JP2006068458A - Device and method for estimating physiological state - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a device and a method for estimating a physiological state by which the physiological state of a subject to be examined can be estimated precisely. <P>SOLUTION: In a vigilance estimation processing section 21, the physiological state (vigilance) of the subject to be examined is computed based on the outputs of a plurality of physiological index measuring sections 20 (instruments for measuring brain waves, heartbeat, electric impedance of the skin, and the like) which is selected from a first physiological index selecting section 13 based on the subject's work content or a second physiological index selecting section 17 based on his/her action characteristics. In this computation, a physiological index is substituted for a vigilance estimation model generated by a vigilance estimation model construction section 14, and the vigilance is calculated. The vigilance estimation model is selected according to the combination of the physiological index measuring sections 20 which are selected in the first physiological index selecting section 13 or the second physiological index selecting section 17. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to a physiological state estimation apparatus and method.

A conventional physiological state estimation device is described in Patent Document 1, for example. In this physiological state estimation device, a driver's drowsiness is detected and a warning is issued by analyzing the awakening level of the human body from the behavior of the pulse wave. That is, the pulse wave signal of the driver's arm is stored in a memory, and the peak is obtained to obtain the time interval of the R wave of the heartbeat, and the awakening level is further obtained from this time interval. The pulse wave does not need to be detected in the arm, and can be detected at another location.
JP-A-8-299443

  However, when the arm is operated, an accurate pulse wave signal cannot be obtained, and the same situation can be considered when detection is performed at another location. That is, the validity of the data changes depending on the work contents of the subject, and the awakening level of the subject cannot be accurately measured.

  An object of the present invention is to provide a physiological state estimation device and method capable of accurately estimating the physiological state of a subject.

  In order to solve the above-described problem, a physiological state estimation device according to a first aspect of the present invention includes a motion detection unit that detects a subject motion and a plurality of physiology that respectively detect a plurality of subject physiological indices in the subject's physiological state estimation device. Select the index detection means, the work estimation means for estimating the subject work from the subject motion detected by the motion detection means, and the physiological index detection means that can be used when the subject performs the subject work estimated by the work estimation means. And a physiological state calculation means for calculating the subject's physiological state based on the output of the physiological index detection means selected by the first physiological index selection means.

  In this device, the subject work (driving, etc.) is estimated from the action of the subject, but the first physiological index selection means selects a physiological index detection means that can give an effective output when performing the work, The subject's physiological state is calculated based on the output of the selected physiological index detection means. In this case, since the physiological index detection means that outputs an ineffective output is not selected depending on the type of the subject work, the physiological state of the subject can be accurately estimated by calculating the physiological state based on the physiological index.

  In the physiological condition estimation device according to the second invention, the physiological condition is a degree of arousal of the subject. That is, there are various physiological states, but here it is assumed that the degree of arousal of the subject. In this case, the driver can be informed of the arousal level when driving the vehicle.

  The physiological state estimation device according to the third invention further comprises wakefulness level estimation model selection means for selecting a wakefulness level estimation model according to the combination of the physiological index detection means selected by the first physiological index selection means, and the physiological state The calculation means calculates the subject arousal level as the subject physiological state by inputting the output of the selected physiological index detection means to the arousal level estimation model selected by the arousal level estimation model selection means. To do.

  The arousal level is obtained by inputting a plurality of physiological indices into a prescribed arousal level estimation model, but the optimal arousal level estimation model differs depending on the combination of physiological indices. In this invention, since the arousal level estimation model is selected according to the combination of the selected physiological index detection means, an accurate arousal level can be obtained.

  In the physiological state estimation device according to the fourth invention, the work estimation means includes a work content database that stores a relationship between the subject work and the subject action, and whether the subject work that provides the subject action detected by the action detection means is correct. If there is a subject work in the work content database that calculates the accuracy and has an accuracy equal to or higher than a predetermined accuracy, a work content determination means that adopts the subject work and obtains a work estimation result is provided.

  The work content database stores the relationship between the subject work and the subject action. Since the subject motion is detected by the motion detection means, there may be a subject task corresponding to this in the work content database, but the relationship between the motion and the task is generally not unique. Therefore, the accuracy of the work content originally performed based on the subject's movement in the possible situation is calculated, and if the accuracy is equal to or greater than the predetermined accuracy, the estimated work content is assumed to be the actual work content, and the work estimation is performed. Adopt as a result. Thereby, the final physiological state of the subject can be obtained more accurately.

  The physiological state estimation device according to a fifth aspect of the present invention further comprises second physiological index selection means for selecting a physiological index detection means that can be used for output from the test subject motion detected by the motion detection means. When the content determination means determines that there is no subject work having an accuracy equal to or greater than the predetermined accuracy in the work content database, the subject physiological state is determined based on the output of the physiological index selection means selected by the second physiological index selection means. It is characterized by calculating.

  When the work content for the motion cannot be specified with high accuracy, a usable physiological index is specified not from the work content but from the motion itself. That is, when it is determined that there is no subject work having an accuracy equal to or higher than the predetermined accuracy in the work content database, based on the output of the physiological index selection means selected by the second physiological index selection means, that is, the subject action The subject's physiological state is calculated.

  The identification of the physiological index based on the operation is a backup process of the identification process of the physiological index based on the work content. In other words, since the subject's movement is instantaneous, the physiological index to be selected changes when the movement changes, but if the physiological index is specified from the work content, as long as the same work content is performed, Judgment can hold validity.

  In the physiological condition estimation apparatus according to the sixth aspect of the invention, the first physiological index selection means selects a physiological index detection means that can be used for output from a database that indicates the relationship between physiological indices that can be used for output and subject work. .

  Since there is a database associating the test subject work with the physiological index, the first physiological index selection means can select the physiological index detection means.

  In the physiological condition estimation apparatus according to the seventh invention, the second physiological index selection means selects physiological index detection means that can be used for output from a database that indicates the relationship between physiological indices that can be used for output and subject movements.

  Since there is a database associating the test subject's action with the physiological index, the second physiological index selection means can select the physiological index detection means.

  The physiological state estimation method according to the present invention is detected in the subject's physiological state estimation method in the motion detection step of detecting the subject motion, the physiological index detection step of detecting a plurality of subject physiological indexes, and the motion detection step. A task estimation step for estimating a subject task from a subject motion, a first physiological index selection step for selecting a physiological index that can be used for output when the subject performs the subject task estimated in the task estimation step, and a first physiological index selection And a physiological state calculating step of calculating a subject's physiological state based on the physiological index selected in the step.

  In this method, the subject's work (driving, etc.) is estimated from the motion of the subject. In the first physiological index selection step, a physiological index that can provide an effective output when performing the work is selected and selected. The subject's physiological state is calculated based on the physiological index. In this case, since the physiological index detection means that outputs an ineffective output is not selected depending on the type of the subject work, the physiological state of the subject can be accurately estimated by calculating the physiological state based on the physiological index.

  According to the present invention, the physiological state of a subject can be accurately estimated.

  Hereinafter, a physiological state estimation device and method according to an embodiment will be described. In addition, the same code | symbol shall be used for the same element and the overlapping description is abbreviate | omitted.

  FIG. 1 is a block diagram of a physiological state estimation apparatus.

  The physiological state estimation apparatus includes a behavior feature detection unit (motion detection means) 10 that directly or indirectly detects a subject's motion. The behavior feature detection unit 10 detects the motion of the subject directly with, for example, a light or an ultrasonic sensor, or detects the steering motion of a device operated by the subject, for example, a vehicle, with a steering angle sensor or the like, Or indirectly detecting the movement of the subject.

  The physiological state estimation apparatus includes a plurality of physiological index measurement units (physiological index detection means) 20 that respectively detect a plurality of subject physiological indices (such as brain waves, heartbeats, and skin impedance) (in the figure, one representative is representative). Only blocks are shown). The physiological index measured by the physiological index measuring unit 20 is input to the arousal level estimation processing unit (physiological state calculation means) 21. The arousal level estimation processing unit 21 calculates the arousal level based on the input physiological index, The calculated awakening level is input to the output unit 22. The output unit 22 is, for example, a monitor or a dozing alarm sensor.

  Even though the calculation of the arousal level is not necessarily accurate, the physiological index used for the calculation is not accurate, and in this embodiment, the physiological index and the arousal level model are selected according to the work contents of the subject.

  The work content determination unit 11 estimates the subject work (during driving, reading, etc.) from the subject movement (leg or head movement) detected by the behavior feature detection unit 10. For this estimation, a matching method using the work content database 15 in which the subject motion and the subject work content are associated is used. That is, the work estimation means includes a work content determination unit 11 and a work content database 15 that stores the relationship between the subject work and the subject motion. The work content determination unit 11 is detected by the behavior feature detection unit 10. The accuracy of whether or not the subject work that gives the subject action is correct is calculated, and if there is a subject work in the work content database 15 having an accuracy equal to or higher than a predetermined accuracy, this subject work is adopted as the work estimation result.

  The work content database 15 stores the relationship between the subject work and the subject action. Since the subject motion is detected in the behavior feature detection unit 10, there can be subject work corresponding to this in the work content database 15. Since the relationship between motion and work is not unique, the probability that the estimated work content is the actual work content based on the subject's motion in a possible situation is calculated, and this accuracy is predetermined. If the accuracy is higher than the accuracy, the estimated work content is assumed to be the actual work content and is adopted as the work estimation result. Thereby, the final physiological state of the subject can be obtained more accurately.

  The determination processing branching unit 12 determines whether the above estimation is possible or impossible. When this estimation is possible, the first physiological index selection unit (first physiological index selection means) 13 is a physiological index whose output can be used when the subject performs the subject work estimated by the work content determination unit 11. A measuring unit (an electroencephalogram measuring instrument, a heart rate measuring instrument, a skin impedance measuring instrument, etc.) 20 is selected. For this selection, a database 16 in which work contents are associated with physiological indices is used. As described above, since there is the database 16 that associates the test subject work with the physiological index, the first physiological index selection unit 13 can select the physiological index measurement unit 20.

  The arousal level estimation processing unit 21 calculates the physiological state of the subject based on the output of the physiological index measurement unit (electroencephalogram measurement device, heart rate measurement device, skin impedance measurement device, etc.) 20 selected by the first physiological index selection unit 13. To do. In this calculation, the physiological index is substituted into the arousal level estimation model generated by the arousal level estimation model construction unit 14 to obtain the arousal level. The arousal level estimation model construction unit 14 selects the arousal level estimation model according to the combination of the physiological index measurement unit 20 selected by the first physiological index selection unit 13 or the second physiological index selection unit 17. The arousal level estimation processing unit 21 adds the output of the selected physiological index measurement unit (an electroencephalogram measurement device, a heart rate measurement device, a skin impedance measurement device, etc.) 20 to the arousal level estimation model selected by the arousal level estimation model construction unit 14. Is input, the subject awakening degree as the subject physiological state is calculated.

  In this example, the physiological state is the degree of arousal of the subject, but various physiological states are conceivable. When the physiological state is the arousal level, it is possible to design a device configuration that informs the driver of the awakening level when driving the vehicle. As described above, in the present apparatus, the subject work (driving, etc.) is estimated from the motion of the subject, but the first physiological index selection unit 13 can provide an effective output when performing the work. 20 is selected, and the subject's physiological state (wakefulness) is calculated based on the output of the selected physiological index measurement unit 20. In this case, the physiological index measuring unit 20 that outputs an invalid output depending on the type of the subject work is not selected. Therefore, if the physiological state is calculated based on the selected physiological index, the physiological state of the subject can be accurately estimated. it can.

  When the determination processing branching unit 12 determines that it is impossible to estimate the work content, a wakefulness estimation model is constructed from the behavior characteristics without using the work content. In other words, when the work content for the operation cannot be specified with high accuracy, the usable physiological index is specified not from the work content but from the operation itself. That is, when it is determined that there is no subject work having an accuracy equal to or higher than the predetermined accuracy in the work content database 15, the output of the physiological index selection database 18 selected by the second physiological index selection unit 17, that is, the subject action Based on the above, the physiological state of the subject is calculated.

  In other words, the second physiological index selection unit 17 selects the physiological index measurement unit 20 that can use the output from the database 18. The database 18 stores the relationship between available physiological indices and subject movements. The second physiological index selection unit 17 selects a physiological index measurement unit (an electroencephalogram measurement device, a heart rate measurement device, a skin impedance measurement device, etc.) 20 that can be used based on the test subject motion detected by the motion measurement unit 20. The awakening level estimation processing unit 21 determines that the work content determination unit 11 does not have a subject work in the work content database 15 having a certain accuracy (a certainty that the actual work content is the estimated work content) or more. In this case, the subject's physiological state is calculated based on the output of the physiological index selection unit 17 selected by the second physiological index selection unit 17.

  Note that the specification of the physiological index based on the above-described operation is a backup process of the specific process of specifying the physiological index based on the work content. In other words, since the subject's movement is instantaneous, the selected physiological index changes when the movement changes, but if the physiological index is specified from the work content, the judgment will be made as long as the work content is being performed. Can retain its effectiveness.

  The arousal level is obtained by inputting a plurality of physiological indices into a prescribed arousal level estimation model, but the optimal arousal level estimation model differs depending on the combination of physiological indices. In this invention, since the arousal level estimation model is selected according to the combination of the selected physiological index measurement unit 20, an accurate arousal level can be obtained.

  The physiological state estimation method according to the present invention includes a motion detection step (behavior feature detection unit 10) for detecting a subject motion and a physiological index detection step (physiology) for detecting a plurality of subject physiological indexes in the subject physiological state estimation method. When the subject performs the index measurement unit 20), the work estimation step (work content determination unit 11) for estimating the subject work from the subject motion detected in the motion detection step, and the subject work estimated in the work estimation step ( When the determination processing branching unit 12 is “possible”, a first physiological index selection step (first physiological index selection unit 13) for selecting a physiological index whose output can be used, and a first physiological index selection step (first physiological index) A physiological state calculation step (wakefulness estimation processing unit 21 and arousal level estimation model construction unit 14) that calculates a subject's physiological state based on the physiological index selected by the selection unit 13). And

  In this method, the subject's work (driving, etc.) is estimated from the motion of the subject. In the first physiological index selection step, a physiological index that can provide an effective output when performing the work is selected and selected. The subject's physiological state is calculated based on the physiological index. In this case, since the physiological index detection means that outputs an ineffective output is not selected depending on the type of the subject work, the physiological state of the subject can be accurately estimated by calculating the physiological state based on the physiological index.

  Hereinafter, each component will be described in detail.

  FIG. 2 is a block diagram of the behavior feature detection unit 10 and its peripheral elements.

  The behavior feature detection unit 10 includes a behavior detection unit (direct motion detection) S1 and a device motion detection unit (indirect motion detection) S2 operated by the subject. The behavior detection unit S1 includes a line-of-sight measurement sensor that measures the line of sight of the subject, a sensor that detects movement of the entire body, movement of the limbs, and the like. The device motion detection unit S2 includes a sensor that detects gripping of a pen, a steering angle sensor that detects steering operation of the vehicle, and the like.

  FIG. 3 is a table in the action content database 15.

  In the action content database 15, the operation (condition) of the subject and the work content are stored in association with each other.

  Conditions include: movement of the whole body (stationary, large movement) detected by the action detection unit S1, movement of the limbs (stationary, movement of arms, movement of arms and legs), line of sight (almost fixed, movement in a narrow range, In addition to a wide range of movements, the surrounding environment (building interior, vehicle interior, indoor, outdoor), equipment used (books, pens, notebooks, vehicles, various types (unspecified), specific equipment), etc. The value detected by the motion detection unit S2 is set.

  As work contents, reading, taking notes, calculation, driving, housework, sports, etc. are set.

  FIG. 4 shows a matching process in the work content determination unit 11 using the table of the action content database 15.

  When the condition that the whole body moves, the arms and legs move, the line of sight moves in a narrow range, the surrounding environment is an indoor environment, and the device used is a vehicle is entered, a circle is added in the table. In this case, it is possible to determine that the work content that most closely matches the condition is that the number of corresponding circles is “driving” most. Whether or not the actual work content is the estimated work content “operation” is given by “Accuracy = number of circles (number of conditions) / total number of conditions in the work content database”.

  When the accuracy exceeds a predetermined value TH, for example, exceeds 60%, it is determined that “estimated work content = actual work content”, and the determination processing branching unit 12 determines the work content. Can be determined to be “possible”.

  When this accuracy is lower than a predetermined value TH, for example, 60% or less, it is determined that “estimated work content = not actual work content”, and the determination processing branching unit 12 It can be determined that the content determination is “impossible”.

  The physiological index measuring unit 20 is an electroencephalogram measuring device, a heart rate measuring device, a skin impedance measuring device, a line-of-sight sensor, a respiration sensor, and a body temperature sensor, and the electroencephalogram, heart rate, skin, skin impedance, eye movement amount per unit time, The number of blinks, the number of breaths per unit time, and body temperature are output. The physiological state (wakefulness level) of the subject is estimated using these physiological indices.

  First, when the work content determination is “possible”, the first physiological index selection unit 13 attempts to associate the work content with an available physiological index and selects an available physiological index.

  FIG. 5 is a table of the database 16 showing correspondence between work contents and physiological indices. Available physiological indices are indicated by circles, and unusable physiological indices are indicated by crosses. The physiological index measurement unit 20 outputs brain waves, blinks, heart beats from the arms, heart beats from the chest, skin impedance, etc., and each work content is reading, making a mistake, taking notes, calculating, driving, housework, sports Etc. are shown.

  FIG. 6 is a table of the database 16 showing the correspondence between work contents and physiological indices, and the selection of available physiological indices using this table is shown. When it is determined that the work content determination is “possible”, an available physiological index can be uniquely determined from the database 16. The figure shows a case where `` driving '' is selected as the work content.In this case, brain waves, blinks, heartbeats detected from the chest, skin impedance can be used for determination of arousal level, It turns out that the heartbeat detected from the arm is not available. What is necessary is just to obtain | require an arousal level using these physiological indexes.

  Next, when the work content determination is “impossible”, the second physiological index selection unit 17 tries to associate the behavior characteristic (motion) of the subject with the available physiological index.

  FIG. 7 is a table of the database 18 showing the correspondence between the behavior feature and the physiological index. Available physiological indices are indicated by circles, and unusable physiological indices are indicated by crosses. The physiological indices measured by the physiological index measuring unit 20 are as described above. Since the work content has not been specified, whether the behavioral features include a lot or little gaze movement (gaze sensor output), a lot of movement of the whole body (light or ultrasonic sensor output), or a lot of limb movement Less (light or ultrasonic sensor output, steering wheel angle sensor output (for vehicles), pedal motion measurement sensor output (for vehicles)), head movement (light or ultrasonic sensor) Is set about. A physiological index corresponding to a circle is selected.

  FIG. 8 is a table showing the relationship between physiological index combinations and arousal level models A, B, C, D... A circle is shown when the physiological index is available, and a cross is shown when it is unavailable.

  FIG. 9 is a table showing the relationship between the combination of physiological indices and the arousal level models A, B, C, D... In the arousal level model construction unit 14 and shows how the arousal level model is selected. The availability of these physiological indices is obtained from the work contents when the work contents can be specified, and from the behavior characteristics when the work contents cannot be specified.

  For example, when brain waves, blinks, heartbeats detectable from the chest, and skin impedance are selected as physiological indices for calculating arousal level, the arousal level model B is selected.

  Development of technology to prevent the driver's arousal level from falling is one of the important research issues from the viewpoint of safety, and researches on methods for detecting a low level of arousal level and warning technology are actively conducted. . Many models are known as arousal level models, and the content itself is publicly known. By repeating the test with the physiological index of the subject, the optimal arousal level model can be associated with the physiological index group. For example, wakefulness model A is wakefulness V = α × (electroencephalogram) + β × (heartbeat) + γ × (skin impedance), and wakefulness model B is wakefulness V = α × (electroencephalogram) + β × (beat) + γ × (Skin impedance) + δ × (number of blinks).

  FIG. 10 is a table showing an example of the subject name, the selected arousal level model, and constants assigned to the arousal level model.

  When the subject HH adopts the arousal level model A, α = 0.018, β = 1.5, γ = 0.21. When the subject HH adopts the arousal level model B, α = 0.58, β = 1.2, γ = 8.3, and δ = 12. When another subject KH adopts the arousal level model A, α = 0.433, β = 5.1, γ = 0.01, and when the awakening level model B is employed, α = 1.83. , Β = 3.5, γ = 0.3, and δ = 7.1 are employed. That is, the constants differ for each subject (HH, KH, YH...) And arousal level model (A, B, C, D...). Note that the coefficients α, β, γ, and δ are coefficients that are conceptually defined and change depending on how each data is taken. For example, the arousal level of the arousal level model B Each coefficient can also be set as V = α × (electroencephalogram) + β × (number of blinks) + γ × (heart rate) + δ × (skin impedance).

  FIG. 11 is a conceptual diagram when the arousal level V is obtained when an electroencephalogram, a heartbeat, and skin impedance are employed as physiological indices.

  The electroencephalogram (a) output from the electroencephalograph, the heartbeat (electrocardiogram) (b) output from the heartbeat measuring instrument, and the skin impedance (c) output from the skin impedance measuring instrument are sampled within a predetermined measurement period T. When these physiological indices are selected, each is subjected to signal processing and substituted into the selected expression of the arousal level V.

The brain wave is subjected to FFT (Fast Fourier Transform) by signal processing, and a signal in a predetermined frequency band is extracted. For example, the total signal intensity integrated value in this frequency band is input to the expression of the arousal level V as brain wave data. The electrocardiogram data is converted into a heart rate or an R wave time interval, and the converted data from the current time t ₀ to m seconds (≈T) before is input to the expression of the arousal level V as heart rate data. For skin impedance, for example, an average value is calculated by signal processing, and this is input to the expression of the arousal level V.

  FIG. 12 is a conceptual diagram illustrating an example of an arousal level estimation device mounted on a vehicle.

  The behavior feature detection unit 10 illustrated in FIG. 1 is a body motion monitor (light or ultrasonic sensor) that performs direct motion measurement, and includes a pedal motion measurement sensor and a steering angle sensor that perform indirect motion measurement. Examples of the physiological index measurement unit 20 include a line-of-sight monitor, a heart rate measurement sensor for a chest, a skin impedance measurement sensor provided on a handle, and an electroencephalogram measurement sensor.

  The computer 100 receives the movement of the subject and the physiological index of the subject. Inside the computer 100, the output from the behavior feature detection unit 10 refers to the work content database 15 to determine whether the work content of the subject is “driving”, and is determined to be “driving”. If possible, select physiological indices (electroencephalogram, heart rate, blink rate, skin impedance) and wakefulness model corresponding to the work content, and read out coefficients α, β, γ, δ corresponding to the subject's personal data These are substituted into the arousal level model, and the arousal level V is calculated.

  If it is not clear whether or not the subject's work is “driving”, a physiological index and a wakefulness model that can be used for output are selected from the behavior characteristics, and coefficients α, β, γ, and δ corresponding to the subject's personal data are selected. Are substituted into the arousal level model, and the arousal level V is calculated. In the case of this example, since each measuring instrument is mounted on the vehicle, it is clear that the work content is specified, but the program of the computer 100 is general-purpose and applied to the home automation, medical care, and security fields. In some cases, it is possible to adopt the core technology of the present invention and make a design change to the database part that specifies the behavioral characteristics and work contents.

  The present invention relates to a physiological state estimation apparatus and method.

It is a block diagram of the physiological condition estimation apparatus. It is a block diagram of the action characteristic detection part 10 and its peripheral element. It is a table in the action content database 15. It is a table | surface of the action content database 15. FIG. It is a table | surface of the database 16 which shows a response | compatibility with work content and a physiological index. It is a table | surface of the database 16 which shows a response | compatibility with work content and a physiological index. It is a table of the database 18 which shows a response | compatibility with a behavior characteristic and a physiological index. It is a table | surface which shows the relationship between the combination of the physiological index in the arousal level model construction part 14, and the arousal level model A, B, C, D .... It is a table | surface which shows the relationship between the combination of the physiological index in the arousal level model construction part 14, and the arousal level model A, B, C, D .... It is a table | surface which shows an example of the constant substituted by a test subject name, the selected arousal level model, and the arousal level model. It is a conceptual diagram at the time of calculating | requiring the arousal level V, when an electroencephalogram, a heart rate, and skin impedance are employ | adopted as a physiological index. It is a conceptual diagram which shows an example of the arousal level estimation apparatus mounted in the vehicle.

Explanation of symbols

DESCRIPTION OF SYMBOLS 21 ... Arousal level estimation process part, 13 ... 1st physiological index selection part 13, 17 ... 2nd physiological index selection part, 20 ... Physiological index measurement part, 14 ... Arousal level estimation model Construction department.

Claims (8)

In the physiological state estimation device,
Motion detection means for detecting subject motion;
A plurality of physiological index detection means for detecting a plurality of subject physiological indices,
Work estimation means for estimating the subject work from the subject motion detected by the motion detection means;
When the subject performs the subject work estimated by the work estimation means, a first physiological index selection means that selects a physiological index detection means that can be used for output;
Physiological state calculation means for calculating a subject's physiological state based on the output of the physiological index detection means selected by the first physiological index selection means;
A physiological state estimating device comprising:   The physiological state estimation device according to claim 1, wherein the physiological state is a degree of arousal of a subject. A wakefulness estimation model selection means for selecting a wakefulness estimation model according to the combination of the physiological index detection means selected by the first physiological index selection means;
The physiological state calculation means calculates a subject arousal level as a subject physiological state by inputting the output of the selected physiological index detection unit to the arousal level estimation model selected by the arousal level estimation model selection unit. The physiological state estimation apparatus according to claim 2, wherein: The work estimation means includes
A work content database for storing the relationship between the subject work and the subject action;
Calculate the accuracy of whether or not the subject work that gives the subject motion detected by the motion detection means is correct, and if there is a subject work in the work content database that has a certain accuracy or more, adopt this subject work. Work content determination means as work estimation results;
The physiological state estimation apparatus according to any one of claims 1 to 3, further comprising: A second physiological index selection means for selecting a physiological index detection means that can be used for output from the test subject movement detected by the movement detection means;
The physiological state calculation means includes
When the work content determination means determines that there is no subject work having an accuracy equal to or greater than a predetermined accuracy in the work content database, based on the output of the physiological index selection means selected by the second physiological index selection means The physiological condition estimation apparatus according to claim 4, wherein the physiological condition of the subject is calculated.   6. The first physiological index selection means selects a physiological index detection means that can be used for output from a database that indicates a relationship between physiological indices that can be used for output and subject work. Arrangement state estimation device of description.   6. The organizing state estimation apparatus according to claim 5, wherein the second physiological index selection unit selects a physiological index detection unit that can be used for output from a database that indicates a relationship between an available physiological index and a subject's movement. In the method for estimating the physiological state of the subject,
A motion detection step of detecting subject motion;
A physiological index detection step of detecting a plurality of subject physiological indices,
A work estimation step for estimating the subject work from the subject motion detected in the motion detection step;
When the subject performs the subject work estimated in the work estimation step, a first physiological index selection step for selecting a physiological index that can be used for output;
A physiological state calculation step of calculating a subject's physiological state based on the physiological index selected in the first physiological index selection step;
A physiological state estimation method comprising:

Images