1 to 3. The computer
From the walking assist device that measures the movement when the user walks while assisting the user's walking, the first data representing the measurement result of the user's movement at the time of the nth walking training is acquired.
From the terminal device available to the user, second data representing the selection result of at least one parameter selected by the user from among a plurality of parameters that serve as a guideline for the nth walking training is acquired.
When at least the first data and the second data of the user who has performed the mth walking training are input, the recommended parameters to be used as a guideline at the m + kth walking training of the user are output. The first data and the second data at the time of the nth walking training of the acquired user were input to the model learned in the above, and the first data and the second data were input. Based on the output result of the model, the recommended parameter to be used as a guideline for the n + kth walking training of the user is determined.
Information processing method. On the computer
Acquiring the first data representing the measurement result of the user's movement at the time of the nth walking training from the walking assist device that measures the movement when the user walks while assisting the user's walking.
Obtaining second data representing the selection result of at least one parameter selected by the user from a plurality of parameters that serve as a guideline for the nth walking training from the terminal device available to the user.
When at least the first data and the second data of the user who has performed the mth walking training are input, the recommended parameters to be used as a guideline at the m + kth walking training of the user are output. The first data and the second data at the time of the nth walking training of the acquired user were input to the model learned in the above, and the first data and the second data were input. Based on the output result of the model, the recommended parameter to be used as a guideline for the n + kth walking training of the user is determined.
A program to execute.

Images