Upper body region position change estimation method

Abstract

To estimate a secular change of a position of each of multiple regions of the upper body, by measuring a secular change of force applied to each of multiple regions of the soles when an estimation object person walks.SOLUTION: A provider measurement step S11 measures a secular change of force applied to each of multiple regions of the soles when a data provider walks, and measures a secular change of a position of each of multiple regions in the upper body. A learning model creation step S12 creates a learning model by learning a relevance between the secular change of the measured force, and the secular change of the positions. An object person measurement step S13 measures the secular change of the force applied to each of the multiple regions of the soles when the estimation object person walks, by using a measurement device which is common to a measurement device used for force measurement of the data provider. An estimation step S14 applies the secular change of the measured force to the learning model, for estimating the secular change of the position of each of the multiple regions of the upper body of the estimation object person.SELECTED DRAWING: Figure 5

Description

Technical Field

[0001] The present invention relates to a method for estimating how the positions of a plurality of parts of the upper body of a person to be estimated change over time with walking, and to a method for estimating the position change of the upper body parts.

Background Art

[0002] In recent years, musculoskeletal syndrome, which is a state of requiring care or a state with a high risk of requiring care due to musculoskeletal disorders, has become a social problem. In order to detect musculoskeletal syndrome at an early stage, technological development regarding the determination of musculoskeletal syndrome has been promoted. For example, Patent Document 1 proposes a system that can easily grasp changes in the walking state of a person to be measured.

[0003] In this system, the pressure applied to each of a plurality of parts of the sole of the foot is measured over time. Feature quantities related to the walking state are acquired from the measured measurement information. Based on the acquired feature quantities, it is determined whether the walking state is a state that is a predetermined alert target. When it is determined that the walking state is a state that is a predetermined alert target, this is notified.

[0004] According to the above system, the person to be measured walks while wearing the measurement unit. Or another person such as a caregiver causes the person to be measured to wear the measurement unit and walk. Then, the person to be measured or the other person checks whether there is a notification that the state is an alert target. The person to be measured or the other person can easily grasp that the walking state of the person to be measured has changed to a state that is an alert target without the need for a direct diagnosis by a medical professional based on the above notification. Then, the person to be measured can take appropriate measures such as visiting a medical institution. Also, the other person can take appropriate measures such as causing the person to be measured to visit a medical institution.

Prior Art Documents

Patent Documents

[0005]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0006] The above Patent Document 1 aims to diagnose only the changes in the state of the lower limbs as the walking state. However, in recent years, there has been a need not only for the lower limbs but also to estimate the changes in the state of the upper body, particularly the changes in the positions of multiple parts over time.

Means for Solving the Problems

[0007] Each aspect of the method for estimating the position change of the upper body part for solving the above problems will be described. [Aspect 1] When the data provider walks, measure the change over time of the force applied to each of a plurality of parts of the sole of the data provider, and synchronously with the measurement, measure the change over time of the position of each of a plurality of parts in the upper body of the data provider. A provider measurement step; a learning model creation step of learning the correlation between the change over time of the force measured in the provider measurement step and the change over time of the position, and creating a learning model; using a measurement device common to the measurement device used for the force measurement of the data provider, when the person to be estimated walks, a subject measurement step of measuring the change over time of the force applied to each of a plurality of parts of the sole of the person to be estimated; and giving the change over time of the force measured in the subject measurement step to the learning model created in the learning model creation step, and estimating the change over time of the position of each of a plurality of parts of the upper body of the person to be estimated. An estimation step, and a method for estimating the position change of the upper body part.

[0008] When the data provider and the person to be estimated walk, the force applied to the sole is affected by the change over time of the position of each part of the upper body and changes over time. A correspondence relationship can be seen between the change over time of the force applied to each of a plurality of parts of the sole and the change over time of the position of each of a plurality of parts in the upper body.

[0009] According to the above method, for the person to be estimated whose change over time of the force on the sole is measured, the change over time of the position of each of a plurality of parts of the upper body is estimated by machine learning. In the provider measurement step, the data provider walks. At this time, the change over time of the force applied to each of a plurality of parts on the sole of the data provider's foot and the change over time of the position of each of a plurality of parts on the upper body of the data provider are measured in a synchronized state.

[0010] In the learning model creation step, the relationship between the change over time of the force measured in the provider measurement step and the change over time of the position is learned. That is, a learning model is created by finding a common feature, a law (rule), hidden in a large amount of data.

[0011] In the subject measurement step, the subject to be estimated walks. At this time, the same measuring device as the measuring device used for the force measurement of the data provider is used, and the change over time of the force applied to each of a plurality of parts on the sole of the subject to be estimated is measured.

[0012] In the estimation step, the change over time of the force measured in the subject measurement step is given as unknown data to the learning model created in the learning model creation step. Then, in the estimation step, the change over time of the position of each of a plurality of parts of the upper body of the subject to be estimated is estimated.

[0013] [Aspect 2] The method for estimating the position change of the upper body part according to [Aspect 1], wherein in the provider measurement step, the change over time of the positions of a plurality of joints is measured as a plurality of parts on the upper body.

[0014] A joint is a part where adjacent bones are connected. The bone can move in various directions with the joint as a fulcrum. A joint is a place that serves as a fulcrum when the bone moves and is the place where the amount of position change is the least when the upper body moves. Therefore, as in the above method, by measuring the change over time of the position of the joints of the upper body, it is possible to measure the change over time of the position of each of a plurality of parts of the upper body at a place where the movement of the upper body is small.

[0015] [Aspect 3] The measuring device includes an insole incorporated with a force sensor. In the provider measurement step, the data provider walks while wearing shoes with the insole laid therein, and the force sensor measures the change over time of the force applied to each of a plurality of parts of the sole. In the subject measurement step, the estimated subject walks while wearing shoes with the insole laid therein, and the force sensor measures the change over time of the force applied to each of a plurality of parts of the sole. The method for estimating the position change of the upper body part according to [Aspect 1] or [Aspect 2].

[0016] According to the above method, when measuring the force for each of a plurality of parts of the sole, shoes provided with an insole incorporated with a force sensor are used. In the provider measurement step, the data provider walks while wearing shoes. In the subject measurement step, the estimated subject walks while wearing shoes. The change over time of the force applied to each of a plurality of parts of the sole is measured by the above force sensor.

[0017] [Aspect 4] In the provider measurement step and the subject measurement step, as the insole, the force sensor is incorporated in each of the part where the heel load is applied and the part where the toe load is applied, and the force sensor is incorporated in each of the inner and outer parts with respect to the virtual line connecting the heel and the toe. The method for estimating the position change of the upper body part according to [Aspect 3], wherein the change over time of the force is measured by each force sensor.

[0018] When walking, among the plurality of parts of the sole, the heel touches the ground first, and the part that touches the ground changes in order from the heel toward the toe. Also, after the heel touches the ground and before the toe touches the ground, at least one of the inner and outer parts with respect to the virtual line connecting the heel and the toe touches the ground. The toe gripping force generated varies depending on the part that touches the ground.

[0019] Therefore, when an insole incorporated with a force sensor at the above locations is used for measuring the force on the sole, it is possible to accurately measure the change over time of the force applied to each of a plurality of parts of the sole with a small number of force sensors.

[0020] [Aspect 5] The method for estimating the position change of the upper body part according to any one of [Aspect 1] to [Aspect 4], wherein the force is pressure. As in the above method, the pressure applied to each of a plurality of parts of the sole may be regarded as the force applied to each of the plurality of parts of the sole.

[0021] [Aspect 6] The method for estimating the position change of the upper body part according to [Aspect 3] or [Aspect 4], wherein the force is pressure and the force sensor is a pressure sensor. As in the above method, as the force sensor, a measuring device including an insole in which a pressure sensor is incorporated may be used, and the change over time of the pressure may be measured as the change over time of the force applied to each of a plurality of parts of the sole.

Advantages of the Invention

[0022] According to the present invention, when the person to be estimated walks, by measuring the change over time of the force applied to each of a plurality of parts of the sole, the change over time of the position of each of a plurality of parts of the upper body can be estimated.

Brief Description of the Drawings

[0023]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Modes for Carrying Out the Invention

[0024] Hereinafter, an embodiment that embodies a method for estimating the position change of the upper body part will be described with reference to the drawings. In the estimation method of this embodiment, a first measuring device 10 shown in FIG. 1 and a second measuring device 20 shown in FIG. 2 are used. Next, each measuring device will be described.

[0025] <The first measuring device 10> As shown in FIG. 1, the first measuring device 10 is installed at a location having a flat surface 8 that is horizontal or nearly horizontal. A part of the flat surface 8, which has a certain length, for example, 10 meters in the left-right direction of FIG. 1, is used as a linear walking path 9 on which the data provider P1 walks. Note that the walking path 9 may be set indoors or outdoors.

[0026] The first measuring device 10 includes a pair of imaging devices 11, 12. An example of each imaging device 11, 12 is a digital camera, a tablet equipped with a digital camera, or the like. The pair of imaging devices 11, 12 are installed on both sides in the length direction of the walking path 9 and sandwich the walking path 9. More specifically, the imaging device 11 is installed near one end 9a of the walking path 9, and the imaging device 12 is installed near the other end 9b of the walking path 9. The two imaging devices 11, 12 are installed facing each other in the extending direction of the walking path 9.

[0027] Each imaging device 11, 12 generates an image by imaging the data provider P1 walking on the walking path 9. Then, each imaging device 11, 12 generates moving image data in which the plurality of images are continuous over time (in time series).

[0028] Each imaging device 11, 12 includes a transmission unit (not shown). The transmission unit for each imaging device 11, 12 transmits the generated moving image data to cloud computing (cloud) via a network.

[0029] In addition, by motion-capturing the above video data, the change over time in the positions of multiple parts of the upper body P1a of the data provider P1 is calculated. In the present embodiment, a plurality of joints at multiple locations on the upper body P1a are set as the multiple parts of the upper body P1a. A joint is a part where adjacent bones are connected. The bones can move in various directions with the joint as a fulcrum. A joint is a location that serves as a fulcrum when the bones move and is the location where the amount of change in position is the smallest when the upper body P1a moves. The calculation of the change over time in the above positions is performed at an arbitrary timing before the creation of the learning model described later.

[0030] <Second measuring device 20> As shown in FIGS. 1 and 2, the second measuring device 20 includes an insole 22 laid on the inner bottom of the shoe 21. A pressure sensor 23 for measuring pressure is incorporated in each insole 22 as a force sensor that measures the force applied to multiple parts of the sole (underside) of the foot 24 during walking. The pressure sensor 23 includes a heel sensor 23a, a toe sensor 23b, an inner sensor 23c, and an outer sensor 23d.

[0031] Note that walking generally consists of a stance phase, which is the period when the sole touches the ground and supports the body, and a swing phase, which is the period when one foot 24 is lifted and swung forward. In the present embodiment, in addition to the above stance phase and swing phase, the period when stopped in a static standing position immediately before starting movement and the period when stopped in a static standing position immediately after finishing movement are also included in walking.

[0032] As shown in FIG. 2, the heel sensor 23a is disposed on the insole 22 at a portion where the load of the heel 24a is applied and measures the pressure applied to the heel 24a of the sole. The toe sensor 23b is disposed on the insole 22 at any one of the portions where the load is applied among the five fingers constituting the toe 24b and measures the pressure applied to the toe 24b of the sole.

[0033] Here, the inner side of the foot 24 is the side closer to the opposite foot 24 in the left-right direction, and the outer side is the side farther from the opposite foot 24 in the same direction. The inner sensor 23c is disposed inside the virtual line L connecting the heel sensor 23a and the toe sensor 23b on the insole 22, at a position where the load on the ball of the thumb is applied, and measures the pressure applied to the inner part of the sole of the foot. The outer sensor 23d is disposed outside the virtual line L on the insole 22, at a position where the load on the ball of the little finger is applied, and measures the pressure applied to the outer part of the sole of the foot.

[0034] The reason why the pressure sensors 23 are arranged at the above four positions on the insole 22 is as follows. When walking, among the plurality of parts of the sole of the foot, the heel 24a touches the ground first, and the part that touches the ground changes in order from the heel 24a toward the toe 24b. Further, after the heel 24a touches the ground and before the toe 24b touches the ground, at least one of the inner and outer parts with respect to the virtual line L touches the ground. The toe gripping force generated varies depending on the part that touches the ground.

[0035] Therefore, in the present embodiment, the pressure sensors 23 are incorporated in the insole 22 at the parts where the loads of the heel 24a and the toe 24b are applied, respectively. Further, the pressure sensors 23 are incorporated in the insole 22 at the inner and outer parts with respect to the virtual line L, respectively. Therefore, the pressure sensors 23 incorporated in the insole 22 appropriately measure the change over time of the pressure applied to each of the plurality of parts of the sole of the foot.

[0036] As the pressure sensor 23, a known pressure-sensitive sensor using a piezoelectric element or the like can be used. In particular, in view of the use situation of being disposed on the sole of the foot, from the viewpoints of stretchability and durability, it is preferable to use an elastomer-based capacitance-type sensor using a dielectric elastomer. Examples of the dielectric elastomer include crosslinked polyrotaxane, silicone elastomer, acrylic elastomer, urethane elastomer, and the like.

[0037] Since the above-described capacitive sensor made of an elastomer can be formed thinly as a whole, there is also an advantage that even when it is placed in the insole 22, the thickness of the insole 22 does not increase significantly. This sensor has a structure in which a dielectric elastomer is disposed between a pair of electrodes. Then, as the dielectric elastomer deforms due to pulling or stress, the amount of electricity (capacitance) stored in the electrodes changes. This change amount is measured as the pressure on the sole of the foot.

[0038] Each measured value measured by each pressure sensor 23 is a measured value that can be converted into a pressure value corresponding to the measurement method of the pressure sensor 23, such as a capacitance value or an electrical resistance value. Figure 3 shows an example of a waveform representing the temporal change in pressure for each part of the sole of the foot measured by each pressure sensor 23. In FIG. 3, the solid line indicates the measured value of the heel sensor 23a, the one-dot chain line indicates the measured value of the toe sensor 23b, the two-dot chain line indicates the measured value of the inner sensor 23c, and the broken line indicates the measured value of the outer sensor 23d.

[0039] The insole 22 incorporated with the above-described pressure sensor 23 is used, and the subjects for measuring the temporal change in pressure applied to each of a plurality of parts of the sole of the foot are a plurality of data providers P1 shown in FIG. 1 and a predetermined estimated subject P2 shown in FIG. 4. In other words, for the data provider P1 and the estimated subject P2, the temporal change in pressure applied to each of a plurality of parts of the sole of the foot is measured using a common insole 22. Here, the common insole 22 refers to the same type of insole 22, and more specifically, an insole 22 in which the same number of pressure sensors 23 as described above are arranged at the same locations as described above.

[0040] The measurement of the temporal change in pressure performed on the data provider P1 shown in FIG. 1 is performed simultaneously at the same location where the measurement of the temporal change in position was performed. Therefore, when the data provider P1 walks on the walking path 9, in addition to the measurement of the temporal change in position, the temporal change in pressure is also measured.

[0041] In addition to the pressure sensor 23, the insole 22 includes a transmission unit (not shown). The transmission unit transmits the pressure data measured by each pressure sensor 23 to the cloud via a network. This transmission is performed at a timing synchronized with the timing at which video data is transmitted from the transmission units of the imaging devices 11 and 12. The synchronization is performed, for example, by using a method defined in a communication protocol such as NTP.

[0042] The measurement of the change over time of the pressure on the sole of the foot performed on the estimated subject P2 shown in FIG. 4 is performed when the estimated subject P2 walks on the walking path 26 set on a flat surface 25 that is horizontal or nearly horizontal.

[0043] Regarding the data provider P1, the change over time of the position measured by the first measuring device 10 and the change over time of the pressure measured by the second measuring device 20 are used when estimating the change over time of the position of each of a plurality of parts of the upper body P2a of the estimated subject P2.

[0044] <Actions of this Embodiment> Next, as an action of this embodiment, a procedure for estimating the change over time of the position of each of a plurality of parts in the upper body P2a of the estimated subject P2 from the change over time of the pressure applied to each of a plurality of parts of the sole of the foot of the estimated subject P2 will be described with reference to the flowchart of FIG. 5.

[0045] For this estimation, a technique called machine learning, which is learned by a computer, is used. Here, as one of the machine learning methods, there is a neural network proposed as a mathematical model that mimics the neural circuits of the human brain. In the human brain, a large number of nerve cells (neurons) cooperate to exchange electrical signals, that is, by neurons transmitting electrical signals to each other, processes such as thinking and recognition are performed.

[0046] Each neuron receives an electrical signal from the neurons on the input side of the electrical signal and accumulates it. When the accumulated amount of electricity exceeds a threshold value, each neuron transmits an electrical signal to the neurons on the output side. The fact that this electrical signal exceeds the threshold value and the transmission of the electrical signal to the neurons on the output side are called "firing". Also, each neuron is connected to a plurality of neurons with respect to any neurons on the input side and the output side. The connection strength varies depending on the combination of neurons.

[0047] On the other hand, a neural network has a structure in which a large number of neurons are combined. A neural network has a structure called a "layer" in which a plurality of neurons gather. This layer includes an "input layer" to which data is input, an "output layer" that outputs a result, and a hidden layer (intermediate layer) that is a layer between the input layer and the output layer. Each layer has a structure in which a plurality of nodes are connected by edges. The nodes of each layer use the outputs of all the nodes of the previous layer as inputs.

[0048] In a neural network, values are converted from layer to layer. A neural network is considered to be one large function formed by a series of such conversions. Then, the number of nodes in the input layer and the output layer is determined by what data is input and what output is desired.

[0049] In a neural network, the value of the electrical signal received by a predetermined neuron is adjusted according to the strength of the electrical signal input to that neuron and the degree of connection between neurons. When the amount of electricity accumulated by the above neuron exceeds a certain threshold value, "1" is output. This corresponds to the firing of the neuron. That is, the above node is activated and data is transmitted to the next layer of the network. If the threshold value is not exceeded, the data is not transmitted to the next layer of the network.

[0050] The above neural network is applied to "learning" and "inference (estimation)". Learning is to gradually adjust the weights of each layer of the neural network to reduce the error from the correct label, that is, to improve the accuracy. Inference (estimation) is the step of obtaining an answer using the adjusted network. In the flowchart of FIG. 5 above, these learning and inference (estimation) are performed.

[0051] In the provider measurement step S11, as shown in FIG. 1, the data provider P1 wears the shoes 21 with the insole 22 shown in FIG. 2. Before starting to walk, the data provider P1 takes a static posture at one end 9a in the length direction of the walking path 9. At this time, that is, in the static standing position, the data provider P1 is imaged by each imaging device 11, 12 for a certain period of time (for example, about 30 seconds) for calibration. The calibration here refers to the work of determining whether the images and settings captured by the imaging devices 11, 12 are correct based on the imaging data in the static standing position, and adjusting the settings etc. as necessary so that accurate images can be captured.

[0052] The data provider P1 walks on the walking path 9 set on a flat surface 8 that is horizontal or nearly horizontal, starting from one end 9a in the length direction and walking towards the other end 9b. When the data provider P1 reaches the end 9b, it changes direction and walks towards the end 9a in the opposite direction. Then, the data provider P1 repeats the above walking path 9 a plurality of times, for example, 2 round trips, and returns to the end 9a.

[0053] Each imaging device 11, 12 generates an image by imaging the data provider P1 during the period of walking on the walking path 9. Then, each imaging device 11, 12 generates moving image data in which the above plurality of images are continuous over time (in time series). The transmission unit for each of the imaging devices 11, 12 transmits the moving image data generated by the imaging devices 11, 12 to the cloud via the network.

[0054] Also, while the data provider P1 is walking on the walking path 9, the pressure sensor 23 measures the change over time of the pressure applied to each of a plurality of parts of the sole of the foot. That is, the heel sensor 23a measures the pressure applied to the heel 24a of the sole of the foot, and the toe sensor 23b measures the pressure applied to the toe 24b of the sole of the foot. The inner sensor 23c measures the pressure applied to the inner part of the sole of the foot, and the outer sensor 23d measures the pressure applied to the outer part of the sole of the foot. Then, the transmission unit of each insole 22 transmits the pressure data measured by each pressure sensor 23 to the cloud via the network.

[0055] The above two types of measurements are made for a large number of data providers P1 in FIG. 1. Through these measurements, video data and sole pressure data regarding a large number of data providers P1 are obtained.

[0056] Next, in the learning model creation step S12 in FIG. 5, the correlation between the change over time of the pressure measured in the provider measurement step S11 and the change over time of the position is learned. That is, a learning model is created by finding laws (rules) and other common features hidden in a large amount of data. Feature quantities, which numerically represent features, are treated as variables that serve as clues for estimation in machine learning.

[0057] The learning model is an algorithm that represents a law, that is, a series of procedures for solving problems, and can also be expressed as a predictor. During the above learning, the weights representing the connection strength between neurons are adjusted so that the error (the value of the error function) between the output value and the correct answer becomes small. In this adjustment, for example, an approximate value of the weight that minimizes the value of the error function is obtained by methods such as the gradient descent method and the error backpropagation method. By repeatedly updating the weights according to the obtained results, the weights are brought closer to the target values.

[0058] As for the architecture of the neural network described above, a long short-term memory (LSTM) network, which is an improvement on the recurrent neural network (RNN), is suitable. The RNN is an extension of the neural network that enables handling of time series data. Time series data is data whose values change over time.

[0059] The LSTM enables retention of the influence of outputs from the distant past by introducing the concept of the length of the memory period with respect to the output of the hidden layer (intermediate layer) in the RNN. In the subject measurement step S13 of FIG. 5, as shown in FIG. 4, the subject to be estimated P2 wears shoes 21 on which an insole 22 common to the insole 22 used for pressure measurement of the data provider P1 is laid. The subject to be estimated P2 walks on a walking path 26 set on a flat surface 25 that is horizontal or nearly horizontal. At this time, the change over time of the pressure applied to each of a plurality of parts of the sole is measured. This measurement is performed in the same manner as the measurement in the provider measurement step S11 described above. That is, while the subject to be estimated P2 is walking, the heel sensor 23a measures the pressure applied to the heel 24a, and the toe sensor 23b measures the pressure applied to the toe 24b. Also, during the above period, the inner sensor 23c measures the pressure applied to the inner part of the sole, and the outer sensor 23d measures the pressure applied to the outer part of the sole. Then, the transmission unit of each insole 22 transmits the pressure data measured by each pressure sensor 23 to the cloud via the network.

[0060] In the estimation step S14 of FIG. 5, the change over time of the pressure measured in the subject measurement step S13 is given as unknown data to the learning model created in the learning model creation step S12. Then, in the estimation step S14, the change over time of the position of each of a plurality of parts of the upper body P2a of the subject to be estimated P2 is estimated.

[0061] Among the plurality of procedures described above in FIG. 5, the learning model creation step S12 and the estimation step S14 are performed by a computer for analysis. <Effects of the Present Embodiment> (1-1) When the data provider P1 walks, the change over time of the pressure applied to each of a plurality of parts of the sole (heel 24a, toe 24b, inner and outer parts with respect to the virtual line L) is measured. In synchronization with this measurement, the change over time of the position of each of a plurality of parts in the upper body P1a is measured. A learning model is created by learning the correlation between the measured change over time of the pressure and the change over time of the position. By using the same insole 22 as that used for the pressure measurement of the data provider P1, when the person to be estimated P2 walks, the change over time of the pressure applied to each of a plurality of parts of the sole is measured. The measured change over time of the pressure is given to the above learning model to estimate the change over time of the position of each of a plurality of parts of the upper body P2a of the person to be estimated P2.

[0062] Therefore, when the person to be estimated P2 walks, by measuring the change over time of the pressure applied to each of a plurality of parts of the sole, the change over time of the position of each of a plurality of parts of the upper body P2a can be estimated. (1-2) From the change over time of the position of each of a plurality of parts of the upper body P2a of the person to be estimated P2 estimated in the above (1-1), it becomes possible to obtain various useful information regarding the state of the body of the person to be estimated P2 shown below.

[0063] · The posture of the person to be estimated P2. For example, the arms are bent, the torso is leaning forward, etc. From this information, it becomes possible to recommend to the person to be estimated P2 to take appropriate measures before getting sick or to undergo a medical check-up for prevention.

[0064] · The movement of the person to be estimated P2. For example, walking while holding a smartphone. · The deviation between the image of one's own body (body image) in the brain of the person to be estimated P2 and the state of the actually functioning body. The body image includes functions such as grasping the outline (shape), size, position, etc. of one's own body, and also includes a function of grasping how much motor ability one has.

[0065] It is considered that there is a correlation between the degree of this deviation and the progression degree of dementia. Therefore, if the above deviation is known, it becomes possible to estimate the progression degree of dementia, for example, mild cognitive impairment (MCI).

[0066] (1 - 3) In the provider measurement step S11, as a plurality of parts in the upper body P1a of the data provider P1, the change over time in the position of the joints of the upper body P1a is measured. A joint is a place where the amount of change in position is the least when the upper body P1a moves. Therefore, it is possible to measure the change over time in the position of each of the plurality of parts of the upper body P1a at a place where the movement of the upper body P1a is small.

[0067] (1 - 4) When measuring the pressure for each of the plurality of parts of the sole, shoes 21 having an insole 22 in which a pressure sensor 23 is incorporated are provided at the inner bottom. Therefore, in the provider measurement step S11, the data provider P1 walks while wearing the shoes 21, and in the subject measurement step S13, the estimated subject P2 walks while wearing the shoes 21. By doing so, it is possible to measure the change over time in the pressure applied to each of the plurality of parts of the sole.

[0068] (1 - 5) The pressure sensor 23 is incorporated in each of the part where the load of the heel 24a is applied and the part where the load of the toe 24b is applied in the insole 22. Also, the pressure sensor 23 is incorporated in each of the parts that are inner and outer with respect to the virtual line L in the insole 22. Therefore, it is possible to accurately measure the change over time in the pressure applied to each of the plurality of parts of the sole with a small number of pressure sensors 23.

[0069] <Modification Example> This embodiment can be implemented with the following modifications. This embodiment and the following modification examples can be implemented in combination with each other within a technically non - conflicting range.

[0070] · Personal authentication may be performed by utilizing the change in position over time estimated in the estimation step S14. The above authentication may be performed only based on the change in position over time of each estimated part. Further, the above authentication may be performed by combining with other authentication techniques, such as ID cards, fingerprint authentication, face authentication, etc. By combining them, it is possible to improve the accuracy of authentication.

[0071] · When the above personal authentication is performed, for example, by walking and passing through the authentication gate, a pressure sensor for measuring the pressure on the sole may be incorporated into the floor surface near the authentication gate instead of the insole 22. In this case, there is an advantage that authentication can be performed even without the insole 22 by walking on the floor.

[0072] Also, a person to be authenticated at the authentication gate may be made to wear the shoes 21 equipped with the insole 22, walk, and pass through the authentication gate. · The feet 24 on which the pressure for each part applied to the sole is measured by the pressure sensor 23 are, in principle, both feet 24, but may be one of the left and right feet 24.

[0073] · The number and arrangement of the pressure sensors 23 incorporated in the insole 22 may be appropriately changed. · As the first measuring device 10, an acceleration sensor may be used instead of the imaging devices 11, 12.

[0074] · The lengths of the walking paths 9, 26 may be appropriately changed. Also, the number of times the data provider P1 travels back and forth on the walking path 9 may be changed. · In the above embodiment and the above modification example, pressure is described as one of the forces, but the force may be acceleration or torque. The force sensor may be an acceleration sensor, a torque sensor, etc. capable of measuring those forces. These sensors may be incorporated into the insole 22, for example.

Explanation of Signs

[0075] 10… First measuring device 11, 12… Imaging devices 20…Second measuring device (measuring device) 21…Shoes 22…Insole 23…Pressure sensor (force sensor) 24a…Heel 24b…Toe L…Virtual line P1…Data provider P1a, P2a…Upper body P2…Subject to be estimated

Claims

1. A provider measurement step of measuring the change over time of the force applied to each of a plurality of parts of the sole of the data provider when the data provider walks, and synchronously with the measurement, measuring the change over time of the position of each of a plurality of parts of the upper body of the data provider; A learning model creation step of creating a learning model by learning the correlation between the change over time of the force and the change over time of the position measured in the provider measurement step; A subject measurement step of measuring the change over time of the force applied to each of a plurality of parts of the sole of the subject when the subject walks, using a measurement device common to the measurement device used for the force measurement of the data provider; An estimation step of giving the change over time of the force measured in the subject measurement step to the learning model created in the learning model creation step, and estimating the change over time of the position of each of a plurality of parts of the upper body of the subject; A method for estimating the position change of the upper body part, comprising:

2. The method for estimating the position change of the upper body part according to claim 1, wherein in the provider measurement step, the change over time of the position of a plurality of joints as a plurality of parts of the upper body is measured.

3. The measurement device includes an insole in which a force sensor is incorporated, In the provider measurement step, the data provider walks wearing shoes with the insole laid thereon, and the change over time of the force applied to each of a plurality of parts of the sole is measured by the force sensor, The method for estimating the position change of the upper body part according to claim 1 or claim 2, wherein in the subject measurement step, the subject walks wearing shoes with the insole laid thereon, and the change over time of the force applied to each of a plurality of parts of the sole is measured by the force sensor.

4. In the provider measurement step and the subject measurement step, as the insole, the force sensor is incorporated in each of the part where the heel load is applied and the part where the toe load is applied, and the force sensor is incorporated in each of the inner and outer parts with respect to the virtual line connecting the heel and the toe, and the change over time of the force is measured by each force sensor. The method for estimating the position change of the upper body part according to claim 3.

5. The method for estimating the position change of the upper body part according to claim 1, wherein the force is pressure.

6. The method for estimating the position change of the upper body part according to claim 3, wherein the force is pressure and the force sensor is a pressure sensor.

Images