JP6798068B2 - Function improvement support device and function improvement support system - Google Patents

Description

The present invention relates to a function improvement support device, a function improvement support system, and a terminal device that support the improvement of the function of a target user.

In recent years, the aging of the population has progressed rapidly, and the number of people requiring long-term care, including the elderly, is increasing. For this reason, it is necessary to take measures to curb the increase in the number of people requiring care or reduce the number of people requiring care.

As an example of the above measures, for example, Patent Document 1 discloses a dementia care support system that improves cognitive function by having a person requiring long-term care for dementia play a game.

In addition, Patent Document 2 discloses a user health maintenance activation support and monitoring system that improves motor function by providing an exercise program to a person requiring long-term care and having them perform actions according to the program. There is.

Japanese Unexamined Patent Publication No. 2016-071897 JP-A-2010-140119

However, the problems of care recipients are diverse, and even if problems are found, it is difficult to select the most effective program from a large number of exercise programs unless you are a rehabilitation profession. In other words, it is extremely difficult for ordinary care workers to choose such a program.

Even in the systems described in Patent Documents 1 and 2, it is difficult to select the optimum game or exercise program, and the same problem arises.

The present invention has been made in view of such circumstances, and is a function improvement support device and function improvement support capable of determining an exercise program for effectively improving the functions of a target user such as a care recipient. It is an object of the present invention to provide a system and a terminal device.

(1) In order to achieve the above object, the function improvement support device according to a certain aspect of the present invention acquires the feature amount related to the function of the target user based on the acquired unit and the learned model. The trained model includes an exercise program determination unit that determines an exercise program for improving the function of the target user from the feature amount, and the trained model includes the feature amount related to the function for a plurality of users and each user. This is a model in which the executed exercise program is machine-learned as learning data, and the identification information of the exercise program corresponding to the feature amount is output.

According to this configuration, the exercise program for improving the function of the target user can be determined by inputting the feature amount related to the function of the target user into the trained model. This trained model is a model obtained by machine learning the feature quantities for a plurality of users and the exercise program executed by each user as learning data. Therefore, by inputting the feature amount of the target user into the trained model, the exercise program expected to have the highest degree of improvement for the target user is output. This makes it possible to determine an exercise program for effectively improving the functions of the target user such as a care recipient.

(2) Preferably, the function is a function related to the exercise or life of the user, and the acquisition unit acquires a distance image showing the distance to the target user as the feature amount and determines the exercise program. Based on the trained model, the unit determines a problem related to the exercise or life of the user and an exercise program for improving the problem from the acquired feature amount, and the trained model determines the problem. , The problem corresponding to the feature amount by machine learning the distance image of each user, the problem related to the exercise or life of each user, and the exercise program executed by each user as learning data for a plurality of users. This is a model that outputs identification information of points and the exercise program.

According to this configuration, by inputting the distance image of the target user into the trained model, the problem of the target user can be extracted and the exercise program for improving the problem can be determined. For example, even if the target user cannot get up, there is a difference in the degree to which the target user cannot get up. However, in this configuration, the same exercise program is not provided to a plurality of target users who cannot get up, but the most effective exercise program is determined and provided according to the degree of inability to get up for each target user. can do.

(3) Further, the function is a function related to the user's language or oral cavity, and the acquisition unit receives an answer to the target user's language or oral question item and examination item, and the target user's language or The evaluation by the oral specialist for the evaluation item of the oral cavity is acquired as the feature quantity, and the learned model is obtained by answering the question item and the inspection item of the language or the oral cavity of each user for a plurality of users and by the oral specialist. The evaluation of each user's evaluation item, and the management guidance item and exercise program executed by each user are machine-learned as learning data, and the identification information of the management guidance item and the identification information of the exercise program corresponding to the feature amount are obtained. It is a model to be output, and the exercise program determination unit may determine management guidance items and exercise programs for improving the function of the target user from the acquired feature amount based on the learned model. ..

According to this configuration, answers to the target user's language or oral questions and examination items and the evaluation of the target user by the oral profession are acquired, and the acquired answers and evaluations are input to the trained model. It is possible to determine management guidance items and exercise programs for improving the language or oral cavity-related functions of the target user. Thereby, for example, instead of providing the same management instruction item and exercise program to a plurality of target users with language disorders, the most effective management instruction item and exercise program are determined for each target user. Can be provided.

(4) Further, the above-mentioned function improvement support device further includes a machine learning unit that machine-learns the learned model, and the machine learning unit further executes the exercise program determined by the exercise program determination unit. The trained model may be machine-learned using the feature amount of the target user and the identification information of the exercise program as learning data.

According to this configuration, the trained model can be relearned by using the feature amount of the target user and the exercise program provided to the target user as learning data. As a result, it is possible to determine and provide an effective exercise program with a higher degree of improvement to other target users.

(5) The function improvement support system according to another aspect of the present invention includes a distance image sensor that acquires a distance image indicating the distance to the target user, the function improvement support device according to (2), and the function improvement support. From the device, a problem related to the exercise or life of the target user and an exercise program for improving the function of the target user are acquired, and the acquired problem and a terminal device for presenting the exercise program are provided.

According to this configuration, the terminal device can present the problems and the exercise program determined by the above-mentioned function improvement support device. Therefore, for example, the target user who uses the terminal device can execute the training for functional improvement based on the most effective exercise program after knowing the problem of the target user. In addition, the caregiver of the target user can make the target user perform training for functional improvement based on the most effective exercise program, knowing the problems of the target user.

(6) The function improvement support system according to another aspect of the present invention is a management guidance item and exercise for improving the function of the target user from the function improvement support device described in (3) and the function improvement support device. It is provided with a terminal device that acquires the program and presents the acquired management guidance item and the exercise program.

According to this configuration, the exercise program determined by the above-mentioned function improvement support device can be presented to the terminal device. Therefore, for example, the target user who uses the terminal device can execute training for functional improvement based on the most effective management guidance items and exercise programs. In addition, the caregiver of the target user can have the target user perform training for functional improvement based on the most effective management guidance items and exercise programs.

The terminal device according to another aspect of the present invention includes an exercise program acquisition unit that acquires an exercise program for improving the function of the target user from the above-mentioned function improvement support device, and the exercise acquired by the exercise program acquisition unit. It is equipped with an exercise program presentation unit that presents the program.

According to this configuration, it is possible to present the exercise program determined by the above-mentioned function improvement support device. Therefore, for example, the target user or the caregiver of the target user who uses the terminal device can execute training for functional improvement based on the most effective exercise program.

According to the present invention, the functions of a target user such as a care recipient can be effectively improved.

It is a figure which shows the structure of the function improvement support system which concerns on Embodiment 1 of this invention. It is a block diagram which shows the structure of the function improvement support apparatus which concerns on Embodiment 1 of this invention. It is a figure which shows an example of the user information stored in the storage part. It is a block diagram which shows the structure of the terminal apparatus which concerns on Embodiment 1 of this invention. It is a flowchart which shows an example of the processing procedure of the function improvement support apparatus which concerns on Embodiment 1 of this invention. It is a flowchart which shows an example of the processing procedure of the function improvement support apparatus which concerns on Embodiment 1 of this invention. It is a figure which shows the other configuration of the function improvement support system. It is a block diagram which shows the structure of the function improvement support apparatus which concerns on Embodiment 2. It is a figure which shows an example of the questionnaire including the question item, the examination item and the answer item for a language or an oral cavity. It is a figure which shows an example of the evaluation report which includes the evaluation item about language or the oral cavity and the evaluation of the oral profession for the said item. It is a figure which shows an example of the user information stored in the storage part. It is a flowchart which shows an example of the processing procedure of the function improvement support apparatus which concerns on Embodiment 2 of this invention. It is a figure which shows an example of the management guidance plan of the oral function improvement service created based on the management guidance item and the content of the exercise program transmitted to the terminal device.

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. In addition, at least a part of the embodiments described below may be arbitrarily combined.

[Embodiment 1]
<Structure of function improvement support system>
FIG. 1 is a diagram showing a configuration of a function improvement support system according to the first embodiment of the present invention.

The function improvement support system 1 is a system for improving the functions of the target user 50 such as a person requiring long-term care, and includes a distance image sensor 10, a function improvement support device 20, and a terminal device 40.

The distance image sensor 10 measures the distances to a plurality of points of the subject to generate a distance image in which the brightness value of each pixel indicates the distance to the subject. That is, the distance image sensor 10 is a camera including a plurality of LEDs (Light Emitting Diodes) regularly arranged on the front surface of the camera and a lens, and the light emitted from the plurality of LEDs all at once is a subject. The reflected and reflected light (eg, near-infrared light) is focused by the lens. At this time, the time from when the LED emits light until the reflected light reaches the lens varies depending on the position of the subject. That is, the closer the subject is to the camera, the shorter the arrival time of the reflected light, and the farther the subject is from the camera, the longer the arrival time of the reflected light. The distance image sensor 10 measures the arrival time for each pixel and outputs the distance to the subject for each pixel. That is, the distance image sensor 10 outputs a distance image in which the distance to the subject is represented by the brightness value of the pixels.

The function improvement support device 20 acquires a distance image from the distance image sensor 10. The function improvement support device 20 determines a problem regarding the function of the target user 50 and an exercise program for improving the problem from the distance image based on the trained model that has been machine-learned in advance.

For example, the function to be improved is a function related to the exercise or life of the target user 50.

The trained model is machine-learned using distance images for multiple users, problems related to the functions of each user, and exercise programs executed by each user as learning data, and identifies problems and exercise programs corresponding to the distance images. This is a model that outputs information (exercise program name).

That is, the trained model is machine-learned for a plurality of users using the distance image of each user, the problem of each user, the attributes of each user, and the exercise program executed by each user as learning data, and the target user 50. This is a model that outputs the problem of the target user 50 corresponding to the distance image and the attribute of the target user 50 and the identification information (exercise program name) of the exercise program.

The model can be created by, for example, deep learning. Here, the user's distance image and the user's attributes are input by performing deep learning with a teacher using the learning data in which the user's distance image is input and the user's problem and the exercise program name are output. As, you can create a model that outputs the problem and the exercise program name. By performing deep learning, the model can be represented by a neural network.

The function improvement support device 20 transmits the determined problem and the exercise program to the terminal device 40 via the network 2.

The terminal device 40 is, for example, a computer device including a display screen of a tablet terminal, a smartphone, a notebook computer, or the like. The terminal device 40 is operated by, for example, a caregiver, a caregiver, a care worker, or the like who assists the function improvement of the target user 50. The target user 50 itself may operate the terminal device 40.

The terminal device 40 receives the problem and the exercise program from the function improvement support device 20. The terminal device 40 displays the received exercise program on the display screen. As a result, the caregiver or the like informs the target user 50 of the problem and causes the target user 50 to implement an exercise program for improving the problem, or the target user 50 knows the problem by himself / herself and implements an exercise program for improving the problem. Can be carried out.

<Configuration of function improvement support device>
FIG. 2 is a block diagram showing a configuration of a function improvement support device according to the first embodiment of the present invention.

The function improvement support device 20 includes a distance image acquisition unit 21, an exercise program determination unit 23, a storage unit 24, a learning data acquisition unit 26, a machine learning unit 27, and a user attribute acquisition unit 28.

The function improvement support device 20 can be realized by a general computer including a CPU (Central Processing Unit), a ROM (Read Only Memory), a RAM (Random Access Memory), an HDD (Hard Disk Drive), a communication interface, and the like. .. For example, the processing units 21, 23 and 26 to 28 of the function improvement support device 20 are functionally realized by loading the computer program stored in the HDD or ROM into the RAM and executing it on the CPU. .. However, a part or all of the processing units 21, 23 and 26 to 28 may be realized by hardware such as a semiconductor device.

The distance image acquisition unit 21 acquires a distance image output from the distance image sensor 10 as a feature amount of the target user 50.

The distance image acquisition unit 21 may perform a labeling process for labeling the three-dimensional coordinates within a predetermined distance range with the same label based on the three-dimensional coordinates indicated by the distance image. The distance image acquisition unit 21 may extract the area of the target user 50 from the result of the labeling process. For example, the distance image acquisition unit 21 may extract the largest area existing in the area as the area of the target user 50 by designating the area where the target user 50 exists in advance.

The distance image acquisition unit 21 stores the acquired distance image in the storage unit 24 as user information 25A in association with the user ID of the target user 50.

The user attribute acquisition unit 28 acquires, for example, the input attribute of the target user 50 by operating the keyboard, the terminal device 40, or the like, and additionally registers it in the user information 25A stored in the storage unit 24. The attributes include, for example, information such as the gender, age, weight and height of the target user 50.

The exercise program determination unit 23 is based on the learned model 25B stored in the storage unit 24, and is based on the distance image acquired by the distance image acquisition unit 21 and the attributes of the target user 50 acquired by the user attribute acquisition unit 28. , The problem related to the function of the target user 50 and the exercise program for improving the problem are determined. The exercise program determination unit 23 additionally registers the determined problem and the exercise program name, which is the identification information of the exercise program, in the user information 25A stored in the storage unit 24.

The problems determined by the exercise program determination unit 23 include, for example, problem 1 indicating a major classification of problems such as being unable to turn over, being unable to stand up, and being unable to crouch, and more detailed problems. There is a problem 2 shown. For example, problem 2 for problem 1 "cannot turn over" includes "neck not raised", "arm not raised", "shoulder not raised", "body does not twist", etc. Is done.

The exercise program is prepared for each problem 2, for example, a neck flexion exercise for a user whose neck is not raised and a neck flexion exercise for a user whose shoulder is not raised, and an abdomen for a user whose shoulder is not raised. Includes various exercise programs such as reach exercises to move the shoulder blades and standing exercises to extend the elbows and tilt the abdomen sideways for users who do not twist their bodies.

The exercise program determination unit 23 transmits the determined problem and the content or the exercise program name of the determined exercise program to the terminal device 40 and registers the determined problem in the user information 25A.

The storage unit 24 is composed of an HDD, a flash memory, or the like, and stores the user information 25A and the learned model 25B.

FIG. 3 is a diagram showing an example of user information stored in the storage unit.
The user information 25A is stored in association with a user ID, gender, age, weight, height, distance image, problem 1, problem 2, and an exercise program name.

For example, the gender of the target user 50 of the user ID "u1" is "male", the age is "81", the weight is "72" kg, the height is "168" cm, the distance image is "img1", and the target user 50. Problem 1 is the inability to "turn over", and problem 2 of the target user 50 is "the neck is not raised" and "the left arm is not raised", and problem 2 of the target user 50. The name of the exercise program decided to improve the above is "p1". Similar information is stored in the storage unit 24 for the target user 50 such as the user IDs “u2” and “u3”.

The learning data acquisition unit 26 acquires learning data for re-learning the trained model 25B. Specifically, the learning data acquisition unit 26 reads the user information 25A from the storage unit 24.

For example, the learning data acquisition unit 26 acquires user information about the target user 50 of the user ID “u1” shown in FIG. 3 as learning data. The learning data acquisition unit 26 similarly acquires user information as learning data for other target users 50 such as user IDs “u2” and “u3”.

The machine learning unit 27 updates the learned model 25B by further machine learning the learned model 25B based on the learning data acquired by the learning data acquisition unit 26. As described above, deep learning can be used as an example of machine learning.
The machine learning unit 27 stores the updated learned model 25B in the storage unit 24.

<Configuration of terminal device>
FIG. 4 is a block diagram showing a configuration of a terminal device according to a first embodiment of the present invention.
The terminal device 40 includes an exercise program acquisition unit 41 and an exercise program presentation unit 42.

The exercise program acquisition unit 41 acquires problems and exercise programs from the function improvement support device 20 via the network 2. When the content of the exercise program is stored in its own storage unit, the exercise program acquisition unit 41 acquires the exercise program name from the function improvement support device 20 and corresponds to the exercise program name from the storage unit. The exercise program may be acquired by reading the exercise program.

The exercise program presentation unit 42 presents the problem acquired by the exercise program acquisition unit 41 and the exercise program to the target user 50 or the caregiver of the target user 50 by displaying the exercise program on the display screen.

<Operation mode>
FIG. 5 is a flowchart showing an example of a processing procedure of the function improvement support device according to the first embodiment of the present invention. FIG. 5 shows a processing procedure of the function improvement support device 20 in the operation mode in which the exercise program of the target user 50 is determined from the distance image captured by the target user 50.

The distance image acquisition unit 21 acquires a distance image output from the distance image sensor 10 (S1).

The exercise program determination unit 23 inputs the distance image acquired by the distance image acquisition unit 21 and the attributes (gender, age, weight, height) of the target user 50 into the trained model 25B, and has the highest probability (functional improvement). (Probability of) problem and exercise program name are determined (S2).

The exercise program determination unit 23 transmits the determined problem and the content of the exercise program to the terminal device 40 (S3). The exercise program determination unit 23 may transmit the exercise program name to the terminal device 40 instead of the content of the exercise program.

As a result, the terminal device 40 can receive and display the problems and the contents of the exercise program transmitted from the function improvement support device 20. Therefore, the caregiver who uses the terminal device 40 can inform the target user 50 of the problem of the target user 50 and execute an exercise program that is most effective for improving the function.

<Learning mode>
FIG. 6 is a flowchart showing an example of a processing procedure of the function improvement support device according to the first embodiment of the present invention. FIG. 6 shows a processing procedure of the function improvement support device 20 in the learning mode in which the trained model for determining the exercise program is relearned.

The learning data acquisition unit 26 acquires the user information 25A by reading it from the storage unit 24 (S11).

The machine learning unit 27 uses the user information 25A acquired by the learning data acquisition unit 26 as learning data, and machine-learns the learned model 25B stored in the storage unit 24 again (S13). In the example shown in FIG. 3, the machine learning unit 27 machine-learns the trained model 25B using each user's gender, age, weight, height, distance image, problem 1, problem 2, and exercise program name as learning data. To do. The machine learning unit 27 overwrites the learned model 25B after the opportunity learning on the storage unit 24.
The learning mode processing is performed, for example, every predetermined period (for example, one month).

<Effect of Embodiment 1>
As described above, according to the first embodiment of the present invention, an exercise program for improving the function of the target user 50 by inputting the feature amount related to the function of the target user 50 into the trained model 25B. Can be determined. The trained model 25B is a model obtained by machine learning the feature quantities for a plurality of users, the problems of each user, and the exercise program executed by each user as learning data. Therefore, by inputting the feature amount and the problem of the target user 50 into the trained model 25B, the exercise program expected to have the highest degree of improvement for the target user 50 is output. Thereby, it is possible to determine an exercise program for effectively improving the function of the target user 50 such as a person requiring nursing care. In addition, conventionally, a rehabilitation profession has measured motion elements such as the flexion angle of the neck and the rotation angle in the body axis of the target user 50, and determines the exercise program based on the measurement results. According to this embodiment, such measurement work is also unnecessary.

More specifically, the problem can be extracted based on the distance image, and an exercise program for improving the problem can be determined. For example, even if the target user 50 cannot get up, there is a difference in the degree to which the target user 50 cannot get up. However, in this configuration, the same exercise program is not provided to a plurality of target users 50 who cannot get up, but the most effective exercise program is determined for each target user 50 according to the degree of being unable to get up. , Can be provided.

Further, the trained model 25B can be relearned by using the distance image and the problem of the target user 50 and the exercise program provided to the target user 50 as learning data. As a result, it is possible to determine and provide an effective exercise program with a higher degree of improvement to the other target user 50.

[Modification 1 of the first embodiment]
In the above-described first embodiment, the model is learned by supervised machine learning, but it is also possible to learn the model by reinforcement learning.

For example, by performing reinforcement learning with the user's distance image and user's attributes as the state in reinforcement learning, the user's problem and exercise program name as the action in reinforcement learning, and the degree of improvement as the reward in reinforcement learning, the user It is possible to create a model that outputs a problem and an exercise program name that maximizes the degree of improvement by inputting a distance image and user attributes. Reinforcement learning can also be referred to as deep reinforcement learning, and the model can be expressed by a neural network by performing deep reinforcement learning.

The learning data acquisition unit 26 of the function improvement support device 20 shown in FIG. 2 reads the user information 25A from the storage unit 24. In addition, the learning data acquisition unit 26 acquires the degree of improvement of the function after executing the exercise program with the exercise program name shown in the user information 25A. The learning data acquisition unit 26 receives, for example, the degree of improvement input by the caregiver by operating the keyboard or the terminal device 40. The degree of improvement is quantified in five stages from 1 to 5, for example, 1 is the highest degree of improvement, the degree of improvement is lower as the numerical value is larger, and 5 is the lowest degree of improvement.

For example, the learning data acquisition unit 26 executes the user information about the target user 50 of the user ID “u1” shown in FIG. 3 and the exercise program name “p1” shown in the user information, and then the target user 50. The degree of improvement of the function of is acquired as learning data. Similarly, the learning data acquisition unit 26 acquires user information and the degree of improvement of the function as learning data for the other target users 50 such as the user IDs “u2” and “u3”.

The machine learning unit 27 machine-learns the learned model 25B stored in the storage unit 24 again, using the user information 25A acquired by the learning data acquisition unit 26 and the degree of improvement as learning data. In the example shown in FIG. 3, the machine learning unit 27 learns the gender, age, weight, height, distance image, problem 1, problem 2, exercise program name of each user, and the degree of improvement of the function of the user. Machine learning of the trained model 25B as data for use. The machine learning unit 27 overwrites the learned model 25B after the opportunity learning on the storage unit 24.

[Modification 2 of the first embodiment]
Further, the function of the distance image sensor 10 may be provided in the terminal device 40, or the distance image sensor 10 is prepared as an attachment of the terminal device 40 and is used by connecting the distance image sensor 10 to the terminal device 40. You may.

FIG. 7 is a diagram showing another configuration of the function improvement support system.
The function improvement support system 1 is a system for improving the functions of the target user 50 such as a person requiring long-term care, and includes a distance image sensor 10, a function improvement support device 20, and a terminal device 40. The distance image sensor 10 is, for example, miniaturized and configured to be connectable to the terminal device 40.

The terminal device 40 executes the same processing as that of the above-described embodiment. In addition, the terminal device 40 transmits the distance image generated by the distance image sensor 10 to the function improvement support device 20 via the network 2.

The function improvement support device 20 performs the same processing as that of the above-described embodiment based on the distance image received from the terminal device 40.

[Embodiment 2]
In the first embodiment, improvement of the function related to the user's exercise or life has been described. However, the functions to be improved are not limited to this. For example, a function related to the user's language or oral cavity may be a function to be improved.

In the second embodiment, improvement of the function related to the user's language or oral cavity will be described. In the following, the points different from those of the first embodiment will be mainly described, and detailed description of the same points will not be repeated.

The function improvement support system according to the second embodiment does not include the distance image sensor 10 in the function improvement support system 1 shown in FIG.

<Configuration of function improvement support device>
FIG. 8 is a block diagram showing a configuration of the function improvement support device according to the second embodiment.
In the configuration of the function improvement support device 20 according to the first embodiment shown in FIG. 2, the function improvement support device 20 includes an answer acquisition unit 29 and an evaluation acquisition unit 30 instead of the distance image acquisition unit 21. The answer acquisition unit 29 and the evaluation acquisition unit 30 are functionally realized by executing a computer program on the CPU, similarly to the distance image acquisition unit 21. However, the answer acquisition unit 29 and the evaluation acquisition unit 30 may be realized by hardware such as a semiconductor device.

The answer acquisition unit 29 acquires the answers to the language or oral question items and the examination items of the target user 50 as feature quantities related to the functions of the target user 50.

FIG. 9 is a diagram showing an example of a questionnaire including question items, test items, and answer items for language or oral cavity. The questionnaire shows question items, test items, and answer options. For example, item IDs "Q1" to "Q10" indicate a set of question item and answer options, and item ID "Q11" and subsequent items indicate a set of inspection items and answer options.

Here, the question item is an item that the target user 50 himself / herself mainly answers subjectively. However, if it is difficult for the target user 50 to answer, a caregiver or the like may answer. Further, the inspection item is an item for objectively observing the target user 50, and is also called an observation item. Therefore, in principle, the inspection items are answered by the caregiver or the like, not by the target user 50 himself.

For example, "1. No" and "2. Yes" are prepared as answers to the question item "Is it difficult to eat hard food?" Of the item ID "Q1". In addition, as answers to the question item "How is your oral health?" Of item ID "Q10", "1. Good", "2. Somewhat good", "3. Normal", "4. Somewhat bad" and "5. Bad" is prepared.

In addition, "1. None", "2. Yes", and "3. Many" are prepared as answers to the inspection item "dirt on teeth and dentures" of the item ID "Q11". Further, for the inspection item "dirt on the tongue" of the item ID "Q12", "1. None", "2. Yes" and "3. Many" are prepared.

The answer acquisition unit 29 acquires the item ID of the questionnaire and the set of the answers as shown in FIG. 9 for each target user 50. The answer acquisition unit 29 acquires, for example, an item ID input by operating the keyboard or the terminal device 40 and a set of the answers from the keyboard or the terminal device 40, and associates the item ID with the user ID of the target user 50. , The user information 25A is stored in the storage unit 24.

The evaluation acquisition unit 30 acquires the evaluation by the oral profession for the language of the target user 50 or the evaluation item of the oral cavity as a feature amount related to the function of the target user 50.

FIG. 10 is a diagram showing an example of an evaluation report including an evaluation item relating to language or oral cavity and an evaluation by an oral profession for the item. The evaluation report shows evaluation items and evaluation options.

Here, the evaluation item is an item in which a speech therapist, a dental hygienist, or a nurse, who is an oral profession, evaluates while observing the target user 50 on a daily basis.

For example, "1. None", "2. Small amount", and "3. Large amount" are prepared as evaluations for the evaluation item "fluctuation" (evaluation item for whether or not drooling occurs) of the item ID "E1". In addition, as evaluation items for the evaluation item "neck flexion range of motion" (evaluation item of the range of motion of the neck up and down) of the item ID "E2", "1. No limit", "2. Move a little", "3. "Is prepared.

The evaluation acquisition unit 30 acquires the item ID of the evaluation report and the evaluation set thereof as shown in FIG. 10 for each target user 50. For example, the evaluation acquisition unit 30 acquires the item ID and the evaluation set thereof input by operating the keyboard or the terminal device 40 from the keyboard or the terminal device 40, and associates the item ID with the user ID of the target user 50. , The user information 25A is stored in the storage unit 24.

The exercise program determination unit 23 determines the item ID and the set of answers acquired by the answer acquisition unit 29 and the item ID and evaluation acquired by the evaluation acquisition unit 30 based on the learned model 25B stored in the storage unit 24. From the group, management guidance items and exercise programs for improving the functions of the target user 50 are determined. The exercise program determination unit 23 additionally registers the management instruction item name, which is the identification information of the determined management instruction item, and the exercise program name, which is the identification information of the exercise program, in the user information 25A stored in the storage unit 24. The management guidance items include various guidance items such as provision of information on improvement of oral function and support for oral cleaning. In addition, the exercise program includes various programs such as oral exercises, swallowing exercises, salivary gland massage, and functional training such as breathing and swallowing. The management guidance items determined by the exercise program determination unit 23 are a combination of various guidance items as described above, and the exercise program determined by the exercise program determination unit 23 is a combination of various programs as described above. It is a thing.

The exercise program determination unit 23 transmits the determined contents of the management guidance item and the content of the exercise program, or the management guidance item name and the exercise program name to the terminal device 40, and registers the contents in the user information 25A.

The trained model 25B stored in the storage unit 24 has, for a plurality of users, each user's item ID and answer set, item ID and evaluation set, user attributes, and management instruction items executed by each user. The exercise program is machine-learned as learning data, and the management guidance item name and exercise program name corresponding to the item ID and answer set of the target user 50, the item ID and evaluation set, and the attribute of the target user 50 are obtained. It is a model to output.

For example, the model can be created using deep learning. Here, learning data is used in which the feature quantity (item ID and answer set, item ID and evaluation set) and attributes related to the user's function are input, and the management guidance item name and exercise program name are output. The model can be created by deep learning with a teacher. By performing deep learning, the model can be represented by a neural network.

FIG. 11 is a diagram showing an example of user information stored in the storage unit.
The user information 25A is stored in association with a user ID, gender, age, weight, height, question item and inspection item item ID, answer, evaluation item item ID, evaluation, management guidance item name, and exercise program name. ing.

For example, the gender of the target user 50 of the user ID "u1" is "male", the age is "81", the weight is "72" kg, the height is "168" cm, and the answer of the item ID "Q1" is "1". , Item ID "Q2" is "3", Item ID "Q3" is "2", Item ID "E1" is "2", Item ID "E2" is "3", Item The evaluation of the ID "E3" is "1", the name of the management guidance item determined for improving the function of the target user 50 is "m1", and the exercise program determined for improving the function of the target user 50. The name of is "t1". Similar information is stored in the storage unit 24 for the target user 50 such as the user ID “u2”.

The learning data acquisition unit 26 acquires learning data for re-learning the trained model 25B. For example, the learning data acquisition unit 26 acquires user information about the target user 50 of the user ID “u1” shown in FIG. 9 as learning data. The learning data acquisition unit 26 similarly acquires user information as learning data for other target users 50 such as user IDs “u2” and “u3”.

The machine learning unit 27 updates the learned model 25B by further machine learning the learned model 25B based on the learning data acquired by the learning data acquisition unit 26. As described above, deep learning can be used as an example of machine learning.
The machine learning unit 27 stores the updated learned model 25B in the storage unit 24.
The configuration of the terminal device 40 is the same as that of the first embodiment.

<Operation mode>
FIG. 12 is a flowchart showing an example of a processing procedure of the function improvement support device according to the second embodiment of the present invention. FIG. 12 shows a processing procedure of the function improvement support device 20 in the operation mode in which the exercise program of the target user 50 is determined from the questions and answers to the target user 50.

The answer acquisition unit 29 acquires the answers to the language or oral question items and the examination items of the target user 50 as feature quantities related to the functions of the target user 50 (S21). That is, the answer acquisition unit 29 acquires the item ID of the questionnaire and the set of the answers as shown in FIG.

The evaluation acquisition unit 30 acquires the evaluation by the oral profession for the language of the target user 50 or the evaluation item of the oral cavity as a feature amount related to the function of the target user 50 (S22). That is, the evaluation acquisition unit 30 acquires the item ID of the evaluation report and the evaluation set thereof as shown in FIG.

The exercise program determination unit 23 includes the item ID and answer set acquired by the answer acquisition unit 29, the item ID and evaluation set acquired by the evaluation acquisition unit 30, and the attributes (gender, age, weight, height) of the target user 50. ) And Is input to the trained model 25B, and the management guidance item name and the exercise program name are determined (S23).

The exercise program determination unit 23 transmits the determined management guidance item and the content of the exercise program to the terminal device 40 (S24). The exercise program determination unit 23 may transmit the management guidance item name and the exercise program name to the terminal device 40 instead of the management guidance item and the content of the exercise program.

Also in the second embodiment, the function improvement support device 20 relearns the trained model in the learning mode. The processing flow is the same as that shown in FIG. 6, except that the position of the center of gravity and the momentum are changed to the set of item ID and answer and the set of item ID and evaluation.

FIG. 13 is a diagram showing an example of a management guidance plan for an oral function improving service created based on the management guidance items and the contents of the exercise program transmitted to the terminal device 40.

In the management guidance plan, management guidance items and exercise programs to be implemented for the target user 50 are shown. For example, as a management guidance item, "providing information on improvement of oral function" by a profession is shown. In addition, as exercise programs, "oral exercises / swallowing exercises" and "salivary gland massage" by professionals, "oral exercises / swallowing exercises" by related occupations, and "oral exercises / swallowing exercises" by the person himself / herself are shown. Has been done.

<Effect of Embodiment 2>
As described above, according to the second embodiment of the present invention, the answer to the question item and the examination item of the language or oral cavity of the target user 50, and the evaluation by the oral profession for the evaluation item of the language or oral cavity of the target user 50. Is acquired, and by inputting the acquired answers and evaluations into the trained model 25B, it is possible to determine management instruction items and exercise programs for improving the language- or oral-related functions of the target user 50. .. Thereby, for example, instead of providing the same management guidance item and exercise program to a plurality of target users 50 with a language disorder, the most effective management guidance item and exercise program are determined for each target user 50. Can be provided.

[Modified Example of Embodiment 2]
In the above-described second embodiment, the model is learned by supervised machine learning, but it is also possible to learn the model by reinforcement learning.

For example, a model can be created using reinforcement learning. Here, the feature quantities (item ID and answer set, item ID and evaluation set) and attributes related to the user's function are set as the states in reinforcement learning, and the management guidance item name and the exercise program name are set as actions in reinforcement learning. , A model can be created by performing reinforcement learning with the degree of improvement as a reward in reinforcement learning. Reinforcement learning can also be referred to as deep reinforcement learning, and the model can be expressed by a neural network by performing deep reinforcement learning.

The learning data acquisition unit 26 of the function improvement support device 20 shown in FIG. 2 acquires learning data for re-learning the trained model 25B. For example, the learning data acquisition unit 26 has user information about the target user 50 of the user ID “u1” shown in FIG. 9, and the management guidance item and exercise program name of the management guidance item name “m1” shown in the user information. The degree of improvement in the function of the target user 50 after executing the exercise program of "t1" is acquired as learning data. Similarly, the learning data acquisition unit 26 acquires user information and the degree of improvement of the function as learning data for the other target users 50 such as the user IDs “u2” and “u3”. The degree of improvement is quantified in five stages from 1 to 5, for example, 1 is the highest degree of improvement, the degree of improvement is lower as the numerical value is larger, and 5 is the lowest degree of improvement.

The machine learning unit 27 updates the learned model 25B by further machine learning the learned model 25B based on the learning data acquired by the learning data acquisition unit 26. As described above, machine learning can use reinforcement learning or deep reinforcement learning as an example.
The machine learning unit 27 stores the updated learned model 25B in the storage unit 24.

[Additional Notes]
Although the embodiment of the function improvement support system according to the present invention has been described above, the present invention is not limited to the above-described embodiment.

For example, in the above-described first embodiment, the position of the center of gravity and the amount of exercise of the target user 50 are set as feature amounts, but the feature amount is not limited to this. For example, the distance from the ground to the center of gravity of the target user 50, the distance from the distance image sensor 10 to the center of gravity of the target user 50, and the like may be used as feature quantities. Further, if more detailed information on the motor function of the target user 50 can be obtained, that information may be used as a feature quantity. For example, indicators related to cervical flexion angle, rising of intraaxial rotation, indicators related to rising such as forward movement, vertical movement or downward movement of the center of gravity, backward movement of the posterior leg, or of the anterior leg. The values of various indicators such as indicators related to crouching such as bending the knee may be used as the feature amount of the target user 50. The index value can be automatically calculated based on the distance image, or the caregiver or the like may input the index value to the function improvement support device 20. Moreover, you may use these feature quantities in combination.

Further, in the second embodiment described above, the management guidance plan and the exercise program for improving the language- or oral-related functions of the target user 50 are determined. For example, either the management guidance plan or the exercise program is used. It may be something to decide.

Note that some or all of the functions of the above-mentioned function improvement support device 20 may be provided by cloud computing. That is, some or all the functions of the function improvement support device 20 may be realized by the cloud server. For example, in the function improvement support device 20, the functions of the exercise program determination unit 23, the storage unit 24, the learning data acquisition unit 26, and the machine learning unit 27 may be realized by the cloud server. The function improvement support device 20 may acquire the most effective exercise program or exercise program name for improving the function of the target user 50 by transmitting the feature amount and the attribute of the target user 50 to the cloud server. ..

Further, the modified examples of the above-described first and second embodiments can be represented by the following appendices 1 to 4.

<Appendix 1>
An acquisition unit that acquires features related to the function of the target user,
It is provided with an exercise program determination unit that determines an exercise program for improving the function of the target user from the acquired feature amount based on the trained model.
The trained model is machine-learned using feature quantities related to functions for a plurality of users, an exercise program executed by each user, and the degree of improvement of the functions after execution of the exercise program as learning data, and the feature quantities are obtained. A function improvement support device that is a model that outputs identification information of the corresponding exercise program.

<Appendix 2>
The function is a function related to the user's exercise or life, and is
The acquisition unit acquires a distance image showing the distance to the target user as the feature amount.
Based on the learned model, the exercise program determination unit determines a problem related to the exercise or life of the user and an exercise program for improving the problem from the acquired feature amount.
The trained model shows, for a plurality of users, the distance image of each user, problems related to each user's exercise or life, the exercise program executed by each user, and the degree of improvement of the function after execution of the exercise program. The function improvement support device according to Appendix 1, which is a model that is machine-learned as learning data and outputs identification information of the problem and the exercise program corresponding to the feature amount.

<Appendix 3>
The function is a function related to the user's language or oral cavity.
The acquisition unit acquires the answers to the question items and examination items of the language or oral cavity of the target user and the evaluations by the oral profession for the evaluation items of the language or oral cavity of the target user as the feature quantities.
In the trained model, for a plurality of users, each user's answer to the language or oral question item and test item, evaluation of each user's evaluation item by an oral specialist, management instruction item and exercise program executed by each user. , And the degree of improvement of the function after executing the management guidance item and the exercise program are machine-learned as learning data, and the identification information of the management guidance item and the identification information of the exercise program corresponding to the feature amount are output. Is a model
The function improvement support according to Appendix 1, wherein the exercise program determination unit determines a management instruction item and an exercise program for improving the function of the target user from the acquired feature amount based on the learned model. apparatus.

<Appendix 4>
Further, a machine learning unit for machine learning the trained model is provided.
The machine learning unit further learns the feature amount of the target user who has executed the exercise program determined by the exercise program determination unit, the identification information of the exercise program, and the degree of improvement of the function after the exercise program is executed. The function improvement support device according to any one of Supplementary note 1 to Supplementary note 3, which machine-learns the trained model as data for use.

It should be considered that the embodiments disclosed this time are exemplary in all respects and not restrictive. The scope of the present invention is indicated by the scope of claims, not the above-mentioned meaning, and is intended to include all modifications within the meaning and scope equivalent to the scope of claims.

1 Function improvement support system 2 Network 10 Distance image sensor 20 Function improvement support device 21 Distance image acquisition unit 23 Exercise program determination unit 24 Storage unit 25A User information 25B Learned model 26 Learning data acquisition unit 27 Machine learning unit 28 User attribute acquisition Unit 29 Answer acquisition unit 30 Evaluation acquisition unit 40 Terminal device 41 Exercise program acquisition unit 42 Exercise program presentation unit 50 Target user

Claims (2)

As feature quantities related to the target user's language or oral cavity-related functions, the answers to the target user's language or oral cavity question items and test items, and the evaluation by the oral profession for the target user's language or oral cavity evaluation items. And the acquisition department to acquire,
Based on the learned model from the acquired feature quantity, and a exercise program determining portion for determining a management guidance items and exercise program to improve the function of the target user,
In the trained model, for a plurality of users , answers to each user's language or oral question items and test items, evaluations of each user's evaluation items by an oral specialist, and management guidance items and exercise programs executed by each user. Is a function improvement support device that is machine-learned as learning data and outputs identification information of the management guidance item and identification information of the exercise program corresponding to the feature amount. The function improvement support device according to claim 1 and
A function improvement support system including a management guidance item and an exercise program for improving the function of a target user from the function improvement support device, and a terminal device for presenting the acquired management guidance item and the exercise program.

Images