Fall risk estimation method, fall risk estimation program, and fall risk estimation system

Abstract

A fall risk estimation method comprising: a step for extracting a feature quantity of movement of an evaluation target portion of a subject from motion data of the subject when the subject is moving the evaluation target portion in a state in which the center of gravity of the subject is stable within a predetermined range; a step for calculating a variation index of the feature quantity; and a step for estimating a fall risk of the subject on the basis of the variation index.

Description

Fall risk estimation method, fall risk estimation program, and fall risk estimation system

[0001] The present disclosure relates to a fall risk estimation method, a fall risk estimation program, and a fall risk estimation system.

[0002] As described in Patent Document 1, a method is known in which the mobility ability of a care recipient is evaluated based on a moving video of the care recipient's movement movements, and the level of performance of the care recipient's daily activities is determined based on the results of the mobility ability evaluation.

[0003] Japanese Patent Application Laid-Open No. 2022-36573

[0004] When filming the walking and other movement of a subject to be evaluated in order to assess the risk of falling, having the subject perform the movement is a burden on the subject, and there is also the risk that the subject may fall while performing the movement.

[0005] An object of the present disclosure is to provide a fall risk estimation method, a fall risk estimation program, and a fall risk estimation system that can estimate a subject's risk of falling without imposing a significant burden on the subject.

[0006] (1) A fall risk estimation method according to one embodiment of the present disclosure includes the steps of: extracting features of the movement of a subject's part to be evaluated from motion data of the subject when the subject is moving the part to be evaluated while the subject's center of gravity is stable within a predetermined range; calculating a variability index of the features; and estimating the subject's risk of falling based on the variability index.

[0007] (2) The fall risk estimation method described in (1) above may further include a step of acquiring an image capturing the subject's movements and analyzing the subject's movement data from the image.

[0008] (3) In the step of extracting the feature of the fall risk estimation method described in (1) or (2) above, the period in which the subject's evaluation target part reaches a predetermined state may be calculated as the feature of the movement of the subject's evaluation target part.

[0009] (4) In the step of extracting the feature of the fall risk estimation method described in any one of (1) to (3) above, the amount of movement of the subject's evaluation target part may be calculated as the feature of the movement of the subject's evaluation target part.

[0010] (5) In the step of extracting the feature quantities in the fall risk estimation method described in any one of (1) to (4) above, multiple evaluation target parts of the subject are selected, and feature quantities of the movement of each of the multiple evaluation target parts of the subject are extracted from a video of the subject simultaneously moving the multiple evaluation target parts; and in the step of calculating the variability index of the feature quantities, each variability index may be calculated from the feature quantities of the movement of each of the multiple evaluation target parts.

[0011] (6) In the fall risk estimation method described in any one of (1) to (5) above, the state in which the subject's center of gravity is stable within a predetermined range may be a state in which the subject is seated.

[0012] (7) A fall risk estimation program according to one embodiment of the present disclosure causes a processor to execute the steps of extracting features of the movement of the subject's part to be evaluated from motion data of the subject when the subject is moving the part to be evaluated while the subject's center of gravity is stable within a predetermined range, calculating a variability index of the features, and estimating the subject's risk of falling based on the variability index.

[0013] (8) A fall risk estimation system according to an embodiment of the present disclosure includes a motion data acquisition device, an estimation device, and a display device. The motion data acquisition device acquires motion data of a subject when the subject moves a body part to be evaluated while the subject's center of gravity is stable within a predetermined range. The estimation device extracts feature quantities of the movement of the body part to be evaluated from the motion data of the subject, calculates a variability index for the feature quantities, and estimates the subject's fall risk based on the variability index. The display device displays the estimated result of the subject's fall risk.

[0014] According to a fall risk estimation method, a fall risk estimation program, and a fall risk estimation system according to an embodiment of the present disclosure, the subject's risk of falling is estimated while reducing the burden on the subject.

[0015] 1 is a block diagram showing an example configuration of a fall risk estimation system according to the present disclosure; FIG. 2 is a schematic diagram showing an example configuration of a fall risk estimation system; FIG. 3 is a diagram showing an example image capturing the movement of a seated subject; FIG. 4 is a diagram showing an example image capturing the movement of a subject on the walking surface of a treadmill; FIG. 5 is a diagram showing an example image capturing the face of a subject; FIG. 6 is a graph showing a coefficient of variation, which is an example of an index of variability in the movement of a subject; and FIG. 7 is a flowchart showing an example procedure of a fall risk estimation method according to the present disclosure.

[0016] As a comparative example, as in the invention described in Patent Document 1, it is conceivable to estimate the subject's risk of falling by capturing footage of a subject, such as a care recipient, actually walking and analyzing the subject's walking motion. However, having the subject walk increases the subject's risk of falling and being injured. Thus, it is necessary to estimate the subject's risk of falling without having the subject perform unstable movements such as walking, which pose a risk to the subject.

[0017] According to the fall risk estimation system and fall risk estimation method of the present disclosure, a subject's risk of falling is estimated based on the results of capturing a video of the subject's physical movements, such as the subject's hands or arms, or feet or legs, while the subject is in a stable posture, such as a seated posture, or the results of the subject's movements being detected by a sensor. By having the subject perform movements in a stable posture, the subject's risk of falling is estimated without worrying about the subject being injured. In this disclosure, "subject's hands" refers to the part of the subject from the wrist onwards. "subject's arms" refers to the part of the subject from the shoulder onwards. "subject's feet" refers to the part of the subject from the ankle onwards. "subject's legs" refers to the part of the subject from the hip joint onwards.

[0018] Furthermore, in the comparative example, when the subject actually walks, the walking distance is limited, which may prevent sufficient data from being obtained about the subject's walking motion. On the other hand, with the fall risk estimation system and fall risk estimation method according to the present disclosure, the subject can perform the motion with a stable posture, without limiting the walking distance, and the subject can perform the motion for a longer period of time than walking. Therefore, sufficient data about the subject's motion can be obtained. As a result, the accuracy of estimating the subject's fall risk can be improved.

[0019] Furthermore, in the comparative example, when capturing a subject's walking motion, it is necessary to capture the subject from the side in order to analyze the walking motion with high accuracy. On the other hand, according to the fall risk estimation system and fall risk estimation method disclosed herein, the subject's fall risk is estimated based on the variability in the movement of a part of the subject's body, such as the subject's hand or arm, or foot or leg, i.e., the part to be evaluated, or a variability index representing the variability. The variability in movement can be analyzed based on an image of the subject, regardless of the direction from which the subject is captured. Therefore, the subject's fall risk is estimated based on an image obtained by simple capture, such as with a mobile device. In other words, data for estimating the subject's fall risk is easily obtained. As a result, the convenience of the fall risk estimation system is improved.

[0020] In the present disclosure, a subject whose fall risk is to be estimated may be, for example, a care recipient. The subject is not limited to a care recipient and may be a person with various other attributes.

[0021] In the present disclosure, a subject's fall risk is an index representing the possibility of the subject falling. A subject's fall refers to the subject unintentionally putting their hands or knees on the floor or striking their upper body on the floor while standing. The subject's fall risk is expressed as the probability of the subject falling while walking. The subject's fall risk may include the probability of the subject falling from a standing state before the subject starts walking, or the probability of the subject falling when the subject stands up from a sitting state. The subject's fall risk may be expressed taking into account various factors that may trigger the subject's fall. The subject's fall risk may be expressed as the probability of the subject falling for each factor, or as the probability of the subject falling regardless of the factor.

[0022] In the present disclosure, a stable posture of a subject refers to a posture in which the subject's center of gravity is maintained within a predetermined range and is stable. The subject's center of gravity may vary depending on the subject's physical condition. The subject's center of gravity may be located near the thoracic spine when the subject is sitting with the buttocks in contact with the seat. The subject's center of gravity may be located near the pelvis when the subject is standing.

[0023] The state in which the subject's center of gravity is stable within the predetermined range includes, for example, a state in which the subject is seated. In this case, the subject is not moving, so the center of gravity does not move and remains within the predetermined range.

[0024] The state in which the subject's center of gravity is stable within a predetermined range includes, for example, a state in which the subject is walking on a treadmill while being supported around the upper body or waist. In this case, the subject is walking but is not moving as seen from outside the treadmill. Therefore, the subject's center of gravity does not move and remains within the predetermined range.

[0025] The state in which the subject's center of gravity is stable within a predetermined range is not limited to the above examples and can be realized in various ways. The predetermined range may be, for example, a range from the center of gravity when the subject is stationary to the body surface in the horizontal direction. The predetermined range may be, for example, a range in the vertical direction in which the center of gravity moves up and down when the subject performs a walking motion. The predetermined range is not limited to these examples and may be determined as appropriate.

[0026] A fall risk estimation method according to the present disclosure will be described below with reference to the drawings. In particular, the description will focus on one embodiment of the fall risk estimation method, which includes the steps of acquiring images of a subject's movements and analyzing the subject's movement data from the images. A fall risk estimation system 1 (see FIG. 1 ) capable of implementing the fall risk estimation method will also be described.

[0027] (Configuration example of fall risk estimation system 1) As shown in Fig. 1 and Fig. 2 , a fall risk estimation system 1 according to one embodiment includes an estimation device 10, a photographing device 20, and a display device 30. The estimation device 10, the photographing device 20, and the display device 30 are communicatively connected via a network 40. At least some of the estimation device 10, the photographing device 20, and the display device 30 may be communicatively connected via P2P (Peer to Peer) without via the network 40. The estimation device 10, the photographing device 20, and the display device 30 may be communicatively connected in part via the network 40, and communicatively connected in other parts via P2P.

[0028] Below, an example of the configuration of each part of the fall risk estimation system 1 will be described.

[0029] 1 , the estimation device 10 includes a data receiving unit 11, a processing unit 12, a storage unit 16, and a data transmitting unit 17. The estimation device 10 is connected to an imaging device 20 via the data receiving unit 11, and is connected to a display device 30 via the data transmitting unit 17. The processing unit 12 includes a skeleton extracting unit 13, a feature extracting unit 14, and an estimation unit 15.

[0030] The data receiving unit 11 may be configured to include a communication module configured to be able to receive information, data, etc. from other devices such as the imaging device 20. The data transmitting unit 17 may be configured to include a communication module configured to be able to transmit information, data, etc. to other devices such as the display device 30. The data receiving unit 11 and the data transmitting unit 17 may be configured as an integrated unit. The integrated configuration of the data receiving unit 11 and the data transmitting unit 17 is also referred to as a data communication unit.

[0031] The communication module may be configured to be able to communicate with other devices via the network 40, or may be configured to be able to communicate with other devices via P2P (Peer to Peer) without via the network 40. The communication module may be configured to be able to communicate with other devices via wired or wireless communication. The communication module may be compatible with mobile communication standards such as 4G (4th Generation) or 5G (5th Generation). The communication module may be compatible with communication standards such as LAN (Local Area Network). The communication module may be compatible with wired or wireless communication standards. The communication module is not limited to these and may be compatible with various communication standards.

[0032] The processing unit 12 may be configured to include one or more processors. The processor may include, but is not limited to, a general-purpose processor or a dedicated processor specialized for a specific process. The processing unit 12 may be configured to include one or more dedicated circuits. The dedicated circuits may include, for example, an FPGA (Field-Programmable Gate Array) or an ASIC (Application Specific Integrated Circuit). The processing unit 12 may be configured to include a dedicated circuit instead of a processor, or may be configured to include a dedicated circuit together with a processor.

[0033] The memory unit 16 may store information or data used by the processing unit 12, or system programs or application programs executed by the processing unit 12. The memory unit 16 may include, for example, a semiconductor memory, a magnetic memory, or an optical memory. The memory unit 16 may include an electromagnetic storage medium such as a magnetic disk. The memory unit 16 may include a non-transitory computer-readable medium. The memory unit 16 may function as, for example, a main storage device, an auxiliary storage device, or a cache memory. The memory unit 16 may function as a work memory for the processing unit 12. The memory unit 16 may be configured integrally with the processing unit 12 or may be configured separately from the processing unit 12.

[0034] The estimation device 10 may be configured to include at least one server. The estimation device 10 may be realized as a computer such as a desktop personal computer (PC), a notebook PC, or a tablet PC. The estimation device 10 may be realized as a mobile terminal such as a smartphone or a tablet. The estimation device 10 may be realized as a wearable device such as a smartwatch. The estimation device 10 is not limited to these examples and may be realized as various types of equipment.

[0035] <Photographing device 20> The photographing device 20 photographs the photographing target and outputs the photographed image to the estimation device 10. The photographing device 20 may be configured to include a camera. The camera may be configured to include an imaging element that performs photoelectric conversion, such as a CCD (Charge Coupled Device) or CMOS (Complementary Metal Oxide Semiconductor) sensor. The camera may be a visible light camera or various other cameras such as an infrared camera. The photographing device 20 may be configured to include a depth camera that can measure three-dimensional data such as point cloud data of the photographing target.

[0036] The image capturing device 20 may be built into a computer such as a desktop PC, a notebook PC, or a tablet PC, or a mobile terminal such as a smartphone or a tablet. The image capturing device 20 may be built into a wearable device such as a smartwatch. The image capturing device 20 is not limited to these examples and may be realized in various forms.

[0037] <Display Device 30> The display device 30 displays information, data, etc. output from the estimation device 10. The display device 30 may be configured to include various displays, such as a liquid crystal display, an organic EL (Electro-Luminescence) or inorganic EL display, or an LED (Light Emitting Diode) display, so as to be able to display information, data, etc. as characters, symbols, images, etc.

[0038] The display device 30 may be built into a computer such as a desktop PC, a notebook PC, or a tablet PC, or a mobile terminal such as a smartphone or a tablet. The display device 30 may be built into a wearable device such as a smartwatch. The display device 30 is not limited to these examples and may be realized in various forms.

[0039] <Other Configuration Examples> The estimation device 10 may further include an input device that accepts input of information, data, etc. from a user. The input device may include, for example, a touch sensor or a pointing device such as a mouse. The input device may include physical keys. The input device may include a voice input device such as a microphone. The input device may include a display device that displays the content of the accepted input. The input device may be configured as a touch display in which a display device and a touch panel are integrated. The estimation device 10 may be configured to be connectable to an external input device.

[0040] The input device may include the imaging device 20. That is, the estimation device 10 may be configured integrally with the imaging device 20. In this case, one device functions as both the estimation device 10 and the imaging device 20.

[0041] The estimation apparatus 10 may further include an output device that outputs information, data, etc. The output device may include, for example, a speaker, headphones, or earphones so that the information, data, etc. can be output as sound. The estimation apparatus 10 may be configured to be connectable to an external output device.

[0042] The output device may include the display device 30. That is, the estimation device 10 may be configured integrally with the display device 30. In this case, one device functions as both the estimation device 10 and the display device 30.

[0043] The estimation device 10 may be configured integrally with the imaging device 20 and the display device 30. In this case, one device functions as the estimation device 10, the imaging device 20, and the display device 30. In other words, the entire functions of the fall risk estimation system 1 may be realized by one device. The entire functions of the fall risk estimation system 1 may be realized by, for example, a smartphone or a smartwatch.

[0044] The functions of the fall risk estimation system 1 or the estimation device 10 described above may be realized using a cloud service or may be realized in an on-premise environment.

[0045] The image capturing device 20 and the display device 30 may be integrated. In this case, the integrated functions of the image capturing device 20 and the display device 30 may be realized by a single device. The integrated functions of the image capturing device 20 and the display device 30 may be realized by, for example, a smartphone or a smartwatch. When the functions of the estimation device 10 are realized using a cloud service, the smartphone or smartwatch that realizes the integrated functions of the image capturing device 20 and the display device 30 may transmit a video of the subject to the cloud service that realizes the functions of the estimation device 10 and receive an assessment result of the subject's fall risk from the cloud service.

[0046] (Example of operation for estimating fall risk) A fall risk estimation system 1 according to the present disclosure calculates feature quantities of the movement of a subject's evaluation target body part from images captured of the subject moving the evaluation target body part while the subject's center of gravity is stable within a predetermined range. The fall risk estimation system 1 calculates a variability index of the feature quantities of the movement of the subject's evaluation target body part, and estimates the subject's fall risk based on the variability index.

[0047] As shown in Fig. 2, the photographing device 20 photographs the movements of the subject 80, who is a target, through operation by the user 70, and outputs the photographed image to the estimation device 10 via the network 40. As shown in Fig. 3A, the subject 80 is sitting with his or her buttocks in contact with the seating surface 72, and is moving his or her right knee 81a and left knee 81b up and down alternately from a state in which his or her right toes 85a and left toes 85b are in contact with the floor surface 74. Furthermore, the subject 80 is placing his or her right arm, including the right elbow 83a, and his or her left arm, including the left elbow 83b, on his or her waist or buttocks.

[0048] The estimation device 10 acquires captured images from the photographing device 20 via the data receiving unit 11, and estimates the subject's risk of falling based on the captured images of the movements of the subject 80 via the processing unit 12. The estimation device 10 may store, in the memory unit 16, videos of the subject's movements acquired via the data receiving unit 11. The estimation device 10 may store, in the memory unit 16, analysis results of the videos by the processing unit 12 or estimation results of the subject's risk of falling. The estimation device 10 outputs the estimation results of the subject's risk of falling from the data transmitting unit 17 to the display device 30. The display device 30 displays the estimation results of the subject's risk of falling.

[0049] The processing unit 12 of the estimation device 10 analyzes an image capturing the movement of the subject 80. The image capturing the movement of the subject 80 is also referred to as a subject image. An example of the operation of the processing unit 12 analyzing the subject image will be described below.

[0050] <Analysis of Subject's Movement> The skeleton extraction unit 13 of the processing unit 12 extracts skeletal data of the subject from the subject image. The skeletal data of the subject may be represented as a set of joints representing the subject's joints and bones representing the bones connecting the subject's joints. The skeletal data of the subject is not limited to the above example and may be represented in various other forms.

[0051] When the subject image is a video including a plurality of frames, the skeleton extraction unit 13 may extract skeleton data of the subject in each frame.

[0052] The feature extraction unit 14 of the processing unit 12 extracts feature amounts of the movement of at least one part of the subject's body as the evaluation target part. The feature amounts of the movement of the subject's body part may include the period of movement of the subject's body part, or the amount of movement of the subject's body part. The period of movement of the subject's body part may be the repetition period when the subject moves that part repeatedly. The amount of movement of the subject's body part represents the magnitude of the movement of the subject's body part. The amount of movement of the subject's body part may be the one-way movement distance when the subject moves that part repeatedly. The feature amounts of the movement of the subject's body part are not limited to the examples described above.

[0053] When the subject image is a video including multiple frames, the feature extraction unit 14 may analyze the movement of the subject's evaluation target part based on the subject's skeletal data in each frame, and extract features of the movement of the evaluation target part.

[0054] In this embodiment, the subject 80 moves his / her right knee 81a and left knee 81b up and down alternately. Therefore, the feature extraction unit 14 extracts a feature of the movement of at least one of the right knee 81a and the left knee 81b of the subject 80 as the evaluation target part of the subject. The feature extraction unit 14 may extract the timing at which at least one of the right knee 81a and the left knee 81b as the evaluation target part of the subject is farthest from the floor surface 74. In other words, the feature extraction unit 14 may extract the timing at which at least one of the right knee 81a and the left knee 81b as the evaluation target part of the subject reaches the highest point.

[0055] The feature extraction unit 14 extracts multiple times the timing at which at least one of the right knee 81a and the left knee 81b reaches its highest point by repeating the movements of the right knee 81a and the left knee 81b. The interval between the multiple times extracted for the subject's movement is the period of the subject's movement. The period of the subject's movement is included in the feature of the subject's movement. The feature extraction unit 14 calculates the interval between the multiple extracted times as the period of the subject's movement, i.e., the feature of the subject's movement. The feature extraction unit 14 may calculate the interval between the times at which at least one of the right knee 81a and the left knee 81b reaches its highest point as the period of the subject's movement, i.e., the feature of the subject's movement. The feature extraction unit 14 may calculate the interval between the times at which the right knee 81a reaches its highest point and the times at which the left knee 81b reaches its highest point as the period of the subject's movement, i.e., the feature of the subject's movement. In the up-and-down movement of the knee, the time during which the knee remains at its highest point is short. Therefore, it is easy to identify the timing at which the knee reaches the highest point. By calculating the interval between the timings at which the knee reaches the highest point as the feature amount of the subject's movement, the feature amount of the subject's movement can be calculated with high accuracy.

[0056] The feature extractor 14 may calculate, as a feature of the subject's movement, a period in which the evaluation target part of the subject reaches a predetermined state. In the above example, the predetermined state is a state in which the right knee 81 a or the left knee 81 b reaches its highest point. The predetermined state is not limited to the above example, and may be a state in which the right knee 81 a or the left knee 81 b reaches its lowest position, or a state in which the right toe 85 a or the left toe 85 b touches the floor surface 74.

[0057] The feature extractor 14 may calculate the amount of movement of the right knee 81 a or the left knee 81 b as the feature of the evaluation target part of the subject. The amount of movement of the right knee 81 a or the left knee 81 b may be calculated as the distance from the floor surface 74 when the right knee 81 a or the left knee 81 b reaches its highest point.

[0058] The feature extraction unit 14 may calculate the movement speed of the right knee 81a or the left knee 81b as the feature of the evaluation target part of the subject. The movement speed of the right knee 81a or the left knee 81b may be calculated by dividing the distance from the floor surface 74 to the highest point of the right knee 81a or the left knee 81b by the time it takes for the right knee 81a or the left knee 81b to reach its highest point from its lowest position. The feature extraction unit 14 may also calculate the time it takes for the right knee 81a or the left knee 81b to reach its highest point from its lowest position as the feature of the evaluation target part of the subject. The feature extraction unit 14 is not limited to these examples and may calculate the feature of the evaluation target part of the subject in various ways.

[0059] The above describes one embodiment of the fall risk estimation system 1, which extracts skeletal data from image data (i.e., the subject image) capturing the subject's movements while the subject is seated with his or her buttocks in contact with the seat surface 72, i.e., while the subject's center of gravity is stable within a predetermined range, and calculates the periodicity of the knee movements, including at least one of the subject's right knee 81 a and left knee 81 b, or the movement speed of the knees, which are the subject's evaluation target parts, as feature quantities of the subject's movements. However, as described above, the subject can also use a walking training device such as a stud mill so that the subject can perform movements while their center of gravity is stable within a predetermined range. As shown in FIG. 3B , the subject 80 can walk while their posture is stable, i.e., while their center of gravity is stable within a predetermined range, by holding onto a handrail 94 on the walking surface 92 of the treadmill 90. By extracting the subject's skeletal data in the same manner as described above from image data capturing the subject's movements in the state shown in Figure 3B, i.e., the subject image, it is possible to calculate the periodicity of the knee movements, including at least one of the right knee 81a and left knee 81b, which are the subject's evaluation target parts, or the movement speed of the knees, which are the evaluation target parts, as a characteristic of the subject's movements.

[0060] In this way, various methods can be used to achieve a state in which the subject's center of gravity is stable within a predetermined range. Among the various methods, the method of having the subject perform a movement while sitting with their buttocks in contact with the seating surface 72 is preferred as a method for estimating the subject's risk of falling, because there is no risk of the subject being injured even if they move the part of the body to be evaluated.

[0061] In addition to extracting the subject's skeletal data from the subject image, it is also possible to calculate movement features from, for example, repetitive eye movements or head movements, and then calculate a variability index from the eye or head movement feature. As shown in FIG. 3C , the movement of the subject's eyes 88, i.e., the eye movement, may be extracted from a subject image capturing the head of the subject 80. From the extracted eye movement 88, the subject's gaze feature is calculated as a feature of the subject's movement, and a variability index is calculated. The subject's risk of falling can also be estimated from the variability index. The head of the subject 80 includes the face 80F and neck 80N of the subject 80. The movement of the face 80F or neck 80N may be extracted from the subject image. In this way, the gaze feature of the subject 80 can also be calculated based on the orientation of the face 80F or the direction of the neck 80N of the subject 80. It is also possible to calculate feature amounts of the movement of the face 80F or the neck 80N and estimate the risk of the subject falling from a variation index calculated from the feature amounts of the movement of the face 80F or the neck 80N. The movements for which the variation index is calculated are not limited to these examples and may include movements of various parts of the subject's body.

[0062] <Estimating the subject's risk of falling> The greater the variation in the movement of the subject's evaluation target body part, the more likely the subject is to fall. The subject's tendency to fall can be understood as the frequency with which the subject, such as a care recipient, actually falls. The subject's tendency to fall may be understood, for example, as the number of times the subject falls within a predetermined period, or as the ratio of the number of times the subject actually falls to the number of times the subject performs walking movements, etc., that may result in a fall.

[0063] The estimation unit 15 of the processing unit 12 calculates a variation index of the feature of the movement of the evaluation target part of the subject extracted by the feature extraction unit 14 as an index representing the degree of variation in the movement of the evaluation target part of the subject. The variation index is the degree of variation from the average of the feature, and includes, for example, a variance value, a standard deviation, or a coefficient of variation. The variance value is the sum of the squares of the differences between each measurement point and the average value. The square root of the variance value is the standard deviation. The value obtained by dividing the standard deviation by the average value is the coefficient of variation.

[0064] The estimation unit 15 can estimate the risk of a subject falling based on the variation index of the feature amount of the movement of the evaluation target part of the subject. Hereinafter, an embodiment will be described in which the risk of a subject falling is estimated by calculating the coefficient of variation, which is one of the variation indexes.

[0065] The estimation unit 15 may determine that the subject has a high risk of falling when the calculated coefficient of variation is equal to or greater than a determination threshold. The determination threshold may be set based on statistics of the coefficients of variation of the feature amounts of the movements of the evaluation target parts of multiple subjects. The determination threshold is not limited to this and may be set based on various criteria.

[0066] The estimation unit 15 may set the judgment threshold to a different value depending on the state of the subject. For example, the estimation unit 15 may set a judgment threshold for determining the risk of falling to a different value in each state of the subject, for example, when the subject is just standing, when the subject is walking, or when the subject is running. The state in which the subject is walking may be a state in which at least one foot of the subject is in contact with the ground while the subject is moving. The state in which the subject is running may be a state in which both feet of the subject are not in contact with the ground while the subject is moving. The estimation unit 15 may set the judgment threshold to a different value depending on the walking speed of the subject.

[0067] The estimation unit 15 may set the judgment threshold to a different value depending on the walking environment of the subject. For example, the estimation unit 15 may set different judgment thresholds for determining the risk of falling in a flat place, a place with a slight step such as a threshold, or a place with a large step such as a staircase. The estimation unit 15 may set different judgment thresholds depending on the size of the step.

[0068] When the estimation unit 15 determines that the subject has a high risk of falling, it may cause the display device 30 to display a message that the subject has a high risk of falling. When the estimation unit 15 determines that the subject has a high risk of falling, it may cause the display device 30 to display, for example, a message that the subject needs to take preventive measures against falling. For example, when the estimation unit 15 determines that the subject has a high risk of falling while running, it may cause the display device 30 to display information urging the subject not to run. For example, when the estimation unit 15 determines that the subject has a high risk of falling while walking fast, it may cause the display device 30 to display information urging the subject to walk slowly.

[0069] The estimation unit 15 may calculate the probability of the subject falling as a value representing the subject's risk of falling based on the calculated coefficient of variation. The estimation unit 15 may calculate the probability of the subject falling depending on the state of the subject. For example, the estimation unit 15 may calculate the probability of the subject falling when the subject is walking and the probability of the subject falling when the subject is jogging. The estimation unit 15 may determine that the subject has a high risk of falling when the probability of the subject falling is equal to or greater than a probability threshold.

[0070] <Example of an action to be performed by a subject> In the above example, the subject, while sitting, performed an action of alternately moving the right knee 81a and the left knee 81b up and down without moving any part of the body other than the feet and legs including the right knee 81a and the left knee 81b. In this case, the subject was instructed to move only the feet and legs including the right knee 81a and the left knee 81b. Moving only the feet and legs including the right knee 81a and the left knee 81b is considered a single action. A single action to be performed by a subject is also referred to as a single task.

[0071] On the other hand, the subject may be instructed to move their hands or arms while moving their feet and legs, including the right knee 81 a and the left knee 81 b. Moving their hands or arms as well as their feet and legs, including the right knee 81 a and the left knee 81 b, is considered a multiple movement. The multiple movements that the subject is instructed to perform are also referred to as multiple tasks. A multiple task includes simultaneously moving the evaluation target area and another part of the evaluation target area.

[0072] When a subject performs multiple tasks, the subject's ability to concentrate on the movements of each part of the body tends to be lower than when the subject performs a single task. Therefore, when a subject performs multiple tasks, the movement of the part of the body being evaluated tends to vary more.

[0073] The imaging device 20 may capture an image of the subject simultaneously moving multiple evaluation target body parts. In other words, the imaging device 20 may capture an image of the subject performing multiple tasks. The estimation device 10 may extract feature amounts of the movement of each evaluation target body part of the subject based on images of the subject performing multiple tasks, calculate a coefficient of variation for each feature amount, and estimate the subject's fall risk based on these coefficients of variation. Extracting feature amounts of the movement of the subject's evaluation target body parts based on images of the subject performing multiple tasks can increase the coefficient of variation of the feature amounts. Increasing the coefficient of variation of the feature amounts increases the sensitivity for determining whether the subject has a high risk of falling. As a result, the estimation accuracy can be improved.

[0074] As described above, the multiple tasks may include simultaneously moving the feet and legs, including the right knee 81 a and the left knee 81 b, and the hands or arms. The multiple tasks may include, for example, moving the head while moving the feet or legs, or the hands or arms. The multiple tasks may include, for example, answering questions while moving the feet or legs, or the hands or arms. Answering questions corresponds to moving the mouth. The multiple tasks may also include, for example, changing the line of sight while moving the feet or legs, or the hands or arms.

[0075] <Calculation example of coefficient of variation> As an example of a case where a subject performed a single task, the coefficient of variation of the period in which the subject's knees rise was calculated when the subject was sitting and alternately moved their right knee 81 a and left knee 81 b up and down without moving their feet or other parts of their legs. In addition, subjects who were known to have a low risk of falling and subjects who were known to have a high risk of falling performed the single task.

[0076] An example of the calculated coefficient of variation of the knee-raising period when subjects performed a single task is shown in the bar graph in Figure 4. In the bar graph in Figure 4, the vertical axis represents the coefficient of variation. Of the two subjects who performed the single task, the bar graph for the coefficient of variation of the knee-raising period of the subject with a high risk of falling is hatched with diagonal lines. The bar graph for the coefficient of variation of the knee-raising period of the subject with a low risk of falling is filled in with white. In addition, error bars are added to the bar graphs for each coefficient of variation.

[0077] The coefficient of variation of the knee-raising period for subjects with a high risk of falling is larger than the coefficient of variation of the knee-raising period for subjects with a low risk of falling. This difference in coefficient of variation has been confirmed to be significant by statistical hypothesis testing. Therefore, the estimation result of a subject's risk of falling based on the coefficient of variation of the subject's knee-raising period is statistically significant.

[0078] As an example of a case where a subject performed multiple tasks, the coefficient of variation of the period in which the subject's knees rise was calculated when the subject was sitting and alternately moved their right knee 81 a and left knee 81 b up and down while swinging their hands or arms back and forth as other parts of their feet and legs. In addition, subjects who were known to have a low risk of falling and subjects who were known to have a high risk of falling performed the multiple tasks.

[0079] The bar graph in Figure 4 shows an example of the calculated coefficient of variation of the knee-raising period when a subject performed multiple tasks. Of the two subjects who performed multiple tasks, the bar graph for the coefficient of variation of the knee-raising period of the subject with a high risk of falling is hatched with diagonal lines. The bar graph for the coefficient of variation of the knee-raising period of the subject with a low risk of falling is white. Error bars are also added to each bar graph for the coefficient of variation.

[0080] Even when subjects performed multiple tasks, the coefficient of variation of the knee-raising period of subjects with a high risk of falling was larger than the coefficient of variation of the knee-raising period of subjects with a low risk of falling. Statistical hypothesis testing confirmed that this difference in coefficient of variation is significant. Therefore, the estimation of a subject's risk of falling based on the coefficient of variation of the subject's knee-raising period is statistically significant.

[0081] Furthermore, the difference between the coefficient of variation of the knee-raising period of a subject with a high risk of falling and the coefficient of variation of the knee-raising period of a subject with a low risk of falling is larger when the subject performs multiple tasks than when the subject performs a single task. Therefore, the accuracy of estimating a subject's fall risk based on images of a subject performing multiple tasks can be higher than the accuracy of estimating a subject's fall risk based on images of a subject performing a single task.

[0082] In the above example, when the subject performed multiple tasks, a variability index was calculated for the feature amount of the movement of one body part. Then, the subject's risk of falling was estimated based on the variability index for the feature amount of the movement of one body part. As another example, multiple evaluation target parts may be selected from the subject's body parts. When the subject simultaneously moves multiple evaluation target parts as part of multiple tasks, the feature amount of the movement of each of the multiple evaluation target parts may be extracted. Furthermore, a variability index for each of the multiple evaluation target parts may be calculated from the feature amount of the movement of each of the multiple evaluation target parts. Then, the subject's risk of falling may be estimated based on the variability index for each of the multiple evaluation target parts.

[0083] When the estimation device 10 extracts feature quantities for the movement of each evaluation target body part of the subject based on images of the subject performing multiple tasks and calculates the variability index for each feature quantity, the fall risk can be estimated from the maximum value of the variability index for each feature quantity, or the fall risk can be estimated by comprehensively assessing the variability index for each feature quantity. For example, the estimation device 10 may calculate an overall variability index by combining the variability index for the feature quantities of the subject's foot or leg movement and the variability index for the feature quantities of the subject's hand or arm movement, and estimate the subject's fall risk based on the overall variability index. The combination of the subject's body parts used to calculate the overall variability index is not limited to the combination of feet or legs and hands or arms described above, and may include various other combinations. Furthermore, when the subject performs multiple tasks, the estimation device 10 may change how the fall risk is estimated, or how the fall risk is estimated from the variability index for each feature quantity, for each subject.

[0084] <Example Flowchart> The processing unit 12 of the estimation device 10 may execute a fall risk estimation method including the steps of the flowchart illustrated in Fig. 5. The fall risk estimation method may be realized as a fall risk estimation program executed by a processor constituting the processing unit 12. The fall risk estimation program may be stored in a non-transitory computer-readable medium.

[0085] The processing unit 12 acquires a video of the subject (step S1). The processing unit 12 calculates, from the video of the subject, feature amounts of the movement of the subject's evaluation target body part (step S2). The processing unit 12 calculates a variability index of the feature amounts of the movement of the subject's evaluation target body part (step S3).

[0086] The processing unit 12 determines whether the calculated variation index is equal to or greater than the determination threshold (step S4). If the calculated variation index is equal to or greater than the determination threshold (step S4: YES), the processing unit 12 determines that the subject has a high risk of falling and outputs an alarm regarding the risk of falling (step S5). After executing the procedure of step S5, the processing unit 12 ends execution of the procedure of the flowchart in Fig. 5. If the calculated variation index is not equal to or greater than the determination threshold (step S4: NO), that is, if the calculated variation index is less than the determination threshold, the processing unit 12 ends execution of the procedure of the flowchart in Fig. 5 without executing the procedure of step S5.

[0087] The processing unit 12 may cause the display device 30 to display the estimated result of the subject's fall risk, regardless of the magnitude of the variability index.

[0088] (Summary) As described above, with the fall risk estimation system 1 according to this embodiment, the fall risk of a subject is estimated based on images of a movement performed by the subject in a stable posture. By allowing the subject to perform a movement in a stable posture, the burden on the subject is reduced. As a result, the fall risk of the subject is estimated while reducing the burden on the subject.

[0089] (Other embodiments) The above has mainly been described as a case where the subject's movement data is obtained by acquiring images of a video of the subject, and the movement data is data of the subject moving their right knee 81a and left knee 81b up and down alternately while sitting. However, other embodiments of the fall risk estimation system 1 will be described below.

[0090] <Example of Body Part to be Evaluated> In the embodiment described above, the processing unit 12 of the estimation device 10 analyzed, as the movement of the body part to be evaluated, the movement of the right knee 81 a or the left knee 81 b of the subject 80 who is the subject of photography. In another embodiment, the processing unit 12 may analyze, as the movement of the body part to be evaluated, the movement of the subject's hand or arm.

[0091] The processing unit 12 may analyze the movement of the subject's right elbow 83 a or left elbow 83 b as the movement of the evaluation target body part. For example, the processing unit 12 may calculate the period of movement of at least one of the right elbow 83 a or left elbow 83 b as the feature amount of the movement of the evaluation target body part of the subject. The processing unit 12 may calculate the movement amount or movement speed of at least one of the right elbow 83 a or left elbow 83 b as the feature amount of the movement of the evaluation target body part of the subject.

[0092] The processing unit 12 may analyze the movement using the subject's right wrist or left wrist as the evaluation target part. The processing unit 12 may analyze the movement using the subject's right ankle or left ankle as the evaluation target part. The processing unit 12 may analyze the movement using the subject's right toe 85a or left toe 85b as the evaluation target part. The processing unit 12 is not limited to these parts, and may analyze the movement using various parts of the subject's body as the evaluation target part.

[0093] <Example of Movement Performed by Subject> In the embodiment described above, the subject performs a movement in which the right knee 81 a and the left knee 81 b are alternately moved up and down while sitting. The movement performed by the subject is not limited to this.

[0094] For example, the subject may, while sitting, perform an action of moving the right knee 81 a or the left knee 81 b from side to side. The subject may, while sitting, perform an action of raising and lowering the heels while keeping the toes on the floor surface 74. The subject may, while sitting, perform an action of raising and lowering the toes while keeping the heels on the floor surface 74.

[0095] For example, while sitting, the subject may perform a motion of swinging their hands or arms back and forth, a motion of opening their hands or arms left and right, a motion of moving their hands or arms up and down, or a motion of rotating their hands or arms.

[0096] While sitting, the subject may perform an action that moves only one of the left hand or arm or the right hand or arm, or may perform an action that moves only one of the left foot or leg or the right foot or leg.

[0097] The subject may perform a movement that combines movements of multiple parts of the body while sitting. For example, the subject may perform a movement of raising and lowering their feet or legs and swinging their hands or arms. The subject may perform a movement of raising their left hand or left arm when raising their right foot or right leg, and raising their right hand or right arm when raising their left foot or left leg, similar to a walking movement. The subject may perform a movement that is not performed in a walking movement, such as raising their right hand or right arm when raising their right foot or right leg, and raising their left hand or left arm when raising their left foot or left leg.

[0098] The subject is not limited to being seated, and may perform actions that move various parts of the body as multiple tasks in other stable postures, as described above. For example, the subject may be standing with the upper body or the lower back supported and perform an action of simultaneously raising the right hand or arm and the right foot or leg, or an action of simultaneously raising the left hand or arm and the left foot or leg, as multiple tasks.

[0099] <Example of Acquisition of Subject's Motion Data> In the above-described embodiment, the estimation device 10 acquires the subject's motion data by capturing images of the subject's motions using the imaging device 20. The estimation device 10 acquires the subject's motion data from various sensors, such as an acceleration sensor, a gyro sensor, an inertial sensor, or a motion sensor. The various sensors of the estimation device 10 may be attached to the subject to acquire the subject's motion data. The various sensors of the estimation device 10 may be installed on a chair where the subject is sitting or on the floor of a room where the subject is present, and acquire the subject's motion data by capturing variations in vibration.

[0100] The estimation device 10 may extract feature amounts of the motion of the evaluation target part of the subject based on the motion data acquired from the sensor, calculate a variation index of the feature amounts, and estimate the subject's risk of falling.

[0101] The estimation device 10 may estimate the risk of falling based solely on the movement data acquired from the sensor, or may estimate the risk of falling using both the movement data detected by the sensor and the movement data analyzed from images of the subject's movements captured by the imaging device 20.

[0102] The sensor that acquires the subject's motion data and the image capturing device 20 are also collectively referred to as a motion data acquisition device.

[0103] According to the fall risk estimation system 1 of this embodiment, the fall risk of a subject is estimated based on motion data obtained when the subject performs a motion in a stable posture. By enabling the subject to perform a motion in a stable posture, the burden on the subject is reduced. As a result, the fall risk of the subject is estimated while reducing the burden on the subject.

[0104] In the fall risk estimation method and fall risk estimation system 1 disclosed herein, the subject's motion data is acquired by an imaging device as shown in Fig. 1. Using a camera mounted on a device such as a smartphone as the imaging device is particularly preferable, as this allows for the construction of an inexpensive system.

[0105] <Comparison of Subject Groups> The estimation device 10 may statistically process the results of extracting feature amounts of the movements of the evaluation target body parts of multiple subjects and output the results so that the trends of the feature amounts for each group classified according to the attributes of the subjects can be compared. The estimation device 10 may output the trends of the feature amounts for each group classified by, for example, age or gender of the subjects. The estimation device 10 may output the trends of the feature amounts for each group classified by, for example, the facility in which the subjects reside. The estimation device 10 may output the trends of the feature amounts for each group classified by, for example, whether the subjects receive care at home or in a facility. The estimation device 10 is not limited to the above examples and may classify subjects into groups according to various attributes.

[0106] <Estimation of Tripping Risk> In the above-described embodiment, the risk of a subject falling is estimated. In the above-described embodiment, a subject falling means that the subject, from a standing position, unintentionally places their hands on the floor, knees on the floor, or hits their upper body on the floor. The subject may trip even without falling. Even if the subject trips, they may be injured due to an unexpected reaction.

[0107] Therefore, the fall risk estimation system 1 according to the present disclosure may estimate not only the risk of a subject falling but also the risk of a subject tripping. Tripping can be a cause of a fall. Therefore, the risk of tripping may be included in the risk of a fall.

[0108] Although the embodiments of the present disclosure have been described based on the drawings and examples, it should be noted that those skilled in the art could make various modifications or alterations based on the present disclosure. Therefore, it should be noted that these modifications and alterations are within the scope of the present disclosure. For example, functions included in each component or step can be rearranged so as not to cause logical inconsistencies, and multiple components or steps can be combined or divided into one. Although the embodiments of the present disclosure have been described mainly in terms of an apparatus, the embodiments of the present disclosure can also be realized as a method including steps executed by each component of the apparatus. The embodiments of the present disclosure can also be realized as a method, a program executed by a processor included in the apparatus, or a storage medium on which a program is recorded. It should be understood that these are also encompassed within the scope of the present disclosure.

[0109] 1 Fall risk estimation system 10 Estimation device (11: data receiving unit, 12: processing unit, 13: skeleton extraction unit, 14: feature extraction unit, 15: estimation unit, 16: memory unit, 17: data transmission unit) 20 Imaging device 30 Display device 40 Network 70 User 72 Seating surface 74 Floor surface 80 Person to be photographed (80F: face, 80N: neck, 81a: right knee, 81b: left knee, 83a: right elbow, 83b: left elbow, 85a: right toe, 85b: left toe, 88: eyes) 90 Treadmill (92: walking surface, 94: handrail)

Claims

1. A method for estimating a risk of falling, comprising the steps of: extracting features of the movement of a part of a subject to be evaluated from motion data of the subject when the subject moves the part to be evaluated while the subject's center of gravity is stable within a predetermined range; calculating a variability index of the features; and estimating the subject's risk of falling based on the variability index.

2. A fall risk estimation method as described in claim 1, further comprising a step of acquiring an image capturing the subject's movements and analyzing the subject's movement data from the image.

3. A fall risk estimation method as described in claim 1 or 2, wherein in the step of extracting the features, a period during which the evaluation part of the subject reaches a predetermined state is calculated as a feature of the movement of the evaluation part of the subject.

4. A fall risk estimation method as described in claim 1 or 2, wherein in the step of extracting the feature, the amount of movement of the subject's evaluation target part is calculated as the feature of the movement of the subject's evaluation target part.

5. A fall risk estimation method as described in claim 1 or 2, wherein in the step of extracting the features, a plurality of evaluation target parts of the subject are selected, and features of the movement of each of the plurality of evaluation target parts are extracted from a video of the subject moving the plurality of evaluation target parts simultaneously, and in the step of calculating the variance index of the features, each variance index is calculated from the feature of the movement of each of the plurality of evaluation target parts.

6. A fall risk estimation method as described in claim 1 or 2, wherein the state in which the subject's center of gravity is stable within a specified range is a state in which the subject is seated.

7. A fall risk estimation program that causes a processor to execute the steps of: extracting features of the movement of the subject's part to be evaluated from the motion data of the subject when the subject moves the part to be evaluated while the subject's center of gravity is stable within a predetermined range; calculating a variability index of the features; and estimating the subject's risk of falling based on the variability index.

8. A fall risk estimation system comprising a motion data acquisition device, an estimation device, and a display device, wherein the motion data acquisition device acquires motion data of the subject when the subject moves a part to be evaluated with the subject's center of gravity stable within a predetermined range, the estimation device extracts features of the movement of the part to be evaluated of the subject from the motion data of the subject, calculates a variability index of the features, and estimates the subject's risk of falling based on the variability index, and the display device displays the estimated result of the subject's risk of falling.

Images