JP2018121930A - Gait evaluation method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To evaluate a gait on an evaluation axis that anyone can easily understand, and to make the evaluation result of the gait useful for improving the gait. SOLUTION: An impression of gait of an arbitrary subject is used by using a multiple regression equation in which an evaluation value of gait impression is used as an objective variable and a gait parameter including a gait characteristic amount obtained by gait measurement is used as an explanatory variable. This is a gait evaluation method for obtaining the evaluation value of gait. included. [Selection diagram] Fig. 6

Description

  The present invention relates to a gait evaluation method.

  The gait is the appearance when walking. Nearly half of women in their 30s and 60s think that their gait is bad, and there are many people who want to improve their gait.

  As an apparatus that can be used to improve gaits, a device that includes a triaxial acceleration sensor carried by a subject and a processing unit that generates evaluation information of the gait of the subject based on the output of the triaxial acceleration sensor has been proposed. . In this device, features such as the back and forth movement of the waist and the tilt of the trunk are detected from the output of the three-axis acceleration sensor, and the features are set to the features in a known ideal gait or in a predetermined category. The degree of similarity of the subject's gait with respect to the ideal gait or the gait classified into a predetermined category is evaluated by collating with the feature amount of the gait classified (Patent Document 1).

  In addition, the foot pressure distribution during walking is measured using a foot pressure sensor, and the mechanical rationality such as balance of posture, symmetry, and temporal continuity of walking is calculated from the walking features obtained from the foot pressure distribution. A device that calculates the beauty and health of walking based on judgment criteria (Patent Document 2), and the walking feature value obtained from the foot pressure distribution is compared with the ideal walking feature value as a model. An apparatus for scoring the beauty of walking (Patent Document 3) has been proposed.

Japanese Unexamined Patent Publication No. 2016-2109 JP 2001-218754 A JP 2002-233517 A

  In general, the posture, balance, stability, step length, step length, and the like of walking are points when evaluating a gait. However, even though the gait is a manifestation of individuality, it cannot be uniformly determined that a specific gait is ideal for any subject. Therefore, simply evaluating the subject's gait by comparing the feature amount of the subject's walking with the ideal feature amount of the walking may result in the evaluation result being difficult for the subject to understand and does not necessarily improve the gait. Not always helpful.

  With respect to such a conventional technique, an object of the present invention is to evaluate a gait with an evaluation axis that anyone can easily understand, and to make the gait evaluation result useful for improving the gait.

The inventor
(i) In order to present the gait impression evaluation results to the subject in an easy-to-understand manner, the gait evaluation results should be presented on an evaluation axis using onomatopoeia that can be easily understood by a wide range of age groups. Is useful,
(ii) In this case, the two poles of the evaluation axis are taken as a pair of sensitivity words having an impression of the opposite electrode, and obtained from the evaluation value of the gait based on this evaluation axis and the measurement value of walking using an acceleration sensor, a foot pressure sensor, etc. The correlation between the evaluation value calculated by the regression equation and the actual evaluation value of the subject shows a good correlation,
(iii) The evaluation value calculated from the characteristic amount of the walking of the subject using this regression formula does not represent the difference between the ideal gait defined uniformly and the gait of the subject. It shows how other people perceive gaits, and is expressed on the evaluation axis using onomatopoeia, so it is easy for subjects to accept,
(iv) Since the evaluation axis having a pair of sensitivity words representing the impression of the opposite electrode can be set in various ways, it becomes possible to evaluate the gait of the subject with an evaluation axis from a plurality of viewpoints. The subject can know his / her gait assessment from multiple points of view, increasing motivation for gait improvement,
The present invention was conceived.

  That is, the present invention uses a multiple regression equation in which an evaluation value of an impression of a gait is an objective variable and a walking parameter including a characteristic amount of walking obtained by measurement of walking is used as an explanatory variable. A gait evaluation method for obtaining an evaluation value of an impression of a gait, wherein the evaluation value of a gait impression is represented by an evaluation axis with a pair of sensitivity words representing the opposite impression of the gait as a bipolar, and a pair of sensitivity An evaluation method is provided wherein onomatopoeia is included in at least one of the words.

  According to the present invention, the evaluation result of the gait of the subject is not a comparison between the ideal gait determined uniformly and the gait of the subject. It will be easy to accept for the subject because it will indicate whether it is captured with a good impression.

  In addition, since the evaluation axis of the gait is a pair of sensitivity words representing the impression of the opposite electrode, and at least one of the pair of sensitivity words includes onomatopoeia, the evaluation result according to the present invention indicates what age group the subject has It will be easy to understand.

  Furthermore, since various evaluation axes can be set by selecting the sensitivity words that are the two extremes of the evaluation axis, the gait of the subject can be evaluated from various aspects, which is useful for improving the gait of the subject.

FIG. 1 is an explanatory diagram of a feature amount of walking that is an index of a spatial feature of a gait. FIG. 2 is an explanatory diagram of a feature amount of walking that serves as an index of temporal features of gaits. FIG. 3A is an explanatory diagram of a head inclination angle and a back muscle angle. FIG. 3B is an explanatory diagram of a shoulder rotation angle. FIG. 3C is an explanatory diagram of a hip rotation angle. FIG. 3D is an explanatory diagram of head up and down shaking. FIG. 3E is an explanatory diagram of head left / right deflection. FIG. 3F is an explanatory diagram of a knee joint movable range. FIG. 3G is an explanatory diagram of hand shaking. FIG. 4 is a diagram illustrating a method for evaluating a gait by VAS. FIG. 5 is a diagram showing the position of each part of the human body determined based on the image signal of the camera constituting the motion capture. FIG. 6 is a diagram illustrating the correlation between the VAS evaluation value and the estimated value of “Ikiiki-Tobotobo”. FIG. 7 is a diagram illustrating the correlation between the VAS evaluation value and the estimated value of “Ikiiki-Tobotobo”. FIG. 8 is a diagram illustrating the correlation between the VAS evaluation value and the estimated value of “Ikiiki-Tobotobo”. FIG. 9 is a diagram illustrating the correlation between the VAS evaluation value and the estimated value of “Ikiiki-Tobotobo”. FIG. 10 is an example of presentation of the impression evaluation result on the evaluation axis of “Ikiiki-Tobotobo”. FIG. 11 is a configuration example of a paper surface or a screen for explaining the feature amount of walking.

<< Summary of Invention >>
The gait evaluation method of the present invention obtains a multiple regression equation using the evaluation value of the gait impression as an objective variable and the walking parameter as an explanatory variable, and uses the multiple regression equation to calculate the gait of any subject. The evaluation value of the gait impression is expressed by an evaluation axis having a pair of sensitivity words representing the opposite impressions of the gait and the sensitivity axis of the pair of sensitivity words. It is characterized by including onomatopoeia in at least one. In this case, multiple types of gait evaluation axes can be set. Therefore, it becomes possible to evaluate the gait from multiple aspects with a plurality of types of evaluation axes.

  In order to create a multiple regression equation, the evaluation value of the gait impression and the data of the walking parameters are required in advance for a plurality of pedestrians. In this case, the pedestrians are preferably classified by age and gender. Moreover, it is preferable that the observer who gives the evaluation value of the impression to the gait of the pedestrian is also classified by age and gender.

<< Assessment axis of gait impression >>
In the multiple regression equation used by the evaluation method of the present invention, the evaluation axis of the evaluation value of the impression of the gait is a pair of sensitivity words representing the opposite impression of the gait, and the pair of sensitivity words. Onomatopoeia is included in at least one of the above.

  Here, the sensitivity word is a word representing the impression of the observer. Onomatopoeia is a general term for onomatopoeia, onomatopoeia, and mimicry words, and is an expression that replicates events and states with linguistic sounds. Therefore, in the present invention, onomatopoeia is an expression that may represent not only the state of the subject's walk itself but also the state of the subject and the emotion of the walking subject.

As an example of a pair of sensitivity words forming both poles of the evaluation axis and including onomatopoeia in at least one of them, the following pairs can be given, for example.
Ikiiki-tobo-boshi-seki--Hero-Hero-Pitsu--Sloppy wild--Stud wild-

  By using the sensitivity poles representing the impression of the opposite electrode as the opposite poles of the evaluation axis, the variation width of the evaluation values when multiple observers observe a given gait, and the same observer with the same gait It is possible to suppress the variation range of the evaluation values when repeatedly observed at different times. This is thought to be because the viewpoint of the evaluation axis is clarified by prescribing the impressions on both sides of the evaluation axis, and the observer can easily show the impression of the gait of the evaluation target on the evaluation axis. . For example, when an arbitrary observer is asked to indicate the degree of “lively” of a predetermined gait at a position between 0 and 10 on the evaluation axis, the same gait is set to 0, and “ In the case where it is calculated as indicated by the position on the evaluation axis when “Iki-iki” is set to 10, the range of variation in evaluation values is narrower when “Tobo-tobo-bo” is set to 0 and “Iki-iki” is set to 10. The correlation between the actual evaluation value and the evaluation value calculated by the regression equation becomes high.

  In addition, by including onomatopoeia in at least one of the two poles of the evaluation axis, the evaluation axis can be easily understood by people of a wide range of age groups and easily accustomed as an evaluation axis representing gaits. It is not always necessary to use both onomatopoeia as a pair of sensitivity words.

  A pair of sensitivity words including the above-mentioned onomatopoeia was obtained as follows. That is, the present inventor observed the gaits of 70 pedestrians on the street, extracted 244 sensitivity words including onomatopoeia as impressions of the gaits, and on the other hand, walked by factor analysis of gait VAS values described later. Sense words 244 words were subjected to factor analysis assuming that there are four factors, and a counter word was formed from the coefficient of each factor and the meaning of each word of the sensitivity word 244 words.

  In the present invention, the pair of sensitivity words representing the impression of the counter electrode is not limited to the above example, and a pair of sensitivity words including onomatopoeia generally recognized as the meaning of the counter electrode can be used.

  In addition, in the above-mentioned pairs of sensitivity words, for example, both “Ikiiki-Tobotobo” and “Ikiiki-Yobo-yobo” have “Iki-Iki” as one end of the axis, but “Iki-Iki-Tobobo” is related to impression. On the other hand, “Yobo-Yobo-Ikiiki” is related to the impression of aging, and these are different evaluation axes. In this way, by using a pair of sensitivity words representing an impression of a counter electrode including onomatopoeia as the two poles of the evaluation axis, it becomes possible to perform a multifaceted evaluation of the gait impression.

<Gathering evaluation data for gait impressions>
When obtaining the evaluation value of the gait impression table and the regression equation of the gait parameter, VAS (Visual Analogue Scale), NRS (Numerical Rating Scale), etc. are used as a method of collecting the gait impression evaluation value data. be able to. For example, when the observer evaluates “Noshinashi-Sutasuta” with a VAS for a predetermined gait, the observer has “Shinoshita” at one end and “Sutasuta” at the other end as shown in FIG. The vertical line 10 corresponding to one's own impression is written in the evaluation axis of the question paper in which the evaluation axis of “ According to VAS, an evaluation value of an observer's impression can be easily obtained as a relative evaluation between “Noshinoshi” and “Sutasuta”, and VRS (Verbal Rating Scale) of four-level evaluation It is possible to obtain a highly reliable evaluation value as compared with the above. Note that, from the VAS data in which the observer has written a vertical line, for example, the evaluation value of the position of the vertical line can be read with one end of the evaluation axis as zero and the other end as 100. it can.

<< Walking parameters >>
The walking parameter is an objective feature amount that affects the gait. The walking parameters include (a) a walking feature value obtained by walking measurement using a measuring device such as a foot pressure sensor, motion capture, and acceleration sensor, and (b) a walking feature value obtained from walking measurement. It does not include basic information such as age, height, weight, BMI, and gender that affect gait. When a feature amount of a certain gait can be measured by a plurality of measuring devices, it is preferable to use a feature amount of a gait by a measuring device with high measurement accuracy.

<Foot pressure sensor>
As the foot pressure sensor, it is possible to use a foot pressure distribution image received on the walking surface by walking, or a device capable of outputting two-dimensional foot pressure data. For example, a seat type lower limb weight meter manufactured by Anima Co., Ltd. A series walk way, a floor reaction force meter manufactured by AMTI Co., Ltd. and the like can be mentioned.

Examples of the feature amount of walking obtained from walking measurement using a foot pressure sensor include those shown in Table 1.

  Of these, for those having a left-right difference, it is preferable to use the left and right average values in terms of the correlation between the walking feature value and the gait evaluation value. Therefore, it is preferable that the left and right average values are used as feature amounts for the stance period ratio, the swing leg period ratio, the both leg support period ratio, the stride, the stride, the step, the walking angle, and the toe angle.

  Of these walking features, the stride, the step, the walking angle, the stride, and the toe angle are spatial indices of the gait as shown in FIG. Further, the stance period ratio, the free leg period ratio, and the both-leg support period ratio are temporal indexes of gaits as shown in FIG.

  Here, the walking ratio is a value obtained by dividing the step length (m) by the cadence (number of steps / minute). It is known that the walking ratio indicates the walking efficiency, which is about 0.0063 on average for adults and low for infants and the elderly. This is an important indicator of walking ability because it decreases with age in connection with the fall history.

  The stance period ratio is a ratio of the stance period time to one walking cycle, and the stance period time is the time during which the foot is in contact with the ground from one heel contact to the other when the one foot leaves the ground.

  The swing period ratio is the ratio of the swing period time to one gait cycle. The swing period time is the time from the time when one foot on the right and left leaves the ground until the foot of the foot touches the ground. It's a floating time.

  The both-leg support period ratio is the ratio of the both-leg support period time in which both the left and right feet are in contact with the ground to one walking cycle. Therefore, the driving action is zero. The ratio of both-leg support period is about 20% in a normal walking action, but becomes longer with aging.

  The stride is the distance from the right and left saddle ground to the ground again. The stride is preferably corrected by height (ie, normalized by dividing by height).

  The stride is the distance from the left and right side of the heel to the ground on the other side. The left and right of the stride are determined by the left and right of the foot that is the axis foot. It is preferable to correct the stride by height (that is, normalize by dividing by height).

  The step is a horizontal distance from one heel contact to the left to the other heel contact, and determines the right and left of the step on the left and right sides of the foot that is the axial foot. It is preferable to correct the step by height (ie, normalize by dividing by height).

  The walking angle is an angle (°) formed by a straight line connecting one heel to the other heel with the traveling direction. The left and right of the walking angle are determined by the left and right of the foot that is on the ground as an axis foot. For example, when the shaft foot is the right foot and the left foot is stepped on and grounded, the angle formed by the straight line connecting the right foot heel and the grounded left foot heel as the shaft foot and the traveling direction is the right walking angle.

  The toe angle is an angle (°) formed by a straight line connecting the heel and the toe with the traveling direction, the outer side being positive and the inner side being negative.

<Motion capture>
For motion capture, use a depth camera (Microsoft's Kinect), a 3D motion analysis system that uses multiple video cameras (3D motion analysis device VICON, VICON MX system, VICON NEXUS manufactured by Interliha Corporation), etc. be able to. Among them, Microsoft's Kinect is equipped with an RGB color video camera, an infrared camera and an infrared light emitting unit for depth measurement, and automatically extracts the position information of the joint point of the subject and displays it on the walking image of the subject. It is possible to do.

  In the measurement of walking using motion capture, it is preferable to extract one walking cycle from the position coordinates and obtain the walking feature amount from the extracted value. Examples of the feature amount of walking by motion capture include those shown in Table 2.

  Here, as shown in FIG. 3A, the head inclination angle is an average value during a walking cycle of an angle θ1 formed by a straight line connecting the shoulder center and the head in the vertical direction in a side view, and the spine angle is the shoulder center angle. The average value during the walking cycle of the angle θ2 formed by the straight line connecting the center of the body and the straight line connecting the center of the shoulder and the head.

  As shown in FIG. 3B, the right shoulder rotation angle θ3R is the difference between the maximum rotation angle Maxθ3R and the minimum rotation angle Minθ3R of the right shoulder with respect to the shoulder center during the walking cycle, and the left shoulder rotation angle θ3L is determined in the same manner. The shoulder rotation angle θ3 is an average value of the right shoulder rotation angle θ3R and the left shoulder rotation angle θ3L.

  As shown in FIG. 3C, the right hip rotation angle θ4R is the difference between the maximum rotation angle Maxθ4R and the minimum rotation angle Minθ4R of the right waist with respect to the waist center during the walking cycle, and the left hip rotation angle θ4L is similarly determined. The waist rotation angle θ4 is an average value of the right waist rotation angle θ4R and the left waist rotation angle θ4L.

  As shown in FIG. 3D, the head vertical shaking is a change amount of the position of the vertical coordinate of the head in a side view.

  As shown in FIG. 3E, the head left-right shake is the amount of change in the position of the horizontal coordinate of the head in a top view.

  As shown in FIG. 3F, the range of motion of the knee joint is the absolute difference between the maximum value and the minimum value of the angle θ5 formed by the straight line connecting the knee and the waist and the straight line connecting the knee and the buttocks in a side view. Left and right average of values.

  As shown in FIG. 3G, the hand shake is a left-right average of absolute values of the difference between the maximum value and the minimum value of the front-rear direction coordinate of the wrist that swings around the shoulder in a side view.

  Here, the head shape, left and right shoulders, shoulder center, left and right hips, waist center, wrist, knee and hips are recognized by the motion capture depth sensor and machine learning is performed from the person shape. It is a site | part determined as shown in FIG. 5 by the well-known method of estimating a skeleton point.

<Acceleration sensor>
As the acceleration sensor, one that can be carried in daily life and that can measure three-axis acceleration of the X axis, the Y axis, and the Z axis is preferable. As the three-dimensional acceleration sensor, a portable terminal capable of extracting a three-axis acceleration waveform can be used.

  Examples of the walking feature amount obtained using the acceleration sensor include walking speed, stride length, cadence, walking ratio, step distance, toe angle, and the like. As a method of obtaining these walking feature amounts from the acceleration sensor, the method described in JP-A-2015-66155 can be cited.

<Combination of walking parameters>
In order to extract the walking parameters that have a large influence on the evaluation axis in advance, a single regression analysis is performed on each of the 11 types of evaluation axes shown in Table 3 using various walking parameters. Table 3 shows the top 10 walking parameters of the absolute value of the correlation coefficient. Here, the data for performing this regression analysis is the observation of 10 healthy women who can walk independently with an age of 40 to 50 years old, who are 20 to 50 year old ordinary people (5 men and 5 women). Moving images of the frontal face and sagittal plane of walking were evaluated, and the observer recorded the VAS evaluation values on 11 types of evaluation axes for each moving image. Further, as a foot pressure sensor, a seat type lower limb weight meter series walk way manufactured by Anima Co., Ltd. was used, and as a motion capture, Kinect of Microsoft Corporation was used.

  From Table 3, it can be seen that walking speed, stride, and stride greatly contribute to most evaluation axes. Further, when the impression evaluation axis includes a word representing a body type, the above 11 evaluations are the walking parameters indicating postures such as the back muscle angle and the head inclination angle, that the evaluation axis has a large contribution of weight and BMI. It can be seen that the contribution is low in any of the axes. In addition, it turns out that it is effective to walk fast and shake your hand with a large thigh to achieve a generally good gait, such as lively, spicy, crunchy, relaxed, and smooth.

  Also, from Table 3, in order to increase the correlation between the evaluation value calculated by the multiple regression equation and the evaluation value of the actual gait impression, the type and number of walking parameters used in the multiple regression equation (that is, The number of explanatory variables) is preferably set appropriately for each type of evaluation axis.

In the multiple regression analysis using the gait evaluation value as the objective variable and the walking parameter as the explanatory variable,
(I) When using both the walking feature value obtained using the foot pressure sensor and the walking feature value obtained using motion capture,
(II) When using multiple types of walking features obtained using a foot pressure sensor,
(III) When using multiple types of walking features obtained using motion capture,
(IV) The case where a plurality of types of walking feature values obtained using an acceleration sensor are used is considered. According to (I), it is possible to improve the accuracy of the multiple regression equation estimation, and according to (IV), it is possible to easily obtain the feature amount of walking in daily life.

  On the other hand, methods for selecting explanatory variables in a multiple regression equation generally include a forced input method, a stepwise method, and a variable reduction method. Therefore, when evaluating the gait impression using the 11 types of evaluation axes shown in Table 2, the forced input method and the stepwise method are used for each of the above-mentioned walking feature value combination patterns (I) to (III). Or, when 20 explanatory variables are used in the variable reduction method, the coefficient of determination (square of correlation coefficient R) between the evaluation value calculated by the regression equation and the evaluation value of the actual gait impression is obtained. It was.

  In this case, as a foot pressure sensor, a seat type lower limb weight meter series walk way manufactured by Anima Co., Ltd. is used, and the walking features obtained using the foot pressure sensor include stride, stride, walking speed, cadence, walking Twelve kinds of distance, walking angle, walking ratio, toe angle, stance ratio, both-leg support period ratio, swing leg period ratio, and walking cycle time were used.

  In addition, Microsoft's Kinect is used for motion capture, and the walking features obtained using motion capture include head left / right shaking, head up / down shaking, shoulder rotation angle, waist rotation angle, and knee joint movement. Eight types of area, hand swing, back muscle angle, and head tilt angle were used.

The results are shown in Table 4.
From Table 4, the features of walking
(I) When using a foot pressure sensor (Walk Way) and motion capture (Kinect),
(II) When using only foot pressure sensor (Walk Way)
(III) When using only motion capture (Kinect),
In any of the above, it can be seen that the evaluation value of the gait impression calculated by the regression equation is correlated with the evaluation value of the actual gait by any of the forced injection method, the stepwise method, and the variable reduction method. .

In addition, when using both (I) the walking feature value obtained using the foot pressure sensor and the walking feature value obtained using the motion capture as the walking feature value, (II) obtained using the foot pressure sensor Compared to the case of using only the walking feature value or (III) only the walking feature value obtained using motion capture, the coefficient of determination is high and the accuracy of the regression equation is high. It can be seen that the coefficient of determination of the evaluation axis is high and the accuracy of the regression equation is high on this evaluation axis.

<< Specific example of multiple regression equation >>
(1) When using the walking feature value by foot pressure sensor and the walking feature value by motion capture Anima Corporation's seat-type lower limb weight meter series Walk Way is used as the foot pressure sensor, and Microsoft's Kinect is used as the motion capture. It was used.

  As for the walking of 10 healthy women aged 40-50, who can walk independently, stride and step are measured by Walk Way, and the head swings and shakes by hand and tilts by Kinect. The corner was measured.

  In addition, a general person (5 males and 5 females) aged 20 to 50 years is considered as an observer, and the above-mentioned animation of the frontal face and sagittal plane during walking is given to the observer on the evaluation axis of “Ikiiki-Tobotobo”. The evaluation was entered on a VAS question form, and the entry position was digitized in the range of 0 to 100 to give an evaluation value of “Ikiiki-Tobotobo”.

  Multiple regression analysis was performed using IBM's statistical data processing software SPSS Statistics 23 with the characteristic value of walking as an explanatory variable and the evaluation value of “Ikiiki-Tobotob” as an objective variable. The results are shown in Table 5 and FIG.

From Table 5, the following equation was obtained as a multiple regression equation for estimating “Ikiiki-Tobotobo”.
(Ikiiki-Tobotobo estimated value)
= -0.681 * (stride) + 3.034 * (step distance) + 369.552 * (head shake) -102.488 * (hand shake) + 0.599 * (head tilt angle) + 100.106 (constant)

  From this multiple regression equation, it can be seen that it is effective to walk straight with your legs stretched straight, swinging your hands wide, without widening the steps, conscious of your crotch than the tempo, to show “lively” .

(2) When only the walking feature value by the foot pressure sensor is used as the walking feature value For the same pedestrian walking as in (1), the stride, step, and both-leg support period ratio measured by Walk Way are the explanatory variables. Except that, multiple regression analysis was performed in the same manner as in (1). The results are shown in Table 6 and FIG.

From Table 6, the following equation was obtained as a multiple regression equation for estimating “Ikiiki-Tobotobo”.
(Ikiiki-Tobotobo estimated value)
= -1.381 * (Stride) + 2.633 * (Step) + 2.663 * (Both-leg support period) + 110.474 (Constant)

  From this multiple regression equation, it can be seen that it is effective to walk straight with your legs straight without expanding your step, conscious of your crotch than the tempo, in order to show “lively”.

(3) When only the walking feature value by motion capture is used as the walking feature value
Same as (1), except for the pedestrian's gait as in (1), with hand swing, head left / right swing, head tilt angle, shoulder rotation angle, and knee joint range of motion measured by Kinect as explanatory variables Multiple regression analysis was performed. The results are shown in Table 7 and FIG.

From Table 7, the following equation is obtained as a multiple regression equation for estimating “Ikiiki-Tobotobo”.
(Ikiiki-Tobotobo estimated value)
= -126.097 * (hand swing) + 586.881 * (head swinging) + 0.580 * (head tilt angle) -1.103 * (shoulder rotation angle) -0.770 * (knee joint range of motion) + 109.536 (constant)

  From this multiple regression equation, it can be seen that it is effective to walk with the shoulders stretched and the back stretched straight without shaking the head and moving the head to the left and right to show “lively”.

(4) When only the feature value of the walking by the acceleration sensor is used as the feature value of the walk For the same pedestrian's gait as in (1), the stride and the step obtained from the measurement data of the acceleration sensor are used as explanatory variables. Was subjected to multiple regression analysis in the same manner as (1). The results are shown in Table 8 and FIG.

From Table 8, the following equation was obtained as a multiple regression equation for estimating “Ikiiki-Tobotobo”.
(Ikiiki-Tobotobo estimated value)
= -3.163 * (step length) + 3.061 * (step distance) + 156.233 (constant)

  From this multiple regression equation, it can be seen that it is effective to walk with the legs straight and without widening the steps, conscious of large thighs, to show “lively”.

  Further, from the results of Tables 5 to 8 and FIGS. 6 to 9, (1) multiple regression analysis was performed using both the walking feature value measured by the foot pressure sensor and the walking feature value measured by the motion capture as walking parameters. In such a case, the regression is compared with the case where a multiple regression analysis is performed using a walking feature amount measured using any one of (2) foot pressure sensor, (3) motion capture, and (4) acceleration sensor alone. It can be seen that the accuracy of the analysis is high.

  In addition, (1) When the walking feature value by the foot pressure sensor and the walking feature value by motion capture are used as the walking parameter, (2) When only the walking feature value by the foot pressure sensor is used as the walking parameter, (3) In the case where only the walking feature value by motion capture is used as the walking parameter, and (4) only the walking feature value by the acceleration sensor is used as the walking parameter, the walking parameter as the explanatory variable of the multiple regression analysis is The measurement values in Table 1 or Table 2 are acquired by the measurement method, and multiple regression analysis is performed on the VAS evaluation value by the stepwise method, forced injection method, etc., from the positive / negative of the coefficient of the multiple regression equation, the significance probability, and the R-square value By judging the explanatory power of the multiple regression equation comprehensively, it seems that (1), (2), (3), (4) is the most relevant. It is a combination of walking parameters.

<< Walking improvement method >>
In the gait evaluation method of the present invention, the subject's gait evaluation value is calculated from the gait's gait parameter using the multiple regression equation of the gait impression evaluation value and the gait parameter. Evaluate the gait.

  As the multiple regression equation, the measurement device that measures the feature amount of walking that constitutes the explanatory variable may be any one of a foot pressure sensor, a motion capture, an acceleration sensor, and the like. However, from the viewpoint of accuracy of regression analysis, it is preferable to use both the walking feature value obtained using the foot pressure sensor and the walking feature value obtained using motion capture as explanatory variables.

  Since the evaluation value of the impression of the gait of the subject obtained from the multiple regression equation is such that at least one of the sensitivity words of both poles consists of onomatopoeia, the subject can easily understand the content being evaluated. The evaluation value corresponds to an impression when another person observes the gait of the subject, and does not indicate a similarity or a gap with the ideal gait defined uniformly. Therefore, the subject can easily accept the evaluation result of his / her gait. Therefore, only presenting the evaluation result of the impression of the subject's gait obtained from the multiple regression equation to any subject, the subject can improve the evaluation result of his / her own gait (that is, at both extremes of the evaluation axis). This makes it possible to make the gait of the subject closer to the impression in the direction desired by the subject).

  In addition, after the subject has received advice for improving the gait on the evaluation axis, the subject can easily calculate the evaluation value of the impression from the multiple regression equation by obtaining the feature amount of the subject's walking again, and the subject can easily , Know the degree of improvement of self gait.

  Furthermore, in the present invention, it is preferable to obtain a single regression equation between the characteristic amount of each gait used in the multiple regression equation for calculating the evaluation value of the gait impression and the evaluation value of the gait impression. . Thereby, the boundary value of the feature amount of the walking when the evaluation of the impression as the counter electrode is reversed is known. Therefore, when the evaluation value of the gait impression is obtained with a single regression equation for the gait of the subject, it can be easily understood whether or not the evaluation value of the subject is closer to the target gait than the boundary value. Therefore, the subject can strive to improve the gait using the boundary value as a guide. In addition, after the subject strives to improve the gait, the feature value of the gait is measured again, and when the evaluation value of the gait is obtained from the single regression equation, the subject should clearly confirm the result of the gait improvement. It is expected that the gait will be further improved.

More specifically, for example, 10 kinds of walking parameters shown in Table 1 as explanatory variables of a multiple regression equation for calculating an evaluation value of “Ikiiki-Tobotobo” and 6 kinds of walking parameters shown in Table 2 When using, as shown in Table 8, individual walking parameters and a single regression equation of “Ikiiki-Tobotobo” are obtained. In Table 8, the boundary value is a numerical value of the feature value of the walking corresponding to the evaluation value 50 when the evaluation value of “Iki-iki-tobo-tobo” is expressed by a numerical value from 0 (live-on side) to 100 (on-the-fly side).

  When the evaluation value of “Ikiiki-Tobotobo” of the subject is calculated by the single regression equation of Table 9, the value is the “estimated value” (before walking instruction) of Table 10. Therefore, the walking advisor determines GOOD when the estimated value based on the single regression equation of the subject is less than 50, and BAD when the estimated value is 50 or more. The estimated value in Table 10 (before walking instruction) and the determination result are It was presented to the subject together with the evaluation value calculated by the regression equation. Then, in order for the gait of the subject to be on the “lively” side in the evaluation axis of “lively”, the estimated value of “lively—bottlebo” for each walking parameter by a single regression equation is less than 50. The subject was advised to keep walking.

Then, when the characteristic amount of the subject's walking was measured again, the measured values shown in Table 9 (after walking guidance) were obtained. Table 9 also shows the estimated value of “Ikiiki-Tobotobo” calculated from the single regression equation using the measured values, and the judgment result of the quality of the estimated values.

  From Table 10, it can be seen that after walking instruction, the estimated value calculated by the single regression equation is GOOD determination for all walking parameters. From this, it can be seen that the subject understood the meaning of the evaluation value of “Iki-Iki-Tobotobo” calculated from the multiple regression equation, tried to improve walking so that “Iki-Iki” could be seen, and the result appeared.

<< Walking improvement tool >>
In order to convey the advice content by the above-mentioned walking advisor to the subject more clearly, the subject on a predetermined evaluation axis according to the method of the present invention, for example, as shown in FIG. 10 on a paper surface or a screen of a personal computer or a portable terminal. Gait features used in the multiple regression equation when evaluating gait impressions, estimated evaluation values calculated from the subject's gait feature parameters using a single regression equation, and boundaries determined by the single regression equation Value and the judgment result of pass / fail determined by the comparison between this estimated value and the boundary value, the evaluation value of the impression calculated by the multiple regression equation is shown as the total score on the evaluation axis, and the gait is improved on the evaluation axis Advice for doing so may be displayed. This advice is based on the multiple gait feature values used when calculating the gait impression evaluation value using multiple regression equations, and the gait calculated using a single regression equation for each gait feature value. The estimated value of the impression and the advice content corresponding to the boundary value may be accumulated in advance, and may be appropriately selected and displayed.

  Moreover, as shown in FIG. 11, the meaning of the feature amount of walking may be displayed on the above-described paper or screen as needed.

1 stride 2 step 3 walking angle 4 stride 5 toe angle 6 stance ratio 7 swing period ratio 8 both legs support ratio 10 vertical line θ1 head tilt angle θ2 spine angle θ3R right shoulder rotation angle θ4R right waist rotation angle θ5 side Angle formed by a straight line connecting the knee and waist and a straight line connecting the knee and buttocks (visual range of knee joint)

Claims (8)

  The evaluation value of the gait impression of any subject can be calculated using a multiple regression equation with the evaluation value of the gait impression as the objective variable and the walking parameters including the walking feature values obtained by walking measurement as explanatory variables. A gait evaluation method to be obtained, in which an evaluation value of a gait impression is represented by an evaluation axis having a pair of sensitivity words representing an impression of the opposite electrode with respect to the gait, and at least one of the pair of sensitivity words has an onomatopoeia. Evaluation method that contains.   The evaluation method according to claim 1, wherein an evaluation value of a gait impression is obtained with two or more evaluation axes in which at least one of the sensitivity words of both poles is different.   The evaluation method according to claim 1 or 2, wherein onomatopoeia is included in both of the pair of sensitivity words.   The evaluation value of the gait impression of the subject is obtained using a single regression equation of the evaluation value of the gait impression and the characteristic amount of the walking included in the multiple regression equation. Evaluation method.   The evaluation method according to any one of claims 1 to 4, wherein an evaluation value of a gait impression used to create a multiple regression equation is obtained by a VAS (Visual Analogue Scale).   Walking feature value obtained by measurement using foot pressure sensor, walking feature value obtained by measurement using motion capture, and walking feature value obtained by measurement using acceleration sensor The evaluation method according to any one of claims 1 to 5, which is selected from:   Walking feature value obtained by measurement using foot pressure sensor, walking feature value obtained by measurement using motion capture, and walking feature value obtained by measurement using acceleration sensor The evaluation method according to claim 6, wherein there are two or more of the above.   The evaluation method according to claim 1, wherein an average value of the left and right walking feature values is used when there is a left-right difference in the walking feature values.

Images