JP2012075708A - Stress state estimation device, stress state estimation method, program, and recording medium - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a technique for estimating a stress state in consideration of the individual difference and in-individual change of body motion data concerning the technique for estimating the stress state of a subject, based on the body motion data.SOLUTION: A stress state estimation device includes: a measurement means for measuring the body motion of a subject so as to acquire body motion data expressing the body motion; a calculation means for extracting a feature amount from the body motion data and calculating intermediate data; a storage means for storing relation data expressing correlation to be estimated between the intermediate data and the stress state of the subject; an estimation means for estimating the stress state of the subject through the use of the relation data, based on the intermediate data; an input means for receiving the input of consciousness stress state data expressing the stress state which is recognized by the subject; and an updating means for inductively obtaining correlation between the intermediate data and the consciousness stress state data through the use of a pair of the intermediate data and the consciousness stress state data when the consciousness stress state data is inputted, and storing the correlation in the storage means as the relation data.

Description

  The present invention relates to a stress state estimation technique for estimating a subject's stress state, and more particularly to a technique for estimating a stress state based on body motion data of a subject.

  In recent years, mental illnesses associated with psychological stress (hereinafter simply referred to as stress) such as depression or sleep disorder and chronic fatigue syndrome are increasing and becoming a social problem. In the prevention of such mental illnesses, it is important to control stress on a daily basis. In order to appropriately perform stress control, it is preferable that the stress state (degree of stress) can be easily and accurately grasped.

  In other words, the stress state refers to the degree of feeling depressed and expresses the mental state, but it naturally includes the case of mental fatigue as a result of physical fatigue or malfunction. It is

  Such a stress state has been conventionally grasped by an inquiry. For example, Patent Document 1 describes a portable biological information collection device that can be worn on a human body for collecting biological information and mental information in time series. With the technique described in Patent Document 1, it is possible to continuously measure biological information in daily life and efficiently collect mental states at that time by timely inquiry. However, the technique for grasping the stress state by the interview cannot grasp the change of the stress state between the interviews.

  On the other hand, in recent years, research on the estimation of the stress state based on the body movement of the subject has been advanced (see Non-Patent Documents 1 and 2).

  Patent Document 2 discloses a fatigue estimation device that estimates a user's fatigue level, and continuously detects the user's activity frequency as an activity level, and estimates the user's fatigue level based on the activity level. A fatigue estimation device is described.

JP 2003-290176 A (released on October 14, 2003) International Application Publication No. WO2007 / 138930 (published December 6, 2007)

Nakamura, T. , K. Kiyono, K. Yoshiuchi, R. Nakahara, Z. R. Struzik, and Y. Yamamoto. Universal Scaling law in human behavioral organization. Physical Review Letters 99: 138103-1-4, 2007. Nakamura, T. , M. Sone, N. Aoyagi, Z. R. Struzik, and Y. Yamamoto. Association of local statistics of locomotor activity with momentary depressive mood. Proceedings of the IFMBE / IMIA 6th International Workshop on Biosignal Interpretation, pp. 48-51, 2009.

  It is preferable that the stress state of the subject can be estimated from the body movement data because the stress state can be easily grasped. However, body movement data not only has great individual differences depending on physical factors such as height and weight, and personal movement habits, sensor wearing conditions, etc. Changes over time, such as changes in body weight or muscle weakening with age, are also significant. Therefore, in the technique for estimating the stress state of a subject from body movement data, a technique for estimating the stress state in consideration of individual differences and intra-individual changes in body movement data is required.

  However, the technique described in Patent Document 2 is a technique for estimating the degree of fatigue, and does not consider individual differences and intra-personal changes in body motion data as described above. The present invention has been made in view of such circumstances, and in a technique for estimating a subject's stress state from body motion data, the stress state is estimated in consideration of individual differences and intra-individual changes in body motion data. The main purpose is to provide technology for this purpose.

  In order to solve the above problems, a stress state estimation device according to the present invention is a stress state estimation device that estimates a stress state of a subject from body motion data representing the body motion of the subject, An acquisition unit to acquire, a calculation unit to extract feature data from the body motion data acquired by the acquisition unit to calculate intermediate data, an intermediate data calculated by the calculation unit, and an estimation between the stress state of the subject Storage means for storing the relationship data representing the correspondence relationship, and estimation means for estimating the stress state of the subject from the intermediate data calculated by the calculation means using the relationship data stored in the storage means; An input means for accepting input of subjective stress state data representing a stress state perceived by the subject, and when the subjective stress state data is input to the input means, the calculation Using the intermediate data calculated by the stage and the set of subjective stress state data input to the input means, the data indicating the correspondence between the intermediate data and the subjective stress state data is obtained inductively. Update means for storing data representing the correspondence relationship in the storage means as the relation data.

  According to said structure, the said stress state estimation apparatus acquires body movement data, and features (for example, the cumulative probability of a subject's rest time frequency, the average value of body movement data, a standard from the acquired body movement data (E.g. statistics such as deviation, skewness, kurtosis, variance, etc.) are extracted to calculate intermediate data, and relational data (e.g., regression line or regression curve) representing the estimated correspondence between the intermediate data and the stress state The stress state of the subject is estimated from the calculated intermediate data using a coefficient, a learning parameter of a machine learner, and the like.

  Here, the stress state estimation apparatus uses the set of the calculated intermediate data and the input subjective stress state data when the subjective stress state data is input to the input unit, and uses the set of intermediate data and the subjective stress state data. Inductively find the correspondence that exists with the data. For example, regression analysis, supervised machine learning, or the like can be used as an inductive method. Then, the above-mentioned relationship data is updated with the obtained correspondence relationship as the estimated correspondence relationship between the intermediate data and the stress state.

  As described above, the stress state estimation device recursively obtains a correspondence relationship between the intermediate data and the subjective stress state data, and estimates a stress state from body motion data using the correspondence relationship. Since the estimated correspondence between the intermediate data and the stress state is stored in the storage means, the stress state is estimated based on the acquired body motion data even when no subjective stress state data is input. It can be performed at any time or continuously. Furthermore, since the correspondence between the intermediate data and the subjective stress state data is acquired from the subject person at any time, accurate estimation of the stress state is possible even if there are individual differences or intra-personal changes in body motion data. It can be carried out. As described above, according to the stress state estimation device, it is possible to estimate a subject's stress state in consideration of individual differences and intra-individual changes in body motion data.

  Note that Patent Document 2 describes that the actual fatigue level information obtained as a result of the inquiry is associated with estimated fatigue level information indicating the estimated fatigue level. Is also not described for use in estimating the fatigue level. Patent Document 2 describes that a coefficient in an arithmetic expression for obtaining a fatigue level and an algorithm for estimating the fatigue level are dynamically calibrated and corrected to improve the fatigue level estimation accuracy. However, it does not describe how to dynamically calibrate and correct the coefficient and algorithm based on the evidence regarding the relationship between stress state and body movement data as listed in Non-Patent Documents 1 and 2. It is also not intended to take into account individual differences and intra-individual changes in body motion data. Therefore, the present invention is different from the technique described in Patent Document 2.

  In the stress state estimation device, when the stress state estimation device estimates the stress state of the subject at an arbitrary timing, the calculation means acquires the body acquired by the measurement means within a predetermined time including the arbitrary timing. Intermediate data is calculated from the movement data, and the updating means calculates the intermediate data calculated by the calculation means from the body movement data acquired by the measurement means within a predetermined time including the timing when the subjective stress state data is input, Further, the correspondence relationship between the intermediate data and the subjective stress state data may be obtained recursively using the set of the subjective stress state data.

  According to said structure, a test subject's stress state can be estimated successfully.

  The stress state estimation device further includes storage means for storing and storing a set of the intermediate data and the subjective stress state data when the subjective stress state data is input to the input means, and the updating means includes: By performing regression analysis using the subjective stress state data as an objective variable and the intermediate data as an explanatory variable for the set of the intermediate data and the subjective stress state data stored in the storage means, intermediate data It is preferable to obtain data representing the correspondence between the consciousness stress state data and the subjective stress state data.

  According to said structure, the said relationship data can be calculated | required successfully by regression analysis.

  In the stress state estimation device, it is preferable that the storage unit discards the set of the intermediate data and the subjective stress state data stored for a predetermined period.

  According to said structure, since a test subject's stress state can be estimated based on the latest measurement data, the influence of the in-person change of the tendency of the body motion data in a test subject is excluded, and a stress state is suitably displayed. Can be estimated.

  In the stress state estimation device, the updating unit includes a machine learner that performs supervised machine learning, and the relationship data and the data representing the correspondence are learning parameters of the machine learner, and the machine learner Is a method for recursively obtaining data representing the correspondence between the intermediate data and the subjective stress state data by performing supervised learning using the subjective stress state data as teacher data and the intermediate data as input data. There may be.

  According to said structure, the said relationship data can be calculated | required successfully by machine learning.

  In the stress state estimation device, the machine learner is preferably a multilayer perceptron.

  According to said structure, the said relationship data can be calculated | required successfully by machine learning.

  In the stress state estimation device, the feature amount may be a length and frequency of rest of the subject.

  According to the above configuration, as shown in Non-Patent Document 1, since there is a relationship between the length and frequency of the subject's rest and the subject's stress state, the subject's stress state is successfully estimated. be able to.

  In the stress state estimation device, the calculation means determines whether the body motion data indicates activity or rest of the subject by comparing a predetermined threshold with the body motion data, and a determination result It is preferable to extract the length and frequency of the rest of the subject based on the above.

  According to said structure, the length and frequency of the said test subject's rest can be extracted successfully.

  In the stress state estimation device, the intermediate data may be stress evaluation value data calculated from the length and frequency of the subject's rest, and the calculating means is configured to allow the subject to rest for a certain length or more. The regression analysis using a linear regression line with the logarithm of the cumulative probability of rest time frequency, which is the probability of occurrence, as the objective variable and the logarithm of the certain length as the explanatory variable, and based on the slope of the obtained regression line Stress evaluation value data may be calculated.

  According to said structure, a test subject's stress state can be estimated successfully by using the above stress evaluation value data as intermediate data.

  In the stress state estimation device, the feature amount may be two or more types of statistics of the body motion data, and the statistics are a group including an average value, a standard deviation, a skewness, a kurtosis, and a variance. It may be more selected.

  According to the above configuration, as shown in Non-Patent Document 2, since there is a correlation between the statistical amount of the body movement data of the subject and the stress state of the subject, the stress state of the subject is successfully estimated. be able to.

  The stress state estimation device includes a trend removing unit that removes a trend from the body movement data, and the calculation unit extracts one or more types of statistics from the body movement data, and the trend removing unit extracts a trend. One or more types of statistics may be extracted from the removed body motion data.

  According to the above configuration, the stress state of the subject can be estimated more accurately by selecting for each statistic whether to use body movement data or trend removal body movement data in the calculation of the statistic. .

  The stress state estimation device includes a trend removing unit that removes a trend from the body movement data, and the calculation unit calculates the intermediate data from the body movement data from which the trend has been removed by the trend removing unit. There may be.

  According to the above configuration, since intermediate data can be calculated based on body movement data from which the trend has been removed, it is possible to more accurately estimate the stress state by eliminating trends that may be caused by factors other than stress. Can do.

  The stress state estimation method according to the present invention is a stress state estimation method in which a stress state estimation device provided with storage means estimates the subject's stress state based on the subject's body motion, A measurement step of measuring and acquiring body motion data representing the body motion, a calculation step of calculating feature values from the body motion data acquired in the measurement step, and calculating intermediate data; and calculating in the calculation step An estimation step for estimating a stress state of the subject using relationship data representing an estimated correspondence relationship between the intermediate data stored in the storage means and the stress state of the subject from the intermediate data; An input step for accepting input of subjective stress state data representing a stress state perceived by the subject, the intermediate data when the subjective stress state data is input, and Using the set of the subjective stress state data, the data representing the correspondence between the intermediate data and the subjective stress state data is obtained recursively, and the data representing the obtained correspondence is stored as the relation data And an update process to be stored.

  According to said method, there exists an effect equivalent to the stress condition estimation apparatus which concerns on this invention.

  Further, a program for operating the apparatus according to the present invention, which causes a computer to realize the functions of each of the above apparatuses, and a computer-readable recording medium recording the program are also included in the scope of the present invention.

  ADVANTAGE OF THE INVENTION According to this invention, a test subject's stress state can be continuously estimated from a test subject's body motion data in consideration of the individual difference and intra-individual change of body motion data.

It is the block diagram which showed the structure of the stress evaluation system which concerns on one Embodiment of this invention. It is the flowchart which showed the process of the stress evaluation part which concerns on one Embodiment of this invention. It is the schematic diagram which showed the relationship between the body movement data and the subjective stress value data in the stress evaluation part which concerns on one Embodiment of this invention. It is the schematic diagram which showed the relationship between the body movement data in the stress evaluation part which concerns on one Embodiment of this invention, and activity state determination data. It is a figure explaining the calculation method of the stress judgment value in the stress evaluation part which concerns on one Embodiment of this invention. It is a figure explaining the calculation method of the parameter for stress index calculation in the stress evaluation part concerning one embodiment of the present invention. It is the block diagram which showed the structure of the stress evaluation system which concerns on one Embodiment of this invention. It is the flowchart which showed the process of the stress evaluation part which concerns on one Embodiment of this invention. It is the block diagram which showed the structure of the stress evaluation system which concerns on one Embodiment of this invention. It is the flowchart which showed the process of the stress evaluation part which concerns on one Embodiment of this invention. It is the block diagram which showed the structure of the stress evaluation system which concerns on one Embodiment of this invention. It is the flowchart which showed the process of the stress evaluation part which concerns on one Embodiment of this invention. It is the block diagram which showed the structure of the stress evaluation system which concerns on one Embodiment of this invention. It is the flowchart which showed the process of the stress evaluation part which concerns on one Embodiment of this invention. It is the block diagram which showed the structure of the stress evaluation system which concerns on one Embodiment of this invention. It is the flowchart which showed the process of the stress evaluation part which concerns on one Embodiment of this invention. It is the block diagram which showed the structure of the stress evaluation system which concerns on one Embodiment of this invention. It is the flowchart which showed the process of the stress evaluation part which concerns on one Embodiment of this invention.

[First Embodiment]
First, a stress index evaluation apparatus (stress state estimation apparatus) 100 according to an embodiment (first embodiment) of the present invention will be described with reference to FIGS.

  FIG. 1 is a functional block diagram showing the configuration of the stress index evaluation apparatus 100. The stress index evaluation apparatus 100 includes a stress evaluation unit 110, a body motion data measurement unit (measurement unit) 11, a subjective stress level input unit (input unit) 12, and a stress index output unit 13.

  The stress evaluation unit 110 receives body motion data representing the body motion of the subject from the body motion data measurement unit 11, and subjective stress level data (aware stress state data) representing a stress state perceived by the subject from the subjective stress level input unit 12. , And based on these, the stress index data (stress state estimation result) data of the subject is calculated, and this stress index data is output to the stress index output unit 13.

  The stress evaluation unit 110 includes a body motion data acquisition unit 20, an activity state determination calculation processing unit (calculation unit) 21, a rest time frequency cumulative probability calculation processing unit (calculation unit) 22, and a stress evaluation value calculation processing unit (calculation unit) 23. , Subjective stress level data acquisition unit 24, stress index calculation parameter update calculation processing unit (update unit) 25, stress index calculation processing unit (estimation unit) 26, stress index output unit 27, body motion data storage unit 40, activity state Determination data storage unit 41, rest time frequency cumulative probability data storage unit 42, stress evaluation value data storage unit (storage unit) 43, subjective stress degree data storage unit (storage unit) 44, stress index calculation parameter storage unit (storage unit) ) 45 and a stress index data storage unit 46.

  The body movement data measuring unit 11 is for measuring and collecting body movement data of a subject. Body motion data is time-series data representing the magnitude of body motion. The body motion data measurement unit 11 includes, for example, an acceleration sensor, an angular velocity sensor, or a positioning device, and is mounted on one or a plurality of locations of the subject's body, and measures the activity (body motion) of the mounting portion. Although it does not specifically limit as a mounting | wearing part, For example, an arm part, a waist | lumbar part, a leg part, a bodily sensation, a head, etc. may be sufficient, and it is preferable that it is especially a wrist. Further, an imaging device that images the subject may be provided, and the body motion of the subject may be detected by analyzing the captured image information.

  In this embodiment, acceleration time series data representing the acceleration of the subject collected at intervals of 0.5 seconds is used, but the time interval of collection and the type of data are not limited to this. The sampling time interval may be arbitrary, and in addition to the acceleration time series data, for example, zero cross time series data, acceleration integrated value time series data, etc. should be used if the data represents the magnitude of body movement. Is possible.

  The subjective stress level input unit 12 is for inputting the level of stress that the subject is aware of at any time. For example, a display device such as an LCD and an input device such as a keyboard or a touch panel are provided, and the subject is prompted to input the stress state that he / she is aware of by displaying “please input the psychological stress level that he / she is aware of”. Although the form which receives the input of the reply from can be considered, it is not limited to this. As for the input format of the subjective stress level data, 0 is set to the state of “not feeling stress at all”, and 100 is set to the state of “feeling severely stress”, so that the input is an integer from 0 to 100. However, the present invention is not limited to this, and it is only necessary to input the degree of the subjective stress with two or more values. For example, “I don't feel stress at all, I don't feel stress, I feel stress, I feel stress, I feel a lot of stress”, etc. You may make it choose what applies to a test subject.

  The stress evaluation unit 110 estimates a subject's stress state based on data acquired from the body motion data measurement unit 11 and the subjective stress level input unit 12 and calculates a stress index.

  The process of calculating the stress index by the stress evaluation unit 110 will be described with reference to FIG. FIG. 2 is a flowchart showing a processing process of the stress evaluation unit 110.

  First, the body motion data measurement unit 11 starts measuring body motion data M_data (step woman S0). The body motion data acquisition unit 20 of the stress evaluation unit 110 acquires the body motion data M_data from the body motion data measurement unit 11 (step S1) and stores it in the body motion data storage unit 40 (step S2).

  In the following description, the body motion data measured by the body motion data measurement unit 11 may be expressed as M_data. Also, the first measured body motion data after the start of measurement is M_data [0], the next measured body motion data is M_data [1], and the n + 1th measured body motion data is M_data. It may be written as [n].

  In addition, the conscious stress level data input to the conscious stress level input unit 12 may be referred to as “self_data”. The subjective stress level data input to the subjective stress level input unit 12 when the body motion data measurement unit 11 is measuring the (n + 1) th body motion data may be referred to as self_data [n].

  FIG. 3 shows M_data and self_data arranged in time series. As described above, in the present embodiment, the body motion data M_data is continuously collected at intervals of 0.5 seconds, and therefore M_data [n] is calculated by the body motion data measurement unit 11 after n / 2 seconds from the start of measurement. The measured body motion data is shown, and self_data [n] indicates the subjective stress degree data input to the subjective stress degree input unit 12 before and after n / 2 seconds from the start of measurement (for example, ± 0.25 seconds). .

  Subsequently, in steps S3 to S7, stress evaluation value data is calculated based on the body motion data stored in the body motion data storage unit 40. Hereinafter, as an example, a process when the stress evaluation unit 110 calculates the stress index data R_data [n] at the time of measuring the n-th body movement data M_data [n] from the start of measurement will be described. Needless to say, the stress evaluation unit 110 can calculate the stress index data at an arbitrary timing or continuously.

  In step S <b> 3, the activity state determination calculation processing unit 21 acquires body motion data from the body motion data storage unit 40, and determines whether the subject is active or resting based on these body motion data in time series. Activity state determination data to be expressed is calculated. In the following description, the activity state determination data calculated by the activity state determination calculation processing unit 21 may be expressed as A_data. Moreover, the activity state determination data corresponding to the body movement data M_data [n] may be expressed as A_data [n].

  FIG. 4 is a diagram illustrating step S3 of the processing flow. As shown in FIG. 4, in step S3, the activity state determination calculation processing unit 21 has X data before and after the body motion data M_data [n] measured nth from the start of measurement, that is, M_data [n -X] to M_data [n + X] are processed to calculate activity state determination data A_data [n−X] to A_data [n + X]. In the present embodiment, 1800 is used as X. However, the present invention is not limited to this, and X may be an integer, and may be appropriately set in consideration of the calculation amount and accuracy. Further, M_data that has been processed may be deleted from the body motion data storage unit 40 as appropriate.

  The activity state determination calculation processing unit 21 calculates the activity state determination data A_data by comparing the body movement data M_data with the activity state determination threshold M_th. In the present embodiment, the activity state determination threshold M_th is set to 0.1 G, but the present invention is not limited to this, and any value can be used as the activity state determination threshold M_th. If the body motion data M_data is greater than the activity state determination threshold M_th, it is determined that the body is in the active state. If the body motion data M_data is less than the activity state determination threshold M_th, it is determined that the body is in the resting state. In this embodiment, the active state determination data A_data is set to 1 when in an active state, and the active state determination data A_data is set to 0 when in a resting state. However, the present invention is not limited to this, and the active state determination data A_data is an active state. Any value that can distinguish between a state and a resting state may be used.

  The activity state determination calculation processing unit 21 stores the calculated activity state determination data in the activity state determination data storage unit 41 (step S4).

  In step S5, the rest time frequency cumulative probability calculation processing unit 22 acquires the activity state determination data A_data [n−X] to A_data [n + X] from the activity state determination data storage unit 41, and based on these activity state determination data. The rest time frequency cumulative probability (feature amount) data is calculated. In this specification, the rest time frequency cumulative probability P (rt ≧ x) refers to the probability of the total frequency that the time rt during which the rest state continues exceeds x. In other words, the rest time frequency cumulative probability P (rt ≧ x) refers to the probability that rest occurs continuously for x seconds or more. For example, in the activity state determination data A_data [n−X] to A_data [n + X], When the total rest time that has been continuous is y seconds, P (rt ≧ x) is y / X. The rest time frequency cumulative probability calculation processing unit 22 analyzes A_data [n−X] to A_data [n + X] and obtains a total rest time of x seconds or more continuous for x (0 <x ≦ X). , Rest time frequency cumulative probability data representing rest time frequency cumulative probability for each x is calculated.

  The rest time frequency cumulative probability calculation processing unit 22 stores the calculated rest time frequency cumulative probability data in the rest time frequency cumulative probability data storage unit 42 (step S6).

  In step S7, the stress evaluation value calculation processing unit 23 receives the rest time frequency cumulative probability data P (rt ≧ 1), P (rt ≧ 2),..., P (rt) from the rest time frequency cumulative probability data storage unit 42. ≧ X), and based on the acquired rest time frequency cumulative probability data, the stress evaluation value data (intermediate data) is calculated as follows (step S7).

  First, the stress evaluation value calculation processing unit 23 performs a regression analysis between the rest time frequency cumulative probability P (rt> x) and x. Specifically, the stress evaluation value calculation processing unit 23 performs log (P) for the rest time frequency cumulative probability data P (rt ≧ 1), P (rt ≧ 2),..., P (rt ≧ X), respectively. (Rt ≧ 1)), log (P (rt ≧ 2)),..., Log (P (rt ≧ X)), and the correspondence between P (rt ≧ x) and x Α1 is calculated by approximating the following equation (1) using a multiplication method.

log (P (rt ≧ x)) = α1 × log (x) + β1 (1)
In this embodiment, the least square method is used. However, the present invention is not limited to this, and a known algorithm for regression analysis can be used.

  FIG. 5A is a diagram illustrating an example in which log (P (rt ≧ x)) is plotted against log (x), and FIG. 5B is represented by the above formula (1). It is the figure which added the regression line 61. FIG.

  The stress evaluation value calculation processing unit 23 calculates stress evaluation value data based on the calculated α1. In the following description, the stress evaluation value data calculated by the stress evaluation value calculation processing unit 23 may be expressed as E_data. In addition, the stress evaluation value data corresponding to n seconds after the start of measurement may be expressed as E_data [n].

  In the present embodiment, the stress evaluation value calculation processing unit 23 calculates the stress evaluation value data E_data [n] based on the following calculation formula (2).

E_data [n] = − α1 (2)
The stress evaluation value calculation processing unit 23 stores the stress evaluation value data E_data [n] thus obtained in the stress evaluation value data storage unit 43 (step S8).

  In step S9, the conscious stress level data acquisition unit 24 checks whether the conscious stress level data self_data [n] is input to the conscious stress level input unit 12 (step S9). If the subjective stress degree data self_data [n] is input, the process proceeds to step S10, and the processes of steps S10 to S13 are performed. If the subjective stress degree data self_data [n] is not input, the process proceeds to step S14 without performing the processes of steps S10 to S13.

  In step S10, the conscious stress level data acquisition unit 24 acquires the conscious stress level data self_data [n] from the conscious stress level input unit 12. Subsequently, in step S11, the conscious stress level data acquisition unit 24 The degree data self_data [n] is stored in the subjective stress degree data storage unit 44.

  Here, the stress evaluation value data E_data [n] stored in the stress evaluation value data storage unit 43 in step S8, and the subjective stress level data self_data [n] stored in the subjective stress level data storage unit 44 in step S11. Are stored in each storage unit in association with each other (for example, with the same time stamp or ID attached), and by the stress index calculation parameter update calculation processing unit 25, the subjective stress degree data and the relevant stress level data are stored. It can be acquired as a set of stress evaluation value data corresponding to the subjective stress degree data.

  In the present embodiment, “a set of the subjective stress degree data and the stress evaluation value data corresponding to the subjective stress degree data” is described in detail. The subjective stress degree data and the subjective stress degree data are input. It is a set of stress evaluation value data corresponding to timing, and stress evaluation value data corresponding to an arbitrary timing is a stress evaluation value calculated from body movement data measured in a predetermined period including the arbitrary timing Point to.

  In step S12, the stress index calculation parameter update calculation processing unit 25 corresponds to the subjective stress degree data and the subjective stress degree data stored in the subjective stress degree data storage unit 44 and the stress evaluation value data storage unit 43. A set of stress evaluation value data (including a set of the subjective stress degree data self_data [n] stored in step S10 and the stress evaluation value data E_data [n] corresponding thereto) is acquired. Subsequently, the stress index calculation parameter update calculation processing unit 25 uses the least square method for the acquired set of the correspondence relationship between the subjective stress degree data and the stress evaluation value data corresponding to the subjective stress degree data. The stress index calculation parameters (relation data) α2 and β2 are calculated by performing linear approximation using the following equation (3).

self_data = α2 × E_data + β2 (3)
In this embodiment, the least square method is used. However, the present invention is not limited to this, and a known algorithm for regression analysis can be used.

  FIG. 6A is a diagram showing an example of the case where the subjective stress degree data is plotted against the stress evaluation value data corresponding to the subjective stress degree data, and FIG. ) To which a regression line 71 represented by the above equation (3) is added.

  Note that the set of the subjective stress level data used for the linear approximation and the stress evaluation value data corresponding to the subjective stress level data is stored in the subjective stress level data storage unit 44 and the stress evaluation value data storage unit 43 within a predetermined period. It is preferable to limit to those accumulated in (for example, the last three months). Thereby, the parameter for stress index calculation corresponding to a test subject's aged constitution change can be obtained. The predetermined period can be set arbitrarily.

  Subsequently, in step S13, the stress index calculation parameter update calculation processing unit 25 uses the calculated stress index calculation parameters α2 and β2 to store the stress index calculation parameter α2 stored in the stress index calculation parameter storage unit 45. And β2 are updated. That is, the stress index calculation parameter storage unit 45 stores stress index calculation parameters that are set in advance or updated by the stress index calculation parameter update calculation processing unit 25, and the subjective stress degree data is stored in the stress index calculation parameter storage unit 45. The stress index calculation parameter is updated each time it is input to the subjective stress level input unit 12. As the stress index calculation parameter set in advance, for example, an average value of a plurality of subjects measured in advance can be appropriately used.

  In step S 14, the stress index calculation processing unit 26 acquires the stress evaluation value data E_data [n] from the stress evaluation value data storage unit 43, and the latest stress index calculation parameter from the stress index calculation parameter storage unit 45. α2 and β2 are acquired, and stress index data (stress state estimation result data) is calculated. In the following description, the stress index data calculated by the stress index calculation processing unit 26 may be expressed as R_data. Further, the stress index data corresponding to the measurement of the body motion data M_data [n] may be expressed as R_data [n].

  The stress index calculation processing unit 26 calculates the stress index data R_data [n] by the following calculation formula (4).

R_data [n] = α2 × E_data [n] + β2 (4)
The stress index calculation processing unit 26 stores the calculated stress index data R_data [n] in the stress index data storage unit 46 (step S15).

  Thus, the process of calculating the stress index by the stress evaluation unit 110 ends (step S16).

  In the above, in calculating the correspondence between the subjective stress level data and the stress evaluation value data corresponding to the subjective stress level data, linear regression is used (approximate by a linear expression). Curve regression may be used. That is, instead of the above formulas (3) and (4), a formula representing a curve may be used, and a parameter included in the formula representing the curve may be used as a stress index calculation parameter.

  The stress index output unit 13 includes a display device such as an LCD or an audio output device such as a speaker. The stress index output unit 13 acquires the stress index calculated by the stress evaluation unit 110 via the stress index output unit 27 of the stress evaluation unit 110 and outputs it. (Display, audio output, etc.).

  In the stress index evaluation apparatus 100, various storage units for storing data can be realized by a RAM (Random Access Memory) such as a flash memory or a hard disk, but is not limited thereto. Further, in the stress index evaluation apparatus 100, various acquisition units, output units, calculation units, processing units, and the like that operate or process data can be realized by, for example, independent circuits. For example, it may be a virtual circuit realized by an arithmetic processing circuit such as a computer. In addition, these processing units are provided in the stress index evaluation apparatus 100, but these processing units do not have to be all in the same device, some treatment units are in other devices, Data may be exchanged between apparatuses using information transmission means such as a network.

[Second Embodiment]
A stress index evaluation apparatus (stress state estimation apparatus) 200 according to one embodiment (second embodiment) of the present invention will be described with reference to FIGS. For convenience of explanation, explanation of the same parts as in the first embodiment will be omitted, and only different parts will be explained.

  FIG. 7 is a block diagram showing a configuration of the stress index evaluation apparatus 200. The stress index evaluation apparatus 200 includes a stress evaluation unit 210 instead of the stress evaluation unit 110.

  Compared with the stress evaluation unit 110, the stress evaluation unit 210 does not include the stress evaluation value calculation processing unit 23 and the stress evaluation value data storage unit 43. Further, instead of the stress index calculation parameter update calculation processing unit 25 and the stress index calculation parameter storage unit 45, a learning parameter update calculation processing unit (update unit) 28 and a learning parameter storage unit (storage unit) 48 are provided. ing.

  In the stress evaluation unit 210, body motion data storage unit 40, activity state determination data storage unit 41, rest time frequency cumulative probability data storage unit 42, subjective stress degree data storage unit 44, learning parameter storage unit 48, and stress index data storage The unit 46 can be realized by a RAM (Random Access Memory) such as a flash memory or a hard disk, but is not limited thereto.

  In the stress evaluation unit 210, the body motion data acquisition unit 20, the activity state determination calculation processing unit 21, the rest time frequency cumulative probability calculation processing unit 22, the subjective stress degree data acquisition unit 24, the learning parameter update calculation processing unit 28, the stress The exponent operation processing unit 26 and the stress index output unit 27 can be realized by, for example, independent circuits, but are not limited thereto, and are, for example, virtual circuits realized by an arithmetic processing circuit such as a computer. Also good. In the present invention, these processing units are provided in the stress evaluation unit 210, but these processing units do not have to be all in the same device, and some treatment units are in other devices, Data may be exchanged between a plurality of devices using an information transmission means such as a network.

  A process of calculating the stress index by the stress evaluation unit 210 will be described with reference to FIG. FIG. 8 is a flowchart showing a processing process of the stress evaluation unit 210.

  The body motion data measurement unit 11 starts measuring the body motion data M_data, and transmits the measured body motion data M_data to the stress evaluation unit 210 (step S0).

  The body motion data acquisition unit 20 of the stress evaluation unit 110 acquires the body motion data M_data from the body motion data measurement unit 11 (step S1), and stores the body motion data M_data in the body motion data storage unit 40 (step S2). ).

  Hereinafter, as an example, a process when the stress evaluation unit 210 calculates the stress index data R_data [n] at the time of measuring the n-th body movement data M_data [n] from the start of measurement will be described. Needless to say, the stress evaluation unit 210 can calculate the stress index data at an arbitrary timing or continuously.

  The activity state determination calculation processing unit 21 acquires body motion data M_data [n−X] to M_data [n + X] from the body motion data storage unit 40, and activity state determination data A_data [n−X] based on these body motion data. ] To A_data [n + X] are calculated (step S3) and stored in the activity state determination data storage unit 41 (step S4). In this embodiment, 1800 is used as X. However, the present invention is not limited to this, and X may be an integer.

  Also in this embodiment, the active state determination threshold M_th is set to 0.1 G, the active state determination data A_data is 1 when in the active state, and the active state determination data A_data is 0 when in the resting state.

  The rest time frequency cumulative probability calculation processing unit 22 acquires the activity state determination data A_data [n−X] to A_data [n + X] from the activity state determination data storage unit 41, and based on these activity state determination data, Rest time frequency cumulative probability data (P (rt ≧ x), 0 <x ≦ X: intermediate data) indicating the cumulative probability (feature value) of the rest time frequency is calculated (step S5), and the rest time frequency cumulative probability The data is stored in the data storage unit 42 (step S6).

  The conscious stress level data acquisition unit 24 checks whether the conscious stress level data self_data [n] is input from the conscious stress level input unit 12 (step S7). If the subjective stress degree data self_data [n] is input, the process proceeds to step S8, and the processes of steps S8 to S11 are performed. If the subjective stress degree data self_data [n] is not input, the process proceeds to step S12 without performing the processes of steps S8 to S11.

  In step S <b> 8, the conscious stress level data acquisition unit 24 acquires conscious stress level data self_data [n] from the conscious stress level input unit 12. In step S <b> 9, the conscious stress level data acquisition unit 24 stores the acquired conscious stress level data self_data [n] in the conscious stress level data storage unit 44.

  In step S <b> 10, the stress evaluation unit 210 performs machine learning using the rest time frequency cumulative probability data as input and the subjective stress level data as a teacher. The learning parameter update calculation processing unit 28 acquires learning parameters (relation data) such as weighting factors that define the internal state of the machine learning device from the learning parameter storage unit 48. Then, the learning parameter update calculation processing unit 28 uses the subjective stress level data self_data [n] acquired from the subjective stress level data storage unit 44 as a teacher, and the rest time frequency cumulative probability acquired from the rest time frequency cumulative probability data storage unit 42. Data (corresponding to the nth body movement data from the start of measurement. More specifically, data calculated from body movement data within a predetermined width including the measurement time of the body movement data) Machine learning.

  In the present embodiment, a multi-layer perceptron using back propagation is used as the machine learner, but the present invention is not limited to this, and a known machine learner that performs supervised machine learning can be used.

  The learning parameter update calculation processing unit 28 stores the updated learning parameter as a result of the machine learning in the learning parameter storage unit 48 (step S11).

  In step S12, the stress index calculation processing unit 26 acquires the latest learning parameter from the learning parameter storage unit 48, and calculates the stress index data R_data [n] using the same machine learner as described in step S10. To do. That is, the stress index calculation processing unit 26 inputs the output obtained by inputting the rest time frequency cumulative probability data acquired from the rest time frequency cumulative probability data storage unit 42 to the machine learner as stress index data R_data [n]. To do.

  The stress index calculation processing unit 26 stores the calculated stress index data R_data [n] in the stress index data storage unit 46 (step S13).

  Thus, the process of calculating the stress index by the stress evaluation unit 210 ends (step S14).

[Third Embodiment]
A stress index evaluation apparatus (stress state estimation apparatus) 300 according to an embodiment (third embodiment) of the present invention will be described with reference to FIGS. For convenience of explanation, the description of the same parts as those in the first and second embodiments will be omitted, and only different parts will be described.

  FIG. 9 is a block diagram showing a configuration of the stress index evaluation apparatus 300. The stress index evaluation apparatus 300 includes a stress evaluation unit 310 instead of the stress evaluation unit 110.

  As shown in FIG. 3, the stress evaluation unit 310 includes a body motion data acquisition unit 20, a statistic calculation processing unit (calculation unit) 29, a subjective stress level data acquisition unit 24, and a learning parameter update calculation processing unit (update unit) 28. , A stress index calculation processing unit (estimating means) 26, a stress index output unit 27, a body motion data storage unit 40, a statistic data storage unit 47, a subjective stress degree data storage unit 44, a learning parameter storage unit (storage unit) 48, And a stress index data storage unit 46.

  In the stress evaluation unit 310, the body motion data storage unit 40, the statistic data storage unit 49, the subjective stress level data storage unit 44, the learning parameter storage unit 48, and the stress index data storage unit 46 are, for example, a flash memory, a hard disk, or the like. Although it is realizable with RAM (random access memory), it is not limited to these.

  In the stress evaluation unit 310, the body motion data acquisition unit 20, the statistic calculation processing unit 29, the subjective stress level data acquisition unit 24, the learning parameter update calculation processing unit 28, the stress index calculation processing unit 26, and the stress index output unit 27. Can be realized by, for example, independent circuits, but is not limited thereto, and may be a virtual circuit realized by an arithmetic processing circuit such as a computer. Further, in the present invention, these processing units are provided in the stress evaluation unit 310, but it is not necessary that all these processing units are in the same device, some treatment units are in other devices, Data may be exchanged between a plurality of devices using an information transmission means such as a network.

  The process of calculating the stress index by the stress evaluation unit 310 will be described with reference to FIG. FIG. 10 is a flowchart showing a processing process of the stress evaluation unit 310.

  The body motion data measurement unit 11 starts measuring the body motion data M_data, and transmits the measured body motion data M_data to the stress evaluation unit 310 (step S0).

  The body motion data acquisition unit 20 of the stress evaluation unit 310 acquires the body motion data M_data from the body motion data measurement unit 11 (step S1), and stores the body motion data in the body motion data storage unit 40 (step S2). .

  Hereinafter, as an example, a process when the stress evaluation unit 310 calculates the stress index data R_data [n] at the time of measuring the n-th body movement data M_data [n] from the start of measurement will be described. Needless to say, the stress evaluation unit 310 can calculate the stress index data at an arbitrary timing or continuously.

  The statistic calculation processing unit 29 acquires body motion data M_data [n−X] to M_data [n + X] from the body motion data storage unit 40, and calculates and processes statistics (features) from these body motion data. Statistical data (intermediate data) is calculated (step S3). In this embodiment, 1800 is used as X, but the present invention is not limited to this, and X may be an integer.

  In the present embodiment, the average value, standard deviation, skewness, and kurtosis of the body motion data M_data are calculated as statistics, but the present invention is not limited to this, and two or more of the above may be calculated. Also, variance can be used instead of standard deviation.

  The statistic calculation processing unit 29 stores the calculated various statistic data in the statistic data storage unit 49 (step S4).

  The conscious stress level data acquisition unit 24 checks whether the conscious stress level data self_data [n] is input from the conscious stress level input unit 12 (step S5). If the subjective stress degree data self_data [n] is input, the process proceeds to step S6, and the processes of steps S6 to S9 are performed. If the subjective stress degree data self_data [n] is not input, the process proceeds to step S10 without performing the processes of steps S6 to S9.

  In step S <b> 6, the conscious stress level data acquisition unit 24 acquires conscious stress level data self_data [n] from the conscious stress level input unit 12. In step S <b> 7, the conscious stress level data acquisition unit 24 stores the acquired conscious stress level data self_data [n] in the conscious stress level data storage unit 44.

  In step S <b> 8, the stress evaluation unit 310 receives the statistical data and performs machine learning using the subjective stress degree data as a teacher. The learning parameter update calculation processing unit 28 acquires learning parameters (relation data) such as weighting factors that define the internal state of the machine learning device from the learning parameter storage unit 48. Then, the learning parameter update calculation processing unit 28 uses the consciousness stress degree data self_data [n] obtained from the consciousness stress degree data storage unit 44 as a teacher to obtain the statistic data (n from the start of measurement). Machine learning corresponding to the second body motion data, more specifically, the body motion data calculated from the body motion data within a predetermined period including the measurement time of the body motion data.

  Also in this embodiment, a multi-layer perceptron using back propagation is used as the machine learning device, but the machine learning device is not limited to this, and a known machine learning device that performs supervised machine learning can be used.

  The learning parameter update calculation processing unit 28 stores the updated learning parameter as a result of machine learning in the learning parameter storage unit 48 (step S9).

  In step S10, the stress index calculation processing unit 26 acquires the latest learning parameter from the learning parameter storage unit 48, and calculates the stress index data R_data [n] using the same machine learner as described in step S8. To do. That is, the stress index calculation processing unit 26 sets the output obtained by inputting the statistics data acquired from the statistics data storage unit 49 to the machine learner as stress index data R_data [n].

  The stress index calculation processing unit 26 stores the calculated stress index data R_data [2n] in the stress index data storage unit 46 (step S11).

  Thus, the process of calculating the stress index by the stress evaluation unit 310 ends (step S12).

[Fourth Embodiment]
A stress index evaluation apparatus (stress state estimation apparatus) 400 according to an embodiment (fourth embodiment) of the present invention will be described with reference to FIGS. For convenience of explanation, the description of the same parts as those in the first to third embodiments will be omitted, and only different parts will be described.

  FIG. 11 is a block diagram illustrating a configuration of the stress index evaluation apparatus 400. The stress index evaluation apparatus 400 is for the stress index evaluation apparatus 100 according to the first embodiment. A stress evaluation unit 410 is provided instead of the stress evaluation unit 110. The stress evaluation unit 410 includes all members of the stress evaluation unit 110, and further includes a trend removal calculation processing unit (trend removal unit) 30 and a trend removal data storage unit 50.

  The process of calculating the stress index by the stress evaluation unit 410 will be described with reference to FIG. FIG. 12 is a flowchart showing a processing process of the stress evaluation unit 410.

  The body motion data measurement unit 11 starts measuring the body motion data M_data, and transmits the measured body motion data M_data to the stress evaluation unit 410 (step S0).

  The body motion data acquisition unit 20 of the stress evaluation unit 410 acquires the body motion data M_data from the body motion data measurement unit 11 (step S1), and stores the body motion data in the body motion data storage unit 40 (step S2). .

  Hereinafter, as an example, a process when the stress evaluation unit 410 calculates the stress index data R_data [n] at the time of measuring the n-th body movement data M_data [n] from the start of measurement will be described. Needless to say, the stress evaluation unit 410 can calculate the stress index data at an arbitrary timing or continuously.

  The trend removal calculation processing unit 30 acquires body motion data M_data [n−X] to M_data [n + X] from the body motion data storage unit 40, performs a trend removal process, and calculates body motion data after the trend removal ( Step S3). This is because, for example, if the subject is exercising or performing work while measuring body motion data, the body motion due to the motion or work becomes a disturbance and the stress evaluation may not be performed correctly. Because there is. By performing the trend removal process, this influence can be avoided. In this embodiment, 1800 is used as X. However, the present invention is not limited to this, and X may be an integer.

  In the present embodiment, the trend removal calculation processing unit 30 performs a Fourier transformation and performs a trend removal process by cutting a frequency of 2 Hz or less. The trend removal process only needs to reduce the macroscopic trend of fetal movement data, and the method to be used is not limited. For example, it is possible to perform a trend removal process by performing orthogonal transform such as Fourier transform and discrete cosine transform to cut a low frequency region of body motion data or lowering the gain. Further, for example, it is also possible to perform a trend removal process by obtaining a regression line or regression curve of body motion data and subtracting the value shown in the regression line or regression curve from the body motion data. In this case, the method for obtaining the regression line or the regression curve is not limited. For example, the least square method may be used to fit a function having a specific order. The regression line or regression curve may be divided into small sections and obtained for each section.

  The trend removal calculation processing unit 30 stores the calculated trend removal body motion data in the trend removal data storage unit 50 (step S4).

  The activity state determination calculation processing unit 21 acquires the trend removal body motion data from the trend removal data storage unit 50, and creates the activity state determination data from the trend removal body motion data (step S5).

  The processing of step S5 to step S18 is the same as step S3 to step S16 in the first embodiment except that the trend-removed body motion data is used instead of the body motion data, and thus description thereof is omitted here. .

  Since the stress evaluation unit 410 uses the trend-removed body motion data as an alternative to the body motion data, the stress evaluation unit 410 can calculate the stress index more accurately.

[Fifth Embodiment]
A stress index evaluation apparatus (stress state estimation apparatus) 500 according to an embodiment (fifth embodiment) of the present invention will be described with reference to FIGS. For convenience of explanation, explanation of the same parts as those in the first to fourth embodiments will be omitted, and only different parts will be explained.

  FIG. 13 is a block diagram illustrating a configuration of the stress index evaluation apparatus 500. The stress index evaluation device 500 includes a stress evaluation unit 510 instead of the stress evaluation unit 210 with respect to the stress index evaluation device 200 according to the second embodiment. The stress evaluation unit 510 includes all members of the stress evaluation unit 210, and further includes a trend removal calculation processing unit (trend removal unit) 30 and a trend removal data storage unit 50.

  The process by which the stress evaluation unit 510 calculates the stress index will be described with reference to FIG. FIG. 14 is a flowchart showing the process of the stress evaluation unit 510.

  The body motion data measurement unit 11 starts measuring body motion data M_data, and transmits the measured body motion data M_data to the stress evaluation unit 510 (step S0).

  The body motion data acquisition unit 20 of the stress evaluation unit 510 acquires the body motion data M_data from the body motion data measurement unit 11 (step S1), and stores the body motion data M_data in the body motion data storage unit 40 (step S2). ).

  Hereinafter, as an example, a process when the stress evaluation unit 510 calculates the stress index data R_data [n] when measuring the n-th body movement data M_data [n] from the start of measurement will be described. Needless to say, the stress evaluation unit 510 can calculate the stress index data at an arbitrary timing or continuously.

  The trend removal calculation processing unit 30 acquires body motion data M_data [n−X] to M_data [n + X] from the body motion data storage unit 40, performs a trend removal process, and calculates body motion data after the trend removal ( Step S3). The trend removal calculation processing unit 30 stores the calculated trend removal body motion data in the trend removal data storage unit 50 (step S4). In this embodiment, 1800 is used as X. However, the present invention is not limited to this, and X may be an integer.

  The activity state determination calculation processing unit 21 acquires the trend removal body motion data from the trend removal data storage unit 50, and creates the activity state determination data from the trend removal body motion data (step S5).

  The processing of Step S5 to Step S17 is the same as Step S3 to Step S14 in the second embodiment except that the trend-removed body motion data is used instead of the body motion data, and thus description thereof is omitted here. .

  Since the stress evaluation unit 510 uses the trend-removed body motion data as an alternative to the body motion data, the stress evaluation unit 510 can calculate the stress index more accurately.

[Sixth Embodiment]
A stress index evaluation apparatus (stress state estimation apparatus) 500 according to an embodiment (sixth embodiment) of the present invention will be described with reference to FIGS. For the sake of convenience of explanation, the same parts as those of the first to fifth embodiments will not be described, and only different parts will be described.

  FIG. 15 is a block diagram showing a configuration of the stress index evaluation apparatus 600. The stress index evaluation device 600 includes a stress evaluation unit 610 instead of the stress evaluation unit 310 with respect to the stress index evaluation device 300 according to the third embodiment. The stress evaluation unit 610 includes all members of the stress evaluation unit 310, and further includes a trend removal calculation processing unit (trend removal unit) 30 and a trend removal data storage unit 50.

  The process of calculating the stress index by the stress evaluation unit 610 will be described with reference to FIG. FIG. 16 is a flowchart showing the process of the stress evaluation unit 610.

  The body motion data measurement unit 11 starts measuring the body motion data M_data, and transmits the measured body motion data M_data to the stress evaluation unit 610 (step S0).

  The body motion data acquisition unit 20 of the stress evaluation unit 610 acquires the body motion data M_data from the body motion data measurement unit 11 (step S1), and stores the body motion data M_data in the body motion data storage unit 40 (step S2). ).

  Hereinafter, as an example, a process when the stress evaluation unit 610 calculates the stress index data R_data [n] when measuring the n-th body movement data M_data [n] from the start of measurement will be described. Needless to say, the stress evaluation unit 610 can calculate the stress index data at an arbitrary timing or continuously.

  The trend removal calculation processing unit 30 acquires body motion data M_data [n−X] to M_data [n + X] from the body motion data storage unit 40, performs a trend removal process, and calculates body motion data after the trend removal ( Step S3). The trend removal calculation processing unit 30 stores the calculated trend removal body motion data in the trend removal data storage unit 50 (step S4). In this embodiment, 1800 is used as X. However, the present invention is not limited to this, and X may be an integer.

  The statistic calculation processing unit 29 acquires the trend removal body motion data from the trend removal data storage unit 50, and creates the statistic data (intermediate data) from the trend removal body motion data (step S5).

  The processing of step S5 to step S14 is the same as step S3 to step S12 in the third embodiment except that the trend-removed body motion data is used instead of the body motion data, and thus the description thereof is omitted here. .

  Since the stress evaluation unit 610 uses the trend-removed body motion data as an alternative to the body motion data, the stress evaluation unit 610 can calculate the stress index more accurately.

[Seventh Embodiment]
A stress index evaluation apparatus (stress state estimation apparatus) 700 according to an embodiment (seventh embodiment) of the present invention will be described with reference to FIGS. For convenience of explanation, explanation of the same parts as those in the first to sixth embodiments will be omitted, and only different parts will be explained.

  FIG. 17 is a block diagram showing a configuration of the stress index evaluation apparatus 700. The stress index evaluation device 700 includes a stress evaluation unit 710 instead of the stress evaluation unit 610 with respect to the stress index evaluation device 600 according to the sixth embodiment. The stress evaluation unit 710 includes a first type statistic calculation processing unit (calculation unit) 31 and a second type statistic calculation processing unit (calculation unit) 32 instead of the statistic calculation processing unit 29.

  The process of calculating the stress index by the stress evaluation unit 710 will be described with reference to FIG. FIG. 18 is a flowchart showing the processing steps of the stress evaluation unit 710.

  The body motion data measurement unit 11 starts measuring the body motion data M_data, and transmits the measured body motion data M_data to the stress evaluation unit 710 (step S0).

  The body motion data acquisition unit 20 of the stress evaluation unit 710 acquires the body motion data M_data from the body motion data measurement unit 11 (step S1), and stores the body motion data M_data in the body motion data storage unit 40 (step S2). ).

  Hereinafter, as an example, a process when the stress evaluation unit 710 calculates the stress index data R_data [n] when measuring the n-th body movement data M_data [n] from the start of measurement will be described. Needless to say, the stress evaluation unit 710 can calculate the stress index data at an arbitrary timing or continuously.

  The trend removal calculation processing unit 30 acquires body motion data M_data [n−X] to M_data [n + X] from the body motion data storage unit 40, performs a trend removal process, and calculates body motion data after the trend removal ( Step S3). The trend removal calculation processing unit 30 stores the calculated trend removal body motion data in the trend removal data storage unit 50 (step S4). In this embodiment, 1800 is used as X. However, the present invention is not limited to this, and X may be an integer.

  The first-class statistic calculation processing unit 31 acquires body motion data from the body motion data storage unit 40 and creates first statistic data (intermediate data) (step S5). The first-class statistic calculation processing unit 31 stores the calculated first statistic data in the statistic data storage unit 49 (step S6).

  The second type statistic calculation processing unit 32 acquires the trend removal body motion data from the trend removal data storage unit 50 and creates second statistic data (intermediate data) (step S7). The second type statistic calculation processing unit 32 stores the calculated second statistic data in the statistic data storage unit 49 (step S8).

  The statistics (features) calculated by the first-class statistic calculation processing unit 31 and the statistics (features) calculated by the second-class statistic calculation processing unit 32 are respectively average values of body motion data. , One or more of standard deviation (or variance), skewness, and kurtosis. Depending on the type of statistic, by selecting either the first type statistic calculation processing unit 31 or the second type statistic calculation processing unit 32, the nature of the statistic or the work contents of the subject, etc. Therefore, the presence or absence of trend removal can be flexibly set according to the characteristics of movement.

  Since the process of step S9-step S16 is the same as step S7-step S14 in 6th Embodiment, description is abbreviate | omitted here.

  The stress evaluation unit 710 can calculate the stress index more accurately by selecting for each statistic whether to use body movement data or trend removal body movement data in the calculation of the statistic.

(Program and recording medium)
Finally, each block of the stress evaluation unit of the stress index evaluation apparatus may be realized in hardware by a logic circuit formed on an integrated circuit (IC chip) or using a CPU (Central Processing Unit). It may be realized by software.

  In the latter case, the stress evaluation unit includes a CPU that executes instructions of a program that realizes each function, a ROM (Read Only Memory) that stores the program, a RAM (Random Access Memory) that expands the program, the program, and various types A storage device (recording medium) such as a memory for storing data is provided. An object of the present invention is to provide a recording medium in which a program code (execution format program, intermediate code program, source program) of a control program of a stress index evaluation apparatus, which is software that realizes the above-described functions, is recorded in a computer-readable manner This can also be achieved by supplying the stress index evaluation apparatus and reading and executing the program code recorded on the recording medium by the computer (or CPU or MPU).

  Examples of the recording medium include tapes such as magnetic tapes and cassette tapes, magnetic disks such as floppy (registered trademark) disks / hard disks, and disks including optical disks such as CD-ROM / MO / MD / DVD / CD-R. IC cards (including memory cards) / optical cards, semiconductor memories such as mask ROM / EPROM / EEPROM / flash ROM, PLD (Programmable logic device), FPGA (Field Programmable Gate Array), etc. Logic circuits can be used.

  The stress index evaluation apparatus may be configured to be connectable to a communication network, and the program code may be supplied via the communication network. The communication network is not particularly limited as long as it can transmit the program code. For example, the Internet, intranet, extranet, LAN, ISDN, VAN, CATV communication network, virtual private network, telephone line network, mobile communication network, satellite communication network, and the like can be used. The transmission medium constituting the communication network may be any medium that can transmit the program code, and is not limited to a specific configuration or type. For example, even with wired lines such as IEEE 1394, USB, power line carrier, cable TV line, telephone line, ADSL (Asymmetric Digital Subscriber Line) line, infrared rays such as IrDA and remote control, Bluetooth (registered trademark), IEEE 802.11 wireless, HDR ( It can also be used by radio such as High Data Rate (NFC), Near Field Communication (NFC), Digital Living Network Alliance (DLNA), mobile phone network, satellite line, and digital terrestrial network.

  The present invention is not limited to the above-described embodiments, and various modifications are possible within the scope shown in the claims, and embodiments obtained by appropriately combining technical means disclosed in different embodiments. Is also included in the technical scope of the present invention.

  The present invention can be used in the field of manufacturing medical devices, health appliances, and the like.

11 Body movement data measuring unit (measuring means)
12 Awareness stress input section (input means)
13 Stress index output unit 20 Body motion data acquisition unit 21 Activity state determination calculation processing unit (calculation means)
22 Rest time frequency cumulative probability calculation processing unit (calculation means)
23 Stress evaluation value calculation processing unit (calculation means)
24 Awareness Stress Level Data Acquisition Unit 25 Stress Index Calculation Parameter Update Calculation Processing Unit (Update Unit)
26 Stress index calculation processing section (estimating means)
27 Stress Index Output Unit 28 Learning Parameter Update Calculation Processing Unit (Update Unit)
29 Statistics calculation processing unit (calculation means)
30 Trend removal calculation processing unit (trend removal means)
31. First-class statistic calculation processing unit (calculation means)
32. Second type statistics calculation processing unit (calculation means)
40 body motion data storage unit 41 activity state determination data storage unit 42 rest time frequency cumulative probability data storage unit 43 stress evaluation value data storage unit (accumulation means)
44 Awareness stress data storage unit (accumulation means)
45 Stress index calculation parameter storage unit (storage means)
46 Stress index data storage section 48 Learning parameter storage section (storage means)
49 Statistics data storage unit 50 Trend removal data storage unit 100, 200, 300, 400, 500, 600, 700 Stress index evaluation device (stress state estimation device)
110, 210, 310, 410, 510, 610, 710 Stress evaluation unit

Claims (17)

A stress state estimation device that estimates a stress state of a subject based on the body movement of the subject,
Measuring means for measuring the body movement of the subject and obtaining body movement data representing the body movement;
Calculating means for extracting intermediate data by extracting feature values from the body movement data acquired by the measuring means;
Storage means for storing relationship data representing the estimated correspondence between the intermediate data calculated by the calculation means and the stress state of the subject;
Estimating means for estimating the stress state of the subject from the intermediate data calculated by the calculating means using the relational data stored in the storage means;
Input means for receiving input of subjective stress state data representing a stress state perceived by the subject;
When the subjective stress state data is input to the input means, the intermediate data and the subjective stress state data are obtained by using the intermediate data calculated by the calculation means and the set of the subjective stress state data input to the input means. And a renewal means for recursively obtaining data representing the correspondence relationship between the data and storing the obtained data representing the correspondence relationship in the storage means as the relation data. Estimating device. When the stress state estimation device estimates the stress state of the subject at an arbitrary timing, the calculation means calculates intermediate data from the body movement data acquired by the measurement means within a predetermined time including the arbitrary timing. And
The update means includes a set of intermediate data calculated by the calculation means from the body motion data acquired by the measurement means within a predetermined time including a timing at which the subjective stress state data is input, and the subjective stress state data. The stress state estimation apparatus according to claim 1, wherein the correspondence relation between the intermediate data and the subjective stress state data is obtained recursively using. A storage means for storing and storing a set of the intermediate data and the subjective stress state data when the subjective stress state data is input to the input means;
The updating unit performs a regression analysis on the set of the intermediate data and the subjective stress state data stored in the storage unit, using the subjective stress state data as an objective variable and the intermediate data as an explanatory variable. The stress state estimation apparatus according to claim 1, wherein data representing a correspondence relationship between the intermediate data and the subjective stress state data is obtained.   4. The stress state estimation apparatus according to claim 3, wherein the storage unit discards the set of the intermediate data and the conscious stress state data stored over a predetermined period. The updating means includes a machine learning device that performs supervised machine learning,
The relationship data and the data representing the correspondence are learning parameters of the machine learner,
The machine learning device uses the supervised stress state data as teacher data and performs supervised learning using the intermediate data as input data, thereby recursively expressing data representing a correspondence relationship between the intermediate data and the perceived stress state data. The stress state estimation device according to claim 1, wherein the stress state estimation device is obtained.   The stress state estimation apparatus according to claim 5, wherein the machine learner is a multi-layer perceptron.   The stress state estimation apparatus according to claim 1, wherein the feature amount is a length and frequency of rest of the subject.   The calculation means determines whether the body motion data indicates activity or rest of the subject by comparing a predetermined threshold value with the body motion data, and rests the subject based on the determination result. The stress state estimation apparatus according to claim 7, wherein a length and a frequency of the stress are extracted.   The stress state estimation apparatus according to claim 7 or 8, wherein the intermediate data is stress evaluation value data calculated from the length and frequency of rest of the subject.   The calculation means uses a logarithm of a resting time frequency cumulative probability that is a probability of the subject to take a rest longer than a certain length as an objective variable, and uses a linear regression line with the logarithm of the certain length as an explanatory variable. The stress state estimation apparatus according to claim 9, wherein the stress evaluation value data is calculated based on an inclination of the obtained regression line.   The stress state estimation apparatus according to any one of claims 1 to 6, wherein the feature amount is a statistical amount of two or more types of the body motion data.   12. The stress state estimation apparatus according to claim 11, wherein the statistic is selected from the group consisting of an average value, standard deviation, skewness, kurtosis and variance. A trend removing means for removing the trend from the body movement data;
The calculation means extracts one or more types of statistics from the body movement data, and extracts one or more types of statistics from the body movement data from which the trend has been removed by the trend removal means. The stress state estimation apparatus according to claim 11 or 12. A trend removing means for removing the trend from the body movement data;
The stress state estimation apparatus according to any one of claims 1 to 12, wherein the calculation means calculates the intermediate data from the body movement data from which the trend has been removed by the trend removal means. A stress state estimation device comprising a storage means is a stress state estimation method for estimating a stress state of a subject based on body movement of the subject,
A measurement step of measuring body movement of the subject and obtaining body movement data representing the body movement;
A calculation step of extracting feature values from the body movement data acquired in the measurement step and calculating intermediate data;
From the intermediate data calculated in the calculating step, the stress state of the subject is estimated using relationship data representing an estimated correspondence relationship between the intermediate data stored in the storage means and the stress state of the subject. An estimation process to
An input step of receiving input of subjective stress state data representing a stress state perceived by the subject;
When the subjective stress state data is input, the intermediate data and the set of the subjective stress state data are used to recursively obtain data representing the correspondence between the intermediate data and the subjective stress state data. An update step of storing data representing the obtained correspondence relationship in the storage means as the relationship data.   A program for causing a computer to operate as the stress state estimation apparatus according to any one of claims 1 to 14, wherein the program causes the computer to function as each unit included in the stress state estimation apparatus.   The computer-readable recording medium which recorded the program of Claim 16.

Images