Rehabilitation Support Device, Control Method of Rehabilitation Support Device, and Program

Abstract

To allow a patient to get a proper instruction in a rehabilitation without requiring an evaluation by an expert who examined the patient face to face.SOLUTION: A rehabilitation support device includes: an input unit for inputting sensor data showing motions of a body; and an output unit for outputting the content of a suggestion regarding a rehabilitation according to the input sensor data by using a learned algorithm having learned in advance teacher data including the sensor data and the content of a suggestion on a rehabilitation.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present invention relates to a rehabilitation support device, Control Method of Rehabilitation Support Device and a program.

Background Art

[0002] Patent Document 1 discloses a support device that can support correct movement during rehabilitation. More specifically, when the "movement during rehabilitation" of the user acquired via a sensor by the support device is different from the "information on correct movement during rehabilitation performed with a therapist", it is described that the support device notifies that the "movement" is incorrect and guides the user to approach the "correct movement".

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] Hitherto, experts have met with paralytic patients, observed their movements, conducted functional evaluations, and provided appropriate advice and goal setting. In recent years, due to the influence of the COVID-19 pandemic, the demand for tele-rehabilitation is expected to increase. Therefore, there is a need for a mechanism that allows patients to receive appropriate instructions in rehabilitation without the need for an in-person evaluation by an expert.

[0005] The present invention has been made in view of such problems, and an object of the present invention is to provide a rehabilitation support device and a program that enable patients to receive appropriate instructions in rehabilitation without the need for an in-person evaluation by an expert. Control Method of Rehabilitation Support Device and a program.

Means for Solving the Problems

[0006] According to a first aspect of the present invention, a rehabilitation support device includes an input unit that inputs sensor data indicating body movement, and an output unit that outputs proposal contents related to rehabilitation according to the input sensor data, using a pre-trained learned algorithm using teacher data including the sensor data and proposal contents related to rehabilitation.

[0007] According to a second aspect of the present invention, the output unit outputs a target value of activity amount as the proposal contents related to the rehabilitation.

[0008] According to a third aspect of the present invention, the output unit outputs a target of a training operation as the proposal contents related to the rehabilitation.

[0009] According to a fourth aspect of the present invention, the output unit outputs achievable daily living activities as the proposal contents related to the rehabilitation.

[0010] According to a fifth aspect of the present invention, the output unit outputs a target accompanying a change in severity as the proposal contents related to the rehabilitation.

[0011] According to a sixth aspect of the present invention, the proposal contents related to rehabilitation are proposal contents determined corresponding to the evaluation result of severity based on FMA.

[0012] According to a seventh aspect of the present invention, a rehabilitation support method includes a step of inputting sensor data indicating body movement, and a step of outputting proposal contents related to rehabilitation according to the input sensor data, using a pre-trained learned algorithm using teacher data including the sensor data and proposal contents related to rehabilitation.

[0013] According to an eighth aspect of the present invention, the program causes a computer of the rehabilitation support device to perform a step of inputting sensor data indicating body movements, and a step of outputting proposed content related to rehabilitation according to the input sensor data, using a pre-trained learned algorithm with teacher data including the sensor data and proposed content related to rehabilitation.

Advantages of the Invention

[0014] According to the above-described rehabilitation support device, rehabilitation support method, and program, a patient can receive appropriate instructions in rehabilitation without requiring an evaluation through a face-to-face meeting with an expert.

Brief Description of the Drawings

[0015]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Modes for Carrying Out the Invention

[0016] <The First Embodiment> Hereinafter, a rehabilitation support system according to a first embodiment of the present invention will be described with reference to FIGS. 1 to 8.

[0017] (Overall Configuration of Rehabilitation Support Device) FIG. 1 is a diagram showing the overall configuration of a rehabilitation support device according to the first embodiment. The rehabilitation support device 1 shown in FIG. 1 is a wristwatch-type wearable terminal device. The rehabilitation support device 1 is used, for example, by being attached to the paralyzed arm of an upper limb paralysis patient. In this embodiment, the rehabilitation support device 1 will be described as a dedicated terminal device, but in other embodiments, it is not limited to this mode. For example, the rehabilitation support device 1 may function as the rehabilitation support device 1 described below by applying a dedicated program (application) to a general smartphone or smartwatch.

[0018] As shown in FIG. 1, the rehabilitation support device 1 includes a CPU 10, a memory 11, a touch sensor 12, a display 13, and a three-axis accelerometer 14.

[0019] The CPU 10 is a processor that operates according to a pre-prepared program. By operating according to this program, the CPU 10 exhibits various functions described later.

[0020] The memory 11 is a so-called main storage device and has a storage area necessary for the operation of the CPU 10.

[0021] The touch sensor 12 is an example of an input means for receiving a touch operation by a user (upper limb paralysis patient).

[0022] The display 13 is an example of a display means for displaying various information and notifying the user.

[0023] The three-axis accelerometer 14 is a sensor capable of detecting the movement (acceleration) of the main body (rehabilitation support device 1) in space.

[0024] The CPU 10 of the rehabilitation support device 1 functions as the input unit 100 and the output unit 101 by operating according to a program.

[0025] The input unit 100 inputs sensor data obtained through the three-axis accelerometer 14, which indicates the movement of the user's body (the upper limb on the paralyzed side in this embodiment).

[0026] The output unit 101 uses the learned machine learning algorithm AL to display (output) the rehabilitation-related proposal content corresponding to the sensor data input by the input unit 100 on the display 13.

[0027] The machine learning algorithm AL is a learned algorithm that has been pre-learned using teacher data including sensor data and rehabilitation-related proposal content.

[0028] (Function of the machine learning algorithm) Figures 2 and 3 are diagrams showing the functions of the machine learning algorithm according to the first embodiment. Figure 2 shows the function of the machine learning algorithm AL in the learning stage. Further, Figure 3 shows the function of the machine learning algorithm AL in the practical stage.

[0029] As shown in Figure 2, machine learning (supervised learning) is performed by inputting the teacher data DT into the machine learning algorithm AL.

[0030] The teacher data DT is teacher data that sets the sensor data acquired from the three-axis accelerometer 14 and the rehabilitation-related proposal content corresponding to the sensor data.

[0031] Here, a general method for evaluating upper limb function for upper limb paralysis will be briefly described. Regarding the evaluation method of upper limb function evaluation, it is common for experts (occupational therapists) to conduct it face-to-face with patients, and usually, it is quantitatively evaluated based on the Fugl-Meyer Assessment (hereinafter referred to as "FMA"). FMA is the most internationally used in the upper limb function evaluation of stroke, has high reliability among examiners, and is an outcome of the effect. The upper limb motor function of FMA is evaluated in three levels of "0 points (completely unable)", "1 point (insufficient)", and "2 points (fully possible)" for each item, and the total score (out of 66 points) becomes the upper limb function evaluation (severity). FMA is evaluated by experts having patients perform various movements of the shoulder, forearm, wrist joint, and fingers. Then, based on the score (severity) of this FMA, experts set appropriate goals and give advice to patients.

[0032] The machine learning algorithm AL according to this embodiment is learned to conform to this FMA. Specifically, as shown in FIG. 2, the teacher data DT is prepared as a pair with a goal setting based on the evaluation result of severity by FMA for various sensor data (input values). Here, the sensor data is a group of data acquired by the three-axis accelerometer 14, and includes the activity amount of the upper limb (the arm to which the rehabilitation support device 1 is attached), the trajectory of the upper limb movement, the moving distance of the upper limb, and the like.

[0033] According to the machine learning algorithm AL learned in this way, in the practical stage, as shown in FIG. 3, when the sensor data D1 of a certain patient is input, based on the severity estimated from the sensor data D1, a patient-specific goal setting D2 conforming to FMA can be output.

[0034] (Processing of the rehabilitation support device) FIGS. 4 to 8 are diagrams showing examples of the processing by the rehabilitation support device according to the first embodiment. Hereinafter, with reference to FIGS. 4 to 8, the specific processing of the rehabilitation support device according to the first embodiment will be described in detail.

[0035] (Upper limb activity amount) As shown in FIG. 4, the rehabilitation support device 1 outputs a target value of the upper limb activity amount according to the severity of the upper limb paralysis as a proposal content related to rehabilitation.

[0036] Specifically, the input unit 100 of the rehabilitation support device 1 receives the input of the patient-specific sensor data obtained through the three-axis accelerometer 14 (FIG. 1). Then, the output unit 101 inputs the patient-specific sensor data to the pre-learned machine learning algorithm AL, and obtains the target value of the upper limb activity amount from the machine learning algorithm AL.

[0037] As shown in FIG. 4, the rehabilitation support device 1 causes the display 13 (FIG. 1) to display the display image G. On this display image G, for example, the upper limb activity amount and the total activity amount by time zone for the current (1st of today) and the target values of the upper limb activity amount and the total activity amount by time zone for the next time zone (1st of tomorrow) are displayed.

[0038] Here, the target value of the upper limb activity amount output is based on the judgment of an expert using the FMA because it is output by the machine learning algorithm AL described above.

[0039] (Achievable Movements) As shown in FIG. 5, the rehabilitation support device 1 outputs a target (training content) of "achievable movements" according to the severity of the upper limb paralysis as a proposal content related to rehabilitation.

[0040] Specifically, the input unit 100 receives the input of the patient-specific sensor data obtained through the three-axis accelerometer 14 (FIG. 1). Then, the output unit 101 inputs the patient-specific sensor data to the pre-learned machine learning algorithm AL, and obtains the target of "achievable movements" from the machine learning algorithm AL.

[0041] As shown in FIG. 5, the rehabilitation support device 1 causes the display 13 (FIG. 1) to display the display image G. On this display image G, for example, the target of achievable movements (training content) is displayed as an animation.

[0042] The target of the "attainable movements" output here is output by the machine learning algorithm AL described above, and thus conforms to the judgment of an expert using the FMA.

[0043] (Attainable daily living activities) As shown in FIGS. 6 and 7, the rehabilitation support device 1 outputs, as proposed content related to rehabilitation, the target of "attainable daily living activities" according to the severity of upper limb paralysis.

[0044] Specifically, the input unit 100 receives the input of patient-specific sensor data obtained through the three-axis accelerometer 14 (FIG. 1). Then, the output unit 101 inputs the patient-specific sensor data to the pre-learned machine learning algorithm AL, and acquires the target of "attainable daily living activities" from the machine learning algorithm AL.

[0045] As shown in FIG. 6, the rehabilitation support device 1 causes the display 13 (FIG. 1) to display a display image G indicating the target of the eating operation. The target of the eating operation is set to, for example, any one of "total assistance", "using a universal cuff spoon", "using a thick grip self-help spoon", "self-supporting with a normal spoon", "self-supporting with a self-help chopstick + normal spoon", and "self-supporting with chopsticks" according to the severity.

[0046] Also, as shown in FIG. 7, the rehabilitation support device 1 causes the display 13 (FIG. 1) to display a display image G indicating the target of the dressing operation. The target of the dressing operation is set to, for example, any one of "total assistance", "able to stretch the arm at a low position to put on the sleeve", "able to put on the front-opening shirt by raising the shoulder slightly and stretching the arm to put on the sleeve", "able to put on the front-opening shirt by raising the shoulder high and stretching the arm to put on the sleeve", and "able to button up" according to the severity.

[0047] The target of the "attainable daily living activities" output as shown in FIGS. 6 and 7 is output by the machine learning algorithm AL described above, and thus conforms to the judgment of an expert using the FMA.

[0048] (Setting of Target Values According to Changes in Severity) As shown in FIG. 8, as a proposal regarding rehabilitation, the rehabilitation support device 1 outputs a target according to changes in the severity of upper limb paralysis.

[0049] Specifically, the input unit 100 receives the input of sensor data specific to the patient obtained through the three-axis accelerometer 14 (FIG. 1). Then, the output unit 101 inputs the sensor data specific to the patient to the pre-learned machine learning algorithm AL, and obtains a target according to changes in severity from the machine learning algorithm AL.

[0050] For example, when the rehabilitation support device 1 compares the total activity amount a week ago with the current total activity amount and finds an improvement trend, as the target value of the total activity amount, a target value higher than the target value (standard) set based only on the current total activity amount is set (see FIG. 8(a)).

[0051] Also, when the rehabilitation support device 1 compares the total activity amount a week ago with the current total activity amount and does not find an improvement trend, as the target value of the total activity amount, a target value lower than the target value (standard) set based only on the current total activity amount is set (see FIG. 8(b)).

[0052] Note that FIG. 8 explains the target setting of "total activity amount" as an example of "setting of target values according to changes in severity", but the target settings of "attainable movements" (FIG. 5) and "attainable activities of daily living" (FIGS. 6 and 7) are processed in the same way.

[0053] Such setting of target values according to changes in severity (longitudinal analysis) is also output by the above-mentioned machine learning algorithm AL, so it conforms to the judgment of experts using FMA.

[0054] (Function and Effect) As described above, the rehabilitation support device 1 according to the first embodiment has "sensor data" from an activity sensor attached to a patient, and a learned AI (machine learning algorithm AL) that has learned proposal contents such as "proposal of activity amount and movement goals", "proposal of achievable daily actions", and "setting of target values according to changes in severity (longitudinal analysis)" through supervised learning. By using this learned AI, a patient can receive advice and instructions equivalent to those of an expert without the need for an evaluation through a face-to-face meeting with an expert.

[0055] In the above-described embodiment, the processes of various processes of the rehabilitation support device 1 are stored in a computer-readable recording medium in the form of a program, and the above-described various processes are performed by the computer reading and executing this program. Also, the computer-readable recording medium refers to a magnetic disk, magneto-optical disk, CD-ROM, DVD-ROM, semiconductor memory, and the like. Further, this computer program may be distributed to a computer via a communication line, and the computer that has received this distribution may execute the program.

[0056] The above program may be for realizing a part of the functions described above. Furthermore, it may be a so-called difference file (difference program) that can be realized in combination with a program already recorded in the computer system for realizing the above-described functions.

[0057] As described above, some embodiments according to the present disclosure have been described, but all of these embodiments are presented as examples and are not intended to limit the scope of the invention. These embodiments can be implemented in various other forms, and various omissions, replacements, and changes can be made without departing from the gist of the invention. These embodiments and their modifications are included in the scope and gist of the invention, and are also included in the invention described in the claims and the equivalent scope thereof.

Explanation of Reference Numerals

[0058] 1 Rehabilitation support device 10 CPU 11 Memory 12 Touch sensor 13 Display 14 Triaxial accelerometer 100 Input section 101 Output section AL Machine learning algorithm

Claims

1. An input unit that inputs sensor data indicating body movement; An output unit that outputs a rehabilitation goal setting according to the input sensor data, using a pre-learned algorithm learned in advance using teacher data including the sensor data and a goal setting related to rehabilitation; Comprising: The pre-learned algorithm is learned using the teacher data prepared as a pair of sensor data capable of acquiring at least one of the activity amount of the upper limb, the trajectory of the upper limb movement, and the movement distance of the upper limb of a certain patient, and the goal setting based on the FMA score evaluated by an expert for the same patient. Rehabilitation support device.

2. The output unit outputs a target value of the activity amount as the proposed content related to the rehabilitation. The rehabilitation support device according to claim 1.

3. The output unit outputs an achievable movement target as the proposed content related to the rehabilitation. The rehabilitation support device according to claim 1 or claim 2.

4. The output unit outputs a target associated with a change in severity as the proposed content related to the rehabilitation. The rehabilitation support device according to any one of claims 1 to 3.

5. The proposed content related to the rehabilitation is the proposed content determined corresponding to the evaluation result of the severity based on the FMA. The rehabilitation support device according to any one of claims 1 to 4.

6. A step in which the rehabilitation support device inputs sensor data indicating body movement; A step in which the rehabilitation support device outputs a rehabilitation-related goal setting according to the input sensor data, using a pre-learned algorithm learned in advance using teacher data including the sensor data and a goal setting related to rehabilitation; Comprising: The pre-learned algorithm is learned using the teacher data prepared as a pair of sensor data capable of acquiring at least one of the activity amount of the upper limb, the trajectory of the upper limb movement, and the movement distance of the upper limb of a certain patient, and the goal setting based on the FMA score evaluated by an expert for the same patient. Control method of the rehabilitation support device.

7. On the computer of the rehabilitation support device, A step of inputting sensor data indicating body movement; Outputting a rehabilitation goal setting according to the input sensor data using a pre-trained algorithm using teacher data including the sensor data and a goal setting related to rehabilitation; A program for causing execution, wherein the pre-trained algorithm is trained using teacher data prepared as a pair of sensor data capable of acquiring at least any one of the activity amount of the upper limb, the trajectory of the upper limb movement, and the movement distance of the upper limb of a certain patient, and the goal setting based on the FMA score evaluated by an expert for the same patient; Program.

Images