Evaluation index calculation method, evaluation index calculation device, and program

Abstract

To evaluate assistance skills quantitatively.SOLUTION: A sensor information acquisition unit acquires sensor information corresponding to an assistance operation for assisting posture change of a care-receiver on a bed. An evaluation index calculation unit calculates evaluation index information for evaluating assistance corresponding to evaluation object information using reference information, which is the sensor information as a reference for evaluation, and the evaluation object information, which is the sensor information as an evaluation object. The sensor information is the information indicating a time-series change in the load applied from the bed to a floor surface, which is measured by a load sensor provided in each of a plurality of places of the bed in correspondence with the assistance operation. The reference information is the sensor information in which a process in which the load decreases according to an operation to lift up the care-receiver in the middle of assistance is not included.SELECTED DRAWING: Figure 2

Description

[Technical field]

[0001] The present invention relates to an evaluation index calculation method, an evaluation index calculation device, and a program. [Background technology]

[0002] There is a technology for acquiring biometric information of a subject lying in bed in order to know the physical condition of a patient or a care recipient in a medical or nursing care setting. The biometric information is, for example, information indicating the subject's body movement, breathing, etc. acquired using a load sensor or the like. Patent Document 1 discloses a technology for monitoring the biometric information of a subject lying in bed using multiple load sensors provided on the bed or under the legs of the bed. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Patent No. 6105703 Summary of the Invention [Problem to be solved by the invention]

[0004] Meanwhile, in the field of medical care and nursing care, the training of caregivers is urgently needed with the increase in the elderly population. In the description of Patent Document 1, it is possible to monitor the biological information of the subject on the bed, but it is not assumed that it will be used for training of caregivers. It is common for caregiving skills to be acquired over time while interacting with the care recipient in the field, and it takes a long time for ordinary people who are not caregivers to acquire the caregiving skills. For this reason, it has been difficult to train caregivers quickly. In the process of proceeding with the study to achieve the early training of caregivers, the inventor recognized and solved the problem of quantitatively evaluating the caregiving skills.

[0005] An object of the present invention is to provide an evaluation index calculation method, an evaluation index calculation device, and a program that make it possible to quantitatively evaluate assistance skills. [Means for solving the problem]

[0006] In one embodiment of the present invention, an evaluation index calculation method includes a sensor information acquisition unit that acquires sensor information corresponding to an assisting action of assisting a person being assisted on a bed to change his / her position, and an evaluation index calculation unit that uses reference information, which is the sensor information used as the basis of evaluation, and evaluation target information, which is the sensor information used as the subject of evaluation, to calculate evaluation index information for evaluating the assistance corresponding to the evaluation target information, wherein the sensor information is information indicating the time-series change in the load applied from the bed to the floor surface, measured by load sensors provided at each of multiple locations on the bed corresponding to the assisting action, and the reference information is sensor information that does not include a process in which the load decreases in response to the action of lifting up the person being assisted during the assistance.

[0007] In the above evaluation index calculation method, the sensor information acquisition unit acquires the sensor information from load sensors respectively arranged between a plurality of legs of the bed and a floor surface.

[0008] An evaluation index calculation device according to one embodiment of the present invention comprises a sensor information acquisition unit that acquires sensor information corresponding to an assisting action of assisting a person being assisted on a bed to change his / her position, and an evaluation index calculation unit that calculates evaluation index information for evaluating the assistance corresponding to the evaluation target information using reference information, which is the sensor information used as the basis of evaluation, and evaluation target information, which is the sensor information used as the object of evaluation, wherein the sensor information is information indicating the time-series change in the load applied from the bed to the floor surface, measured by load sensors provided at each of a number of locations on the bed corresponding to the assisting action, and the reference information is sensor information that does not include a process in which the load decreases in response to the action of lifting up the person being assisted during assistance.

[0009] A program according to one embodiment of the present invention causes a computer, which is an evaluation index calculation device, to function as a sensor information acquisition means for acquiring sensor information corresponding to an assisting action of assisting a person being assisted on a bed to change his / her position, and an evaluation index calculation means for calculating evaluation index information for evaluating the assistance corresponding to the evaluation target information using reference information, which is the sensor information used as the basis of evaluation, and evaluation target information, which is the sensor information used as the evaluation target, wherein the sensor information is information indicating the time-series change in the load applied from the bed to the floor surface, measured by load sensors provided at each of a number of locations on the bed corresponding to the assistance action, and the reference information is sensor information that does not include a process in which the load decreases in response to the action of lifting up the person being assisted during the assistance action. Effect of the Invention

[0010] According to the present invention, it is possible to quantitatively evaluate assistance skills. [Brief description of the drawings]

[0011] [Figure 1] FIG. 1 is a diagram showing an outline of an evaluation system 1 according to an embodiment. [Diagram 2] 1 is a block diagram showing an example of the configuration of a load sensor 10 and an evaluation device 20 according to an embodiment. [Diagram 3] FIG. 2 is a diagram showing an example of the configuration of evaluation target information 250 in the embodiment. [Figure 4] FIG. 2 is a diagram showing an example of the configuration of reference information 251 in the embodiment. [Diagram 5] FIG. 2 is a diagram showing an example of evaluation target information 250 in the embodiment. [Figure 6] FIG. 2 is a diagram showing an example of evaluation target information 250 in the embodiment. [Figure 7] FIG. 2 is a diagram illustrating an example of evaluation index information 252 in the embodiment. [Figure 8] FIG. 2 is a diagram illustrating an example of evaluation index information 252 in the embodiment. [Figure 9]FIG. 2 is a diagram illustrating an example of evaluation index information 252 in the embodiment. [Figure 10] FIG. 2 is a diagram illustrating an example of evaluation index information 252 in the embodiment. [Figure 11] FIG. 2 is a diagram illustrating an example of evaluation index information 252 in the embodiment. [Figure 12] 2 is a sequence diagram showing the flow of processing performed by the evaluation system 1 in the embodiment. FIG. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0012] Hereinafter, an embodiment of the present invention will be described in detail with reference to the drawings. Fig. 1 is a diagram showing an outline of an evaluation system 1 in an embodiment. The evaluation system 1 is composed of, for example, a plurality of load sensors 10 (load sensors 10-1 to 10-4), an evaluation device 20, a person being assisted 30, a caregiver 40, and a bed B. In the evaluation system 1, the load sensors 10 and the evaluation device 20 are connected so as to be able to communicate with each other.

[0013] 1 illustrates a case where four load sensors 10 are provided on bed B, but the number of load sensors 10 provided on bed B may be any number and is not limited to four. In the following, an example will be described in which the load sensors 10 are provided on the undersides of the legs at the four corners of bed B, but the present invention is not limited to this. The load sensors 10 are only required to measure a load acting downward from the top surface of the mattress of bed B, and may be provided, for example, between the bottom surface of the mattress and the top surfaces of the legs.

[0014] The caregiver 40 assists the person being assisted 30 who is in bed B. The assistance here is to change the position of the person being assisted 30. Changing the position means changing the position of the person being assisted 30, for example, between a supine position (lying on one's back), a lateral position (lying on one's side), and an edge-sitting position (sitting on the edge of the bed with the legs below the knees lowered to the bottom) and vice versa.

[0015] If the person being assisted 30 stays in the same position on bed B for a long time, their weight will continue to be placed on the same part of their body. In this case, they will be more likely to develop blood circulation disorders and bedsores. If the person being assisted 30 is healthy and has normal muscle strength, they will be able to change their position by turning over in bed, eliminating the discomfort that comes from staying in the same position and preventing bedsores and blood circulation disorders. However, if the person being assisted 30 has lost muscle strength or stamina, or if they have a physical disability, they will need to be assisted to change their position regularly.

[0016] The load sensors 10 are disposed at the four corners of the bed B. Specifically, when viewed from the person being assisted 30 lying on his / her back on the bed B, the load sensor 10-1 is disposed at the upper left of the bed B, the load sensor 10-2 is disposed at the lower left of the bed B, the load sensor 10-3 is disposed at the lower right of the bed B, and the load sensor 10-4 is disposed at the upper right of the bed B.

[0017] The load sensor 10 includes, for example, a measurement unit and a communication unit. The measurement unit is a functional unit that measures the load and is composed of, for example, a piezoelectric element such as a load cell. The measurement unit converts the pressure acting on the sensor into an electric signal corresponding to the pressure and outputs the converted electric signal to the communication unit. The communication unit is a functional unit that communicates with the evaluation device 20 and is composed of, for example, a communication module that uses a wired connection such as a LAN (Local Area Network) or a wireless connection such as a wireless LAN or Bluetooth. The communication unit transmits the electric signal acquired from the measurement unit to the evaluation device 20.

[0018] The evaluation device 20 acquires an electric signal (information indicating the load) from each of the load sensors 10, and analyzes the acquired information. As the evaluation device 20, a PC (Personal Computer) or the like is used.

[0019] Conventionally, the posture (position) of the person being assisted 30 and the state of sleep have been estimated based on the time series change of the load measured by the load sensor 10 (hereinafter, also referred to as sensor information or waveform). This makes it possible to estimate the health condition of the person being assisted 30 and provide assistance as necessary. For example, the number of times the person turns over in bed can be grasped from the waveform. Using this, it is determined that the person being assisted 30 who turns over a number of times during nighttime sleep is below a threshold value is likely to develop a bedsore. Then, a care plan is made so that the person being assisted 30 who is likely to develop a bedsore is changed in position by the caregiver 40 during the day. This makes it possible to prevent bedsores in the person being assisted 30. Alternatively, it is also possible to determine the content and number of changes in position made by the caregiver 40 based on the time series change of the load from the load sensor 10. The determination result can be used, for example, as evidence in a care record.

[0020] The inventors have studied the waveforms obtained from the load sensor 10 in detail and have noticed that even when the same position change is assisted, each waveform has its own characteristics. For example, when assistance is given to turn leftward (from supine to lateral), a waveform like data HD1 in FIG. 5 is obtained at one time, and a waveform like data HD3 in FIG. 6 is obtained at another time. When assistance is given to turn rightward, a waveform like data HD2 in FIG. 5 is obtained at one time, and a waveform like data HD4 in FIG. 6 is obtained at another time. The inventors have studied the characteristics of the waveforms obtained when assisting with position changes. As a result, they have found that an index for evaluating the assistance action can be derived from the characteristics of the waveforms.

[0021] 2 is a block diagram showing an example of the configuration of the load sensor 10 and the evaluation device 20 according to the embodiment. The evaluation device 20 includes, for example, a sensor information acquisition unit 21, an evaluation index calculation unit 22, a calculation result output unit 23, a device control unit 24, a storage unit 25, and a display unit 26.

[0022] The sensor information acquiring unit 21 acquires sensor information from each of the load sensors 10. The sensor information is information indicating a time series change of the load acting on the floor surface from the bed B, measured by the load sensor 10. Specifically, the sensor information acquiring unit 21 acquires, as the sensor information, information indicating a time series change of the load acting on the floor surface from the upper end of the left side of the bed B from the load sensor 10-1, information indicating a time series change of the load acting on the floor surface from the lower end of the left side of the bed B from the load sensor 10-2, information indicating a time series change of the load acting on the floor surface from the lower end of the right side of the bed B from the load sensor 10-3, and information indicating a time series change of the load acting on the floor surface from the upper end of the right side of the bed B from the load sensor 10-4.

[0023] The evaluation index calculation unit 22 calculates an evaluation index based on the sensor information. The evaluation index is an index for evaluating the assisting action. The evaluation index calculation unit 22 outputs information indicating a difference from a reference assisting action as the evaluation index. The reference assisting action is an action that serves as a model of an assisting action, such as an assisting action performed by an experienced caregiver, an assisting action that the assisted person 30 feels safe with, an assisting action that the caregiver 40 feels is less burdensome, or an action that is determined to be a model of an assisting action in a comprehensive manner based on a combination of these.

[0024] In this embodiment, in order to clarify the difference between the assisting motion that is the evaluation standard and the assisting motion that is the subject of evaluation, sensor information with common measurement conditions is used for the information indicating both motions. That is, sensor information corresponding to a model motion of assistance is used as the information indicating the assisting motion that is the evaluation standard. Also, sensor information corresponding to the assisting motion by the caregiver 40 that is the subject of evaluation is used as the information indicating the assisting motion that is the subject of evaluation.

[0025] The evaluation index calculation unit 22 outputs, for example, information indicating the difference from a reference care action, which indicates the matching, similar or different parts between the sensor information used as the reference for evaluation (hereinafter referred to as reference information) and the sensor information to be evaluated (hereinafter referred to as evaluation target information). This allows the caregiver 40 to recognize the difference between the care action performed by himself / herself and the model action, and to grasp the quality of the care action performed by himself / herself and points to be improved.

[0026] When calculating the evaluation index, the evaluation index calculation unit 22 first selects sensor information (reference information) to be used as a standard for evaluation. For example, the evaluation index calculation unit 22 acquires sensor information corresponding to each of a plurality of care actions from the sensor information acquisition unit 21. The evaluation index calculation unit 22 sets one or a plurality of pieces of sensor information selected from the acquired plurality of pieces of sensor information as reference information.

[0027] In this case, the evaluation index calculation unit 22 may select the reference information based on any selection criterion. For example, the evaluation index calculation unit 22 sets the sensor information selected by the operation input by the operator as the reference information. Alternatively, the evaluation index calculation unit 22 may select the sensor information corresponding to the assisting action performed by the expert as the reference information. In this case, the evaluation index calculation unit 22 acquires, for example, information indicating the proficiency of the assistant 40 together with the sensor information, and determines whether or not to select the sensor information acquired from the sensor information acquisition unit 21 as the reference information based on the acquired information indicating the proficiency of the assistant 40.

[0028] Furthermore, the evaluation index calculation unit 22 selects sensor information to be evaluated (evaluation target information). For example, the evaluation index calculation unit 22 acquires sensor information corresponding to each of a plurality of care actions from the sensor information acquisition unit 21. The evaluation index calculation unit 22 selects one or a plurality of pieces of sensor information from among the acquired plurality of pieces of sensor information excluding the sensor information set as the reference information, as the evaluation target information.

[0029] For example, the evaluation index calculation unit 22 sets the sensor information selected by the operation input by the operator as the evaluation target information. In this case, the operator is, for example, the caregiver 40, and after the caregiver performs care, the sensor information corresponding to the care action is set as the evaluation target information. This enables the caregiver 40 to obtain the evaluation index of the care action performed by the caregiver 40.

[0030] The evaluation index calculation unit 22 displays, for example, both the sensor information corresponding to the reference information and the evaluation target information side by side (or overlapping) as the evaluation index. This makes it possible to present the difference from the model movement in a recognizable manner. In this case, the evaluation index calculation unit 22 displays both the reference information and the evaluation target information side by side for four waveforms corresponding to each of the multiple sensor information (for example, see FIG. 8).

[0031] Alternatively, the evaluation index calculation unit 22 may display both the reference information and the evaluation target information side by side for one waveform representative of the plurality of pieces of sensor information (see, for example, FIG. 9). The method of selecting one waveform representative of the plurality of pieces of sensor information may be any method. For example, the evaluation index calculation unit 22 may select one piece of sensor information selected from the plurality of pieces of sensor information as one waveform representative of the plurality of pieces of sensor information. Alternatively, the evaluation index calculation unit 22 may select a waveform indicating the statistics of the plurality of pieces of sensor information as one waveform representative of the plurality of pieces of sensor information. The statistics of the sensor information here is, for example, information indicating the statistics of the loads measured by each of the plurality of load sensors 10 at the same time in a time series. In addition, the statistics here is information indicating the tendency of an aggregate formed by the plurality of loads, and is information calculated using a statistical method. The statistics here is, for example, information indicating a simple arithmetic average, a weighted average, a square mean value, a square root of the square mean value, a maximum value, a minimum value, a mode, or the like, calculated using each of the plurality of loads or the amount of change in the plurality of loads.

[0032] Alternatively, the evaluation index calculation unit 22 calculates the degree of similarity (similarity) between the waveforms of the reference information and the evaluation target information. This allows the caregiver 40 to recognize the difference from the model movement.

[0033] When calculating the similarity, the evaluation index calculation unit 22 first acquires waveform feature quantities of both the reference information and the evaluation target information. The feature quantity here is the degree of the feature of the waveform. The evaluation index calculation unit 22 acquires, as the waveform feature quantities, feature quantities regarding features that can be recognized by humans, such as the time required for assistance, the load extracted from the waveform, and the amount of change in the load. Alternatively, the evaluation index calculation unit 22 may acquire, as the waveform feature quantities, feature quantities regarding multidimensional features that cannot necessarily be recognized by humans, which are acquired using a machine learning technique described later.

[0034] The evaluation index calculation unit 22, for example, expresses (quantifies) the feature amount acquired from the waveform (the feature amount of the waveform of both the reference information and the evaluation target information) as a vector and arranges the feature amount expressed as a vector in a vector space. Then, the evaluation index calculation unit 22 calculates the distance (for example, cosine similarity, etc.) between the vectors corresponding to the feature amounts of both the waveforms arranged in the vector space as the similarity (for example, see FIG. 10).

[0035] Alternatively, the evaluation index calculation unit 22 may use a trained model when calculating the similarity. In this case, the evaluation index calculation unit 22 first creates a trained model. For example, the evaluation index calculation unit 22 creates a learning data set in which already acquired sensor information (for example, basic information) is associated with a label indicating the quality of the assisting action corresponding to the sensor information. The quality of the assisting action corresponding to the sensor information may be determined according to the proficiency of the caregiver 40, or may be determined based on the shape of the waveform.

[0036] The evaluation index calculation unit 22 creates a trained model by, for example, having a learning model such as a CNN (Convolutional Neural Network) learn a learning dataset by machine learning. The evaluation index calculation unit 22 inputs the input (sensor information) of the learning dataset to the learning model and acquires a value output from the learning model. The evaluation index calculation unit 22 adjusts the internal parameters of the learning model so that the value output from the learning model approaches the output of the learning dataset (a label indicating the quality of the care action). The evaluation index calculation unit 22 adjusts the internal parameters of the learning model, for example, by backpropagation. The evaluation index calculation unit 22 creates a trained model by determining, as a trained model, a model whose internal parameters have been adjusted so as to satisfy a predetermined termination condition. The predetermined termination condition here is, for example, a condition in which the probability that the output value output from the learning model and the output of the learning dataset match is equal to or greater than a predetermined threshold.

[0037] The evaluation index calculation unit 22 calculates the similarity using the created trained model. The evaluation index calculation unit 22 inputs evaluation target information to the trained model and acquires a value output from the trained model. The trained model outputs information estimating an output (a label indicating whether the caregiving action is good or bad) for the input (evaluation target information). When the trained model outputs a probability that the caregiving action is estimated to be good, the evaluation index calculation unit 22 regards the probability as the similarity with a waveform group of training data to which a label indicating that the caregiving action is good has been assigned.

[0038] The calculation result output unit 23 outputs the evaluation index (information indicating the waveform and / or similarity of the basic information and the evaluation target information). The calculation result output unit 23 outputs, for example, image information indicating the evaluation index to the display unit 26, and displays the waveform on the display screen of the display unit 26. This allows the waveform and similarity of the basic information and the evaluation target information to be visually recognized by the caregiver 40 or the like. Alternatively, the calculation result output unit 23 may output the evaluation index to a printer (not shown) or the like, thereby printing out the waveform and similarity indicating the evaluation index. Alternatively, the calculation result output unit 23 may output the evaluation index to a speaker (not shown) or the like, thereby outputting the similarity as a voice, an alarm sound, or the like.

[0039] The device control unit 24 comprehensively controls the evaluation device 20. For example, the device control unit 24 outputs the sensor information acquired by the sensor information acquisition unit 21 to the evaluation index calculation unit 22. The device control unit 24 stores the evaluation target information selected by the evaluation index calculation unit 22 as evaluation target information 250 in the storage unit 25. The device control unit 24 stores the basic information selected by the evaluation index calculation unit 22 as reference information 251 in the storage unit 25. The device control unit 24 stores information indicating the evaluation index calculated by the evaluation index calculation unit 22 as evaluation index information 252 in the storage unit 25. The device control unit 24 outputs the information indicating the evaluation index to the calculation result output unit 23.

[0040] The storage unit 25 stores programs for executing various processes of the evaluation device 20 and temporary data used when performing various processes. The storage unit 25 is configured by a storage medium, for example, a hard disk drive (HDD), a flash memory, an electrically erasable programmable read only memory (EEPROM), a random access read / write memory (RAM), a read only memory (ROM), or an arbitrary combination of these storage media. The storage unit 25 includes, for example, evaluation target information 250, reference information 251, and evaluation index information 252. The reference information 251 is information that is a reference for evaluation, and is one or more pieces of sensor information selected from a plurality of pieces of sensor information (see FIG. 4). The evaluation target information 250 is information that is the subject of evaluation, and is one or more pieces of sensor information selected from a plurality of pieces of sensor information excluding the sensor information that is the reference information (see FIG. 3). The evaluation index information 252 is information that is an index for evaluation. The evaluation here is an evaluation on the quality of the assistance action.

[0041] The display unit 26 is a functional unit that displays an image. The display unit 26 includes, for example, a display such as a liquid crystal screen. The display unit 26 acquires, for example, image information output from the calculation result output unit 23, and displays the image based on the acquired information.

[0042] FIG. 3 is a diagram showing an example of the configuration of evaluation target information 250 in an embodiment. Evaluation target information 250 includes items such as a measurement ID, a caregiver ID, and measurement data. The measurement ID is information that uniquely identifies the measurement target information. The caregiver ID is information that uniquely identifies the caregiver 40 who performed the assisting action corresponding to the evaluation target information. The measurement data is information indicating a time series change in the load measured as the evaluation target information. The measurement data includes items such as time and measurement value. The time is information indicating the time when the load was measured. The measurement value includes items for each of sensors 10-1 to 10-4. Sensor 10-1 is information indicating the load measured by load sensor 10-1. Sensor 10-2 is information indicating the load measured by load sensor 10-2. The same applies to sensors 10-3 and 10-4.

[0043] FIG. 4 is a diagram showing an example of the configuration of the reference information 251 in the embodiment. The reference information 251 includes items such as a reference ID, an assistance type, and reference data. The reference ID is information that uniquely identifies the reference information. The assistance type is information indicating the type of assistance motion corresponding to the reference information, and in this example, it is indicated that it is assistance for turning over. The assistance type may include more detailed information. For example, the assistance type may include information indicating whether the turning over is on the left or right side. The reference data is information indicating a time series change in the load measured as the reference information. The reference data includes items such as time and measurement value. The time is information indicating the time when the load was measured. The measurement value includes items for each of the sensors 10-1 to 10-4. The sensor 10-1 is information indicating the load measured by the load sensor 10-1. The sensor 10-2 is information indicating the load measured by the load sensor 10-2. The same applies to the sensors 10-3 and 10-4.

[0044] 5 and 6 are diagrams showing an example of evaluation target information 250 in an embodiment. The horizontal axis of FIG. 5 and FIG. 6 indicates time T, and the vertical axis indicates load W. In FIG. 5 and FIG. 6, key A indicates a waveform measured by a sensor installed at the top end of the left side (i.e., load sensor 10-1). Key B indicates a waveform measured by a sensor installed at the top end of the right side (i.e., load sensor 10-4). Key C indicates a waveform measured by a sensor installed at the bottom end of the left side (i.e., load sensor 10-2). Key D indicates a waveform measured by a sensor installed at the bottom end of the right side (i.e., load sensor 10-3).

[0045] Fig. 5 shows sensor information (HD1) corresponding to the motion of assisting a caregiver 40 to turn to the left, and sensor information (HD2) corresponding to the motion of assisting a caregiver 40 to turn to the left. Fig. 6 shows sensor information (HD3) corresponding to the motion of assisting a caregiver 40 to turn to the left, and sensor information (HD4) corresponding to the motion of assisting a caregiver 40 to turn to the left.

[0046] 5 and 6, for example, the waveforms of symbols HD1 and HD3 have in common the fact that the load W of legend A and legend C increases at the beginning. However, in the waveform of symbol HD1, a state in which the load W increases and then decreases is observed. In the waveform of symbol HD3, a state in which the load W increases and then decreases is not observed. The state in which the load increases and then decreases as shown in symbol HD1 may be, for example, when the caregiver 40 adjusts the position of the assisted person 30 lying on the bed B by lifting the assisted person 30. Since this type of assistance, such as the caregiver 40 lifting the assisted person 30, places a burden on the caregiver 40, it is desirable to avoid this type of assistance as much as possible. On the other hand, in symbol HD3, the load is stable, and it is considered that the caregiver 40 did not lift the assisted person 30 in the middle of the process, and the caregiver 40 provided smooth assistance. In this way, even when the same type of assistance is provided, the characteristics of the sensor information may differ, and by utilizing this difference, it is possible to calculate an index for evaluating the assistance action.

[0047] Fig. 7 is information showing an example of sensor information in an embodiment. The horizontal axis of Fig. 7 indicates time T, and the vertical axis indicates load W. Legend A of Fig. 7 indicates waveforms measured by a sensor installed at the upper left end (i.e., load sensor 10-1), legend B indicates waveforms measured by a sensor installed at the upper right end (i.e., load sensor 10-4), legend C indicates waveforms measured by a sensor installed at the lower left end (i.e., load sensor 10-2), and legend D indicates waveforms measured by a sensor installed at the lower right end (i.e., load sensor 10-4). As shown in Fig. 7, for example, the sensor information is information showing time-series changes in loads measured by each load sensor 10.

[0048] 8 to 11 are diagrams showing examples of evaluation index information 252 in an embodiment. The horizontal axis of FIG. 8 indicates time T, and the vertical axis indicates load W. The horizontal axis of FIG. 9 indicates time T, and the vertical axis indicates combined load GW. The combined load GW is information indicating the load of one waveform representative of a plurality of sensor information. Legend A in FIG. 8 and FIG. 9 indicates waveforms measured by a sensor installed at the upper left end (i.e., load sensor 10-1), legend B indicates a sensor installed at the upper right end (i.e., load sensor 10-3), legend C indicates a sensor installed at the lower left end (i.e., load sensor 10-2), and legend D indicates a sensor installed at the lower right end (i.e., load sensor 10-4).

[0049] 8, for example, evaluation index information 252 is information showing an image in which a waveform of reference sensor information (reference information 251) and a waveform of evaluation target sensor information (evaluation target information 250) are displayed side by side. As shown in this example, the time required for assistance (reference numerals 251T and 250T) may be displayed as evaluation index information 252.

[0050] As shown in FIG. 9, for example, evaluation index information 252 is information showing an image in which one waveform (symbol 251Z) representative of the reference sensor information and one waveform (symbol 250Z) representative of the sensor information to be evaluated are displayed superimposed on each other.

[0051] 10 shows a schematic diagram of a process of calculating the similarity as the evaluation index information 252. As shown in this example, for example, the evaluation index calculation unit 22 calculates an n-dimensional vector v=(a, b, c, d) based on one waveform (reference symbol 251Z) representative of the sensor information to be used as a reference, or a representative value of the load (points a to d in this example) extracted from the waveform (reference symbol 251Z). The evaluation index calculation unit 22 also calculates an n-dimensional vector t=(a#, b#, c#, d#) based on one waveform (reference symbol 250Z) representative of the sensor information to be evaluated, or a representative value of the load (points a# to d# in this example) extracted from the waveform (reference symbol 250Z). Then, the evaluation index calculation unit 22 calculates the cosine similarity of the n-dimensional vectors v and t as the evaluation index information 252.

[0052] FIG. 11 shows a schematic diagram of a process of calculating the similarity using a trained model as the evaluation index information 252. As shown in this example, for example, the evaluation index calculation unit 22 generates a learning data set in which a label is added to the waveforms (reference numerals 251Z-1, 251Z-2, ...) of a plurality of pieces of sensor information to be used as a reference. The label is information indicating the quality or poorness of the assisting action corresponding to the waveform. As shown in this example, the label may be information indicating a judgment result corresponding to a plurality of levels such as good, slightly good, normal, and poor, instead of the binary values ​​of good and bad. The evaluation index calculation unit 22 creates a trained model M2 by having the trained model M1 machine-learn the generated learning data set. The evaluation index calculation unit 22 sets the output (in this example, the evaluation estimated value: goodness 83%) obtained by inputting the waveform (reference numeral 250Z) of the sensor information to be evaluated into the trained model M2 as the similarity.

[0053] 12 is a sequence diagram showing the flow of processing performed by the evaluation system 1 in the embodiment. The processing performed by the evaluation system 1 includes a process of acquiring sensor information (step S100) and a process of calculating an evaluation index using the acquired sensor information (step S200). For example, the process of acquiring sensor information (step S100) is repeatedly performed, and when a plurality of pieces of sensor information are accumulated, a process of calculating an evaluation index (step S200) is performed.

[0054] In step S100, first, the caregiver 40 provides assistance (step S10). In this case, for example, a caregiver ID indicating identification information of the caregiver 40 is notified to the load sensor 10. The load sensor 10 periodically measures the load acting on the floor surface from the bed B during the period in which the caregiver 40 provides assistance (step S11). The load sensor 10 transmits the measured load to the evaluation device 20 in association with the caregiver ID, etc. The evaluation device 20 periodically and repeatedly receives the load measured by the load sensor 10 to acquire sensor information (step S12).

[0055] In step S200, the evaluation device 20 selects evaluation target information 250 from the multiple sensor information (step S20). For example, the evaluation device 20 selects one or more pieces of sensor information as evaluation target information 250 from the multiple sensor information excluding the sensor information selected as the reference information. Next, the evaluation device 20 selects reference information 251 from the multiple sensor information (step S21). For example, the evaluation device 20 selects one or more pieces of sensor information as reference information 251 from the multiple sensor information corresponding to the assisting action performed by the expert. The evaluation index calculation unit 22 calculates an evaluation index using the selected reference information 251 and the evaluation target information 250 (step S22).

[0056] In the above, the case where the evaluation target information 250 is selected and then the reference information 251 is selected has been described as an example, but the present invention is not limited to this. The evaluation device 20 may select the evaluation target information 250 after selecting the reference information 251.

[0057] As described above, the evaluation system 1 of the embodiment is an evaluation index calculation system that calculates an evaluation index for evaluating the action of the caregiver 40 in assisting the change of position of a human or doll on the bed B (an example of a "bed"). The evaluation system 1 includes a sensor information acquisition unit 21 and an evaluation index calculation unit 22. The sensor information acquisition unit 21 acquires sensor information. The sensor information is information indicating a time-series change in the load acting on the floor surface from each of a plurality of points on the bed during a period in which the caregiving action was performed. The evaluation index calculation unit 22 selects, from the plurality of pieces of sensor information acquired by the sensor information acquisition unit 21, reference information 251 as a basis for evaluation and evaluation target information 250 different from the reference information. The evaluation index calculation unit 22 calculates evaluation index information 252 using the selected reference information 251 and evaluation target information 250. The evaluation index information 252 is information for evaluating the assistance corresponding to the evaluation target information 250.

[0058] In the embodiment, the sensor information acquisition unit 21 acquires sensor information indicating a time series change in the load acting on the floor surface from the bed B, measured by the load sensors 10 provided at multiple locations on the bed B during the period when the assisting action was performed. This makes it possible to acquire quantitative information that allows the assisting action to be evaluated. Moreover, the configuration in which the load sensors 10 are provided on the bed B is a configuration that has been used conventionally to acquire bio-information of the person being assisted 30. Therefore, without providing dedicated equipment, existing equipment can be utilized to easily acquire information that allows the assisting action to be evaluated.

[0059] In the embodiment, the sensor information acquisition unit 21 acquires sensor information from the load sensors 10-1 to 10-4 arranged between the floor surface and a plurality of legs of the bed B. This makes it possible to acquire information that allows for more accurate evaluation of an assisting action in which different loads can be measured on the left and right, such as turning over from side to side.

[0060] In the evaluation system 1 of the embodiment, the evaluation index calculation unit 22 calculates, as the evaluation index information, an image (of a waveform) showing the reference information 251 and the evaluation target information 250 in a manner that allows them to be compared with each other. Alternatively, the evaluation index calculation unit 22 calculates, as the evaluation index information, an image showing one waveform (an example of a "representative value of reference information") representative of a plurality of pieces of sensor information used as a reference, and one waveform (an example of a "representative value of evaluation target information") representative of a plurality of pieces of sensor information used as an evaluation target, in a manner that allows them to be compared with each other. This allows the evaluation system 1 of the embodiment to allow the caregiver 40 or the like to recognize the difference between the assistance movement and the sample movement.

[0061] In the evaluation system 1 of the embodiment, the evaluation index calculation unit 22 calculates the similarity (one example of the "degree of similarity") between the reference information 251 and the evaluation target information 250 as the evaluation index information. This allows the evaluation system 1 of the embodiment to allow the caregiver 40 to recognize the similarity between the care movement and the sample movement.

[0062] Furthermore, in the evaluation system 1 of the embodiment, the evaluation index calculation unit 22 inputs the evaluation target information 250 to the trained model, and calculates an output value output from the trained model as evaluation index information. The trained model is a model that has learned the input / output relationship in a learning dataset in which the evaluation target information 250 is input and a label (an example of an "evaluation value") indicating the quality of the assistance corresponding to the evaluation target information 250 is output. As a result, in the evaluation system 1 of the embodiment, it is possible for the caregiver 40 or the like to recognize the quality of the assistance action estimated using the trained model.

[0063] The evaluation system 1 and the evaluation device 20 in the above-mentioned embodiment may be realized in whole or in part by a computer. In that case, a program for realizing this function may be recorded in a computer-readable recording medium, and the program recorded in the recording medium may be read into a computer system and executed to realize the function. The term "computer system" as used herein includes hardware such as an OS and peripheral devices. The term "computer-readable recording medium" refers to portable media such as flexible disks, optical magnetic disks, ROMs, and CD-ROMs, and storage devices such as hard disks built into a computer system. The term "computer-readable recording medium" may also include a medium that dynamically holds a program for a short period of time, such as a communication line when a program is transmitted via a network such as the Internet or a communication line such as a telephone line, and a medium that holds a program for a certain period of time, such as a volatile memory inside a computer system that is a server or client in that case. The above-mentioned program may be a program for realizing a part of the above-mentioned function, or may be a program that can realize the above-mentioned function in combination with a program already recorded in the computer system, or may be a program that is realized using a programmable logic device such as an FPGA.

[0064] Although an embodiment of the present invention has been described in detail above with reference to the drawings, the specific configuration is not limited to this embodiment, and designs that do not deviate from the gist of the present invention are also included. [Explanation of symbols]

[0065] 1... evaluation system (evaluation index calculation system), 10... load sensor, 20... evaluation device (evaluation index calculation device), 21... sensor information acquisition unit, 22... evaluation index calculation unit, 23... calculation result output unit, 24... device control unit, 25... storage unit, 250... evaluation target information, 251... reference information, 252... evaluation index information, 26... display unit

Claims

1. The sensor information acquisition unit acquires sensor information corresponding to an assisting action for assisting a change in position of the person being assisted who is on the bed; The evaluation index calculation unit calculates evaluation index information for evaluating assistance corresponding to the evaluation target information using reference information, which is the sensor information used as a standard for evaluation, and evaluation target information, which is the sensor information used as an evaluation target; the sensor information is information indicating a time series change in a load acting from the bed to a floor surface, the change being measured by load sensors provided at a plurality of locations on the bed in response to the care action; The reference information is the sensor information that does not include a process in which the load is reduced in response to a motion of lifting up the person being assisted during the course of assistance. Evaluation index calculation method.

2. the sensor information acquisition unit acquires the sensor information from load sensors respectively disposed between a plurality of legs of the bed and a floor surface; The evaluation index calculation method according to claim 1 .

3. The sensor information acquisition unit acquires a plurality of pieces of sensor information corresponding to each of a plurality of types of assistance, an evaluation index calculation unit selects the reference information and the evaluation target information from the plurality of pieces of sensor information acquired by the sensor information acquisition unit; The evaluation index calculation method according to claim 1 .

4. The evaluation index calculation unit calculates, as the evaluation index information, an image showing the reference information and the evaluation target information in a manner that allows comparison between them, or an image showing a representative value of the reference information and a representative value of the evaluation target information in a manner that allows comparison between them, The representative value is a value that represents the measurement values ​​by the multiple load sensors included in the sensor information. The evaluation index calculation method according to claim 1 .

5. The evaluation index calculation unit calculates, as the evaluation index information, a degree of similarity between the reference information and the evaluation target information, or a degree of similarity between a representative value of the reference information and a representative value of the evaluation target information; The representative value is a value that represents the measurement values ​​by the multiple load sensors included in the sensor information. The evaluation index calculation method according to claim 1 .

6. The evaluation index calculation unit calculates, as the evaluation index information, an output value output from the trained model by inputting the evaluation target information into the trained model; The trained model is a model that has learned the input / output relationship in a training dataset in which the reference information is input and an evaluation value of assistance corresponding to the reference information is output. The evaluation index calculation method according to claim 1 .

7. The evaluation index calculation unit calculates, as the evaluation index information, an output value output from the trained model by inputting a representative value of the evaluation target information to the trained model; the representative value is a value that represents the measurement values ​​obtained by the multiple load sensors included in the sensor information, The trained model is a model that has learned the input / output relationship in a training dataset in which a representative value of the reference information is input and an evaluation value of assistance corresponding to the representative value of the reference information is output. The evaluation index calculation method according to claim 1 .

8. a sensor information acquiring unit that acquires sensor information corresponding to an assisting action for assisting a person being assisted on a bed to change his / her position; an evaluation index calculation unit that calculates evaluation index information for evaluating assistance corresponding to the evaluation target information using reference information, which is the sensor information used as a standard for evaluation, and evaluation target information, which is the sensor information used as an evaluation target; Equipped with the sensor information is information indicating a time series change in a load acting from the bed to a floor surface, the change being measured by load sensors provided at a plurality of locations on the bed in response to the care action; The reference information is the sensor information that does not include a process in which the load is reduced in response to a motion of lifting up the person being assisted during the course of assistance. Evaluation index calculation device.

9. A computer that is an evaluation index calculation device, a sensor information acquiring means for acquiring sensor information corresponding to an assisting action for assisting a person being assisted on a bed to change his / her position; an evaluation index calculation means for calculating evaluation index information for evaluating assistance corresponding to the evaluation target information, using reference information, which is the sensor information used as a standard for evaluation, and evaluation target information, which is the sensor information used as an evaluation target; A program that functions as the sensor information is information indicating a time series change in a load acting from the bed to a floor surface, the change being measured by load sensors provided at a plurality of locations on the bed in response to the care action; The reference information is the sensor information that does not include a process in which the load is reduced in response to a motion of lifting up the person being assisted during the course of assistance. program.

Images