JP7183901B2 - SLEEP STATE DETERMINATION DEVICE, SLEEP STATE DETERMINATION METHOD AND PROGRAM - Google Patents

Description

The present invention relates to a sleep state determination device, a sleep state determination method, and a program for determining a sleep state of a user.

Techniques for determining a sleep state based on a person's biological information are known. For example, in Patent Document 1, biometric information such as a person's heart rate or pulse rate is measured, a trend curve representing the tendency of change over time in the biometric information is calculated, and based on this trend curve A sleep state determination device for determining sleep states is described.

Japanese Patent Application Laid-Open No. 2001-61820

By the way, it is known that the sleep state is related to the movement of the human body (body movement). Therefore, if the sleep state is determined using not only the biological information described above but also the information related to body movement, it is expected that the determination accuracy will be improved. However, in this case, in addition to the sensor for measuring biological information, it is necessary to newly configure a device such as a motion sensor for measuring body movement.

SUMMARY OF THE INVENTION It is an object of the present invention to provide a sleep state determination device, a sleep state determination method, and a program that can determine a sleep state using a user's heartbeat interval .

In order to achieve the above object, the sleep state determination device according to the present invention includes:
measuring means for measuring the user's biometric information including the heartbeat interval ;
Acquisition means for acquiring a body motion feature amount , which is an index indicating body motion of the user, based on the biometric information not measured from a sleeping person, which is included in the biometric information measured by the measurement means;
determination means for determining a sleep state of the user based on the body motion feature amount acquired by the acquisition means;
characterized by comprising

According to the present invention, it is possible to determine the sleep state using the heartbeat interval of the user .

BRIEF DESCRIPTION OF THE DRAWINGS It is a figure which shows the whole sleep state determination system structure which concerns on embodiment of this invention. It is a block diagram which shows the hardware constitutions of the sleep state determination apparatus which concerns on embodiment of this invention. It is a block diagram showing a functional configuration of a sleep state determination device according to an embodiment of the present invention. (A) and (B) are diagrams showing an example of extracting outliers from biometric information. It is a flowchart which shows the flow of the learning process performed by the sleep state determination apparatus which concerns on embodiment of this invention. It is a flowchart which shows the flow of the sleep state determination process performed by the sleep state determination apparatus which concerns on embodiment of this invention. It is a figure for demonstrating the other example which acquires a sleep feature-value.

BEST MODE FOR CARRYING OUT THE INVENTION Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. The same reference numerals are given to the same or corresponding parts in the drawings.

A sleep state determination device 100 according to an embodiment of the present invention is a device that determines the sleep state of a user who is a subject. Specifically, the sleep state determination device 100 according to the present embodiment can determine any of wakefulness, REM sleep, and non-REM sleep (stages 1 to 3) as the sleep state of the user. . As shown in FIG. 1, the sleep state determination device 100 is communicably connected to the pulse wave sensor 200 via a communication line 300 conforming to standards such as USB. A sleep state determination system 1 is configured by the sleep state determination device 100 and the pulse wave sensor 200 .

The pulse wave sensor 200 is worn on the user's finger or the like, measures a waveform (volume pulse wave) indicating changes in the volume of the blood vessel at the attachment site in real time, and transmits the data indicating the volume pulse wave via the communication line 300 to sleep. It is transmitted to the state determination device 100 at any time. Specifically, the pulse wave sensor 200 includes a light emitting element such as an LED (Light Emitting Diode) and a light receiving element such as a photodiode. The volume pulse wave is measured by converting the reflected light into an electrical signal with a light receiving element.

Next, the hardware configuration of the sleep state determination device 100 will be described. As shown in FIG. 2, the sleep state determination device 100 includes a CPU (Central Processing Unit) 101, a ROM (Read Only Memory) 102, a RAM (Random Access Memory) 103, a display device 104, and an input device 105. , a storage device 106 and an I/F unit 107 . These units are electrically connected to each other via a bus line BL.

The CPU 101 reads programs and data stored in the ROM 102 and storage device 106 onto the RAM 103 and executes processing, thereby controlling the sleep state determination device 100 as a whole.

The ROM 102 is a non-volatile memory that stores programs to be executed by the CPU 101 and data used in executing the programs. The ROM 102 stores, for example, control programs used for learning processing and sleep state determination processing, which will be described later.

A RAM 103 is a volatile memory that temporarily holds various programs and various data read from the ROM 102 and the storage device 106 and is used as a work area for the CPU 101 .

The display device 104 is a display device such as an LCD (Liquid Crystal Display), a PDP (Plasma Display Panel), an organic EL (Electro-Luminescence) display device, etc., and displays various images under the control of the CPU 101 .

The input device 105 is a mouse, a keyboard, or the like, receives an operation from a user, and outputs a signal corresponding to the received operation to the CPU.

The storage device 106 is a rewritable large-capacity, non-volatile storage device, such as an HDD (Hard Disk Drive). The storage device 106 stores various programs executed by the CPU 101 and various data. For example, storage device 106 stores biometric information DB 1061 and teacher data DB 1062 .

The biometric information DB 1061 is a database in which user's biometric information is stored. In this embodiment, the biometric information is information indicating the heartbeat interval of the user. In the biological information DB 1061, biological information indicating the heartbeat interval of the user is stored in chronological order in association with information indicating the time when the heartbeat interval appeared.

The teacher data DB 1062 stores teacher data for creating (learning) a machine learning model for determining a sleep state from biological information. The teacher data is data in which information indicating a person's sleep state in the predetermined period is added to biometric information (heartbeat interval) for a predetermined period (for example, 30 seconds). For example, a person's heartbeat interval and sleep state at that time actually measured in a predetermined period may be used as teacher data, or empirically known typical person's heartbeat interval and sleep state Based on this, teacher data may be created. In order to improve the accuracy of learning, it is desirable that a large number of different teacher data are stored in the teacher data DB 1062 .

I/F section 107 is an interface for receiving data from pulse wave sensor 200 via communication line 300 .

Next, the functional configuration of the sleep state determination device 100 will be described with reference to FIG. 3 . The sleep state determination device 100 has a functional configuration including a biological information measurement unit 110, an outlier extraction unit 120, a feature amount acquisition unit 130, a learning unit 140, a sleep state determination unit 150, and a determination result output unit. 160. These units are realized by cooperative operation of the hardware shown in FIG.

The biological information measurement unit 110 measures the heartbeat interval, which is the user's biological information, from the volume pulse wave measured by the pulse wave sensor 200 . Specifically, the biological information measurement unit 110 may measure, as the heartbeat interval, the interval between the peaks of the waveform of the volume pulse wave filtered by a bandpass filter or the like. Then, the biological information measurement unit 110 stores the biological information indicating the measured heartbeat interval in the biological information DB 1061 in chronological order in association with the information indicating the time when the heartbeat interval appeared. The biological information measuring unit 110 is an example of measuring means of the present invention.

The outlier extracting unit 120 extracts outliers, which are biometric information that does not meet predetermined criteria, from the biometric information for a predetermined period included in the teacher data stored in the teacher data DB 1062 in a learning process to be described later. Extract. Outliers are biometric information that is not measured from a normally sleeping person. Further, the outlier extraction unit 120 similarly extracts outliers from the biometric information stored in the biometric information DB 1061 in the sleep state determination process described later. The outlier extraction unit 120 is an example of the extraction means of the present invention.

It is empirically known that the heartbeat interval of a sleeping person is in the range of 500-1500 msec. Therefore, for example, as shown in FIG. 4A, the outlier extracting unit 120 outputs the biological information P3, P4, and P7 of the heartbeat intervals outside the range of 500 to 1500 msec among the biological information P1 to P8. Extract it as a lyre.

In addition, it is empirically known that the heartbeat interval of a sleeping person is generally stable, and that the heartbeat interval rarely changes suddenly during sleep. Therefore, for example, as shown in FIG. 4B, the outlier extracting unit 120 extracts biometric data from among the biometric information P9 to P16 in which the heartbeat interval has changed by a predetermined value or more (for example, 500 msec or more) from the previous one. Information P11 and P12 may be extracted as outliers.

It should be noted that the method for extracting outliers from biometric information is not limited to the above-described FIGS. be.

Returning to FIG. 3, the feature amount acquisition unit 130 acquires feature amounts to be input to the machine learning model from the biological information in the learning process and the sleep state determination process. Specifically, the feature amount acquisition unit 130 acquires body motion feature amounts from the outliers extracted by the extraction unit. For example, if the feature amount acquisition unit 130 acquires the ratio of the number of outliers extracted in a predetermined period and the number of inliers excluding the outliers from the biometric information extracted in the predetermined period as the body movement feature amount, good. The feature quantity acquisition unit 130 is an example of acquisition means of the present invention.

As described above, since the feature amount acquisition unit 130 extracts biological information that is not normally measured from a sleeping person as an outlier, when many outliers are extracted, the measurement position of the pulse wave sensor 200 and There is a high possibility that an accurate signal cannot be obtained due to an angle deviation, or that the pulse wave contains a signal due to body movement. In such cases, it is assumed that there is a lot of body movement. be. Therefore, the body motion feature amount (the ratio of the number of outliers to the number of inliers) acquired from the outliers of the biometric information serves as an index indicating the features of a person's body motion.

In addition, the feature amount acquisition unit 130 extracts sleep feature amounts from the inliers of the biometric information. Inliers do not include outliers that are not normally measured from sleeping people. Therefore, the sleep feature amount obtained only from the inliers of the biological information serves as an index that accurately indicates the features of the user's sleep.

The learning unit 140 uses teacher data stored in the teacher data DB 1062 to create (learn) a machine learning model for determining a sleep state. Note that the machine learning model learned by the learning unit 140 is a support vector machine, a random forest, or the like. Learning unit 140 is an example of a learning means of the present invention.

The sleep state determination unit 150 inputs the body movement feature amount acquired from the outlier of the biological information and the sleep feature amount acquired from the inlier by the acquisition unit into the machine learning model learned by the learning unit 140, thereby determining whether the user sleeps. determine the state. The sleep state determination unit 150 is an example of determination means of the present invention.

The determination result output unit 160 outputs information indicating the sleep state of the user determined by the sleep state determination unit 150 .

(learning process)
Next, processing of the sleep state determination device 100 configured as above will be described. First, the learning process executed by the sleep state determination device 100 will be described using the flowchart shown in FIG. Note that the teacher data DB 1062 stores a sufficient amount of teacher data for machine learning in advance. For example, the user performs a predetermined operation on the input device 105 to instruct the start of the learning process, and the learning process is started when the CPU receives the instruction.

First, the outlier extraction unit 120 selects one unselected teacher data from the teacher data DB 1062 (step S101). Then, the outlier extraction unit 120 extracts outliers, which are biometric information that does not meet a predetermined criterion, from a plurality of biometric information for a predetermined period included in the selected teacher data (step S102).

Subsequently, the feature amount acquiring unit 130 acquires body motion feature amounts based on the extracted outliers (step S103).

Specifically, the feature amount acquiring unit 130 calculates the ratio of the number of outliers extracted in step S102 among the biometric information included in the selected teacher data and the number of non-extracted biometric information (inliers). Acquire as a dynamic feature. For example, when biometric information P1 to P8 as shown in FIG. 4A is included in teacher data, as described above, three biometric information P3, P4, and P7 are outliers, and the other five are outliers. Pieces of biometric information P1, P2, P5, P6, and P8 are inliers. Therefore, 3/5 is calculated as the body motion feature amount.

Returning to FIG. 5, subsequently, the feature amount acquisition unit 130 acquires a sleep feature amount based on the inlier biometric information (heartbeat interval) obtained by excluding outliers from the biometric information included in the teacher data (step S104). . The sleep feature amount is, for example, the average value of the heartbeat interval, the variance, the low frequency component (LF) obtained by frequency-analyzing the change in the heartbeat interval by fast Fourier transform (FFT), the high frequency component (HF), and their ratio (LF/ HF).

In addition, when obtaining the low-frequency component (LF) and the high-frequency component (HF) described above, the feature amount acquisition unit 130 approximates the change in the measured heart rate with a spline curve, and calculates the frequency based on the spline curve. Analysis should be performed.

Subsequently, the outlier extraction unit 120 determines whether or not all the teacher data stored in the teacher data DB 1062 in step S101 have been selected (step S105). If there is unselected teacher data (step S105; No), the process returns to step S101, unselected teacher data is selected, and the body movement feature amount is calculated from the biological information for a predetermined period included in the teacher data. The process of acquiring the sleep feature quantity is repeated.

On the other hand, if all the teacher data have been selected (step S105; Yes), the learning unit 140 calculates feature amounts (body movement feature amount and sleep feature amount ) and the information indicating the sleep state attached to the teacher data, machine learning is performed to create a machine learning model (step S106). As a result, the values of the parameters defining the generated machine learning model are stored in the RAM 103 or the like. The learning process ends here.

(Sleep state determination processing)
Next, sleep state determination processing will be described using the flowchart of FIG. 6 . Note that the pulse wave sensor 200 measures the volume pulse wave in the period from when the user goes to bed to when he or she wakes up in advance. It is assumed that the biological information measurement unit 110 measures heartbeat intervals as biological information from the measured plethysmogram, attaches time information, and stores them in the biological information DB 1061 in chronological order. In addition, the learning process described above has also been executed, and a machine learning model optimal for sleep state determination has been created. In such a state, for example, the user performs a predetermined operation on the input device 105 to instruct the start of the sleep state determination process, and when the CPU 101 receives the instruction, the sleep state determination process is started.

First, the outlier extraction unit 120 extracts biometric information for a predetermined period (for example, 30 seconds) from the biometric information stored in the biometric information DB 1061 that has not yet been selected and has the oldest time information. Select (step S201). Then, the outlier extracting unit 120 extracts outliers, which are biometric information that does not match a predetermined criterion, from the biometric information for the selected predetermined period (step S202).

Subsequently, the feature amount acquiring unit 130 acquires body motion feature amounts based on the extracted outliers (step S203). The body motion feature quantity acquired here is the same type of feature quantity as the body motion feature quantity acquired in step S103 of the learning process.

Subsequently, the feature amount acquisition unit 130 acquires sleep feature amounts based on the inlier biometric information (heartbeat interval) obtained by excluding outliers from the biometric information for the selected predetermined period (step S204). The sleep feature quantity acquired here is the same type of feature quantity as the sleep feature quantity acquired in step S104 of the learning process.

Subsequently, the sleep state determination unit 150 determines the sleep state by inputting the body movement feature amount acquired in step S203 and the sleep feature amount acquired in step S204 into the machine learning model created in the learning process ( step S205). The sleep state determined here means the sleep state during the predetermined period selected in step S201.

Subsequently, the outlier extraction unit 120 determines whether or not all biometric information stored in the biometric information DB 1061 in step S201 has been selected (step S206). If there is unselected biometric information (step S206; No), the process returns to step S201, the biometric information for the unselected predetermined period is selected in chronological order, and the body movement feature amount and the sleep data are calculated from the selected biometric information. The process of acquiring the feature amount and determining the sleep state is repeated.

On the other hand, if all biological information has been selected (step S206; Yes), the determination result output unit 160 outputs the sleep state determination result (step S207). For example, the determination result output unit 160 creates a graph showing changes in the sleep state during the period from when the user goes to bed to when the user wakes up, based on the sleep state determined for each predetermined period in step S205, and causes the display device 104 to display the graph. With this, the sleep state determination processing ends.

As described above, according to the sleep state determination device 100 according to the present embodiment, outliers, which are biological information that does not match the predetermined criteria among the measured biological information, are used as indicators of the body movement of the user. A motion feature is acquired, and a sleep state is determined based on the body motion feature. That is, according to the present embodiment, the sleep state is determined using the information (body motion feature amount) regarding the user's body motion acquired from the biological information without using a dedicated device such as a motion sensor for measuring body motion. can be performed with high accuracy. In addition, since the present embodiment does not require a motion sensor or the like, it is possible to reduce the size and cost of the entire system.

Further, according to the sleep state determination device 100 according to the present embodiment, the sleep feature amount, which is an index indicating the user's sleep characteristics, is acquired based on the inliers obtained by excluding the outliers from the biological information for a predetermined period. be done. Since the sleep state is determined using not only the body movement feature amount but also the sleep feature amount, it is possible to further improve the accuracy of the sleep state determination.

(Modification)
It should be noted that the present invention is not limited to the above-described embodiments, and various modifications are of course possible without departing from the gist of the present invention.

For example, in the above embodiment, the sleep state is determined using the body motion feature amount and the sleep feature amount, but the sleep state may be determined using only the body motion feature amount. By doing so, it is possible to shorten the time required for the sleep state determination process.

For example, in step S104 of the learning process (see FIG. 5) and step S204 of the sleep state determination process (see FIG. 6), the feature amount acquisition unit 130 acquires the sleep feature amount from the biological information of the inliers excluding the outliers. did. However, the feature amount acquisition unit 130 not only excludes the outliers, but also replaces the excluded outliers with biological information having an average value of the inliers, etc. Based on the biological information replaced with the inliers, the sleep feature amount may be obtained. By doing so, the number of biometric information to be referred to in order to obtain the sleep feature amount increases, so it becomes possible to acquire a more accurate sleep feature amount.

Acquisition of the sleep feature amount described above will now be described with an example. FIG. 7 shows a biometric system in which outlier biometric information P3, P4, and P7 among the biometric information P1 to P8 shown in FIG. An example of replacement with information Q3, Q4, and Q7 is shown. In this case, the feature amount acquiring unit 130 acquires the sleep feature amount using not only the inlier biometric information P1, P2, P5, P6, and P8 but also the replaced biometric information Q3, Q4, and Q7.

Further, the feature amount acquisition unit 130 may acquire sleep feature amounts from all biological information without excluding outliers.

In the above embodiment, the ratio of the number of biometric information outliers to the number of inliers in a predetermined period of time is obtained as a body motion feature amount. Various information can be used as the body motion feature amount. For example, the number of outliers extracted in a predetermined period, the variance of the extracted outliers, or the like may be used as the body motion feature amount.

Also, in the above-described embodiment, the biometric information is the heartbeat interval, but the biometric information in the present invention is not limited to the heartbeat interval. For example, various information such as pulse intervals, pulse wave peak intervals, heart rate, and pulse rate can be applied to the present invention as biological information.

Further, in the above-described embodiment, the sleep state is determined by machine learning, but the sleep state may be determined by other methods. For example, the storage device 106 stores a table showing the relationship between the body movement feature amount and the sleep feature amount and the sleep state. Then, in the sleep state determination process, the sleep state determination unit 150 may determine the sleep state based on the body movement feature amount and the sleep feature amount obtained from the biological information and this table. In this case, it is not necessary to execute the learning process.

Further, in the above embodiment, sleep state determination apparatus 100 is connected to pulse wave sensor 200 via communication line 300, but may be connected to pulse wave sensor 200 wirelessly. Moreover, a pulse wave sensor may be incorporated in the sleep state determination device 100 .

In addition to being able to provide the sleep state determination device 100 having a configuration for realizing the function according to the present invention in advance, by applying a program, an existing information processing device or the like can be provided as the sleep state determination device 100 according to the present invention. can also function as That is, by applying a program for realizing each functional configuration of the sleep state determination device 100 exemplified in the above embodiment so that a CPU or the like that controls an existing information processing device can execute it, the present invention can be achieved. It can function as the sleep state determination device 100 .

Moreover, the application method of such a program is arbitrary. The program can be applied by storing it in a computer-readable storage medium such as a flexible disk, CD (Compact Disc)-ROM, DVD (Digital Versatile Disc)-ROM, memory card, or the like. Furthermore, a program can be superimposed on a carrier wave and applied via a communication medium such as the Internet. For example, the program may be posted and distributed on a bulletin board (BBS: Bulletin Board System) on a communication network. Then, the above processing may be executed by starting this program and executing it in the same manner as other application programs under the control of an OS (Operating System).

Although the preferred embodiments of the present invention have been described above, the present invention is not limited to such specific embodiments, and the present invention includes the invention described in the claims and their equivalents. included. The invention described in the original claims of the present application is appended below.

(Appendix 1)
a device other than a device that directly measures body movement, measuring means for measuring the user's biological information;
Acquisition means for acquiring a body motion feature quantity, which is an index indicating the body movement of the user, based on the biometric information measured by the measurement means;
determination means for determining a sleep state of the user based on the body motion feature amount acquired by the acquisition means;
A sleep state determination device comprising:

(Appendix 2)
further comprising extraction means for extracting outliers, which are biological information that does not match a predetermined criterion, from the biological information measured by the measuring means;
The acquisition means acquires the body motion feature quantity based on the outliers extracted by the extraction means.
The sleep state determination device according to Supplementary Note 1, characterized by:

(Appendix 3)
The extracting means extracts the outlier from the biometric information measured in the predetermined period for each predetermined period,
The acquiring means acquires the body motion feature amount based on the extracted outliers for each predetermined period of time,
The determining means determines the sleep state of the user in the predetermined period based on the body motion feature amount for each predetermined period.
The sleep state determination device according to appendix 2, characterized by:

(Appendix 4)
Based on the ratio between the number of outliers extracted by the extracting means and the number of inliers that are biological information from which the outliers have been deleted from the biological information measured by the measuring means, the acquisition means Get the body motion feature quantity,
The sleep state determination device according to appendix 2 or 3, characterized by:

(Appendix 5)
The acquiring means acquires the body motion feature quantity based on the number of the outliers extracted by the extracting means.
The sleep state determination device according to appendix 2 or 3, characterized by:

(Appendix 6)
The acquisition means further acquires a sleep feature amount that is an index indicating the sleep characteristics of the user based on the inlier, which is the biometric information obtained by removing the outlier from the biometric information measured by the measurement means,
The determination means determines a sleep state based on the body movement feature amount and the sleep feature amount,
The sleep state determination device according to any one of appendices 2 to 5, characterized by:

(Appendix 7)
The acquisition means acquires the sleep feature amount based on the inliers and the biometric information obtained by replacing the outliers with biometric information that matches a predetermined criterion.
The sleep state determination device according to appendix 6, characterized by:

(Appendix 8)
The acquisition means acquires the body motion feature amount and the sleep feature amount, which is an index indicating the sleep feature of the user, from the biological information measured by the measurement means,
The determination means determines a sleep state based on the body movement feature amount and the sleep feature amount.
The sleep state determination device according to any one of appendices 1 to 7, characterized by:

(Appendix 9)
The measurement means measures biological information including the heartbeat interval of the user.
The sleep state determination device according to any one of appendices 1 to 8, characterized by:

(Appendix 10)
Further comprising learning means for learning a machine learning model for determining a sleep state based on the body motion feature amount in a predetermined period and information indicating the sleep state in the predetermined period,
The determination means determines the sleep state using the machine learning model learned by the learning means.
The sleep state determination device according to any one of appendices 1 to 9, characterized by:

(Appendix 11)
a measurement step of measuring the biological information of the user by a device other than a device that directly measures body movement;
an acquisition step of acquiring a body motion feature quantity, which is an index indicating body motion of the user, based on the biometric information measured in the measurement step;
a determination step of determining the sleep state of the user based on the body motion feature amount acquired in the acquisition step;
A sleep state determination method comprising:

(Appendix 12)
the computer,
Acquisition means for acquiring a body movement feature amount, which is an index indicating body movement of the user, based on biological information measured by a device other than a device that directly measures body movement;
determination means for determining a sleep state of the user based on the body motion feature amount acquired by the acquisition means;
A program characterized by functioning as

DESCRIPTION OF SYMBOLS 1... Sleep state determination system, 100... Sleep state determination apparatus, 200... Pulse wave sensor, 300... Communication line, 101... CPU, 102... ROM, 103... RAM, 104... Display device, 105... Input device, 106... Memory Apparatus 1061 Biological information DB 1062 Teacher data DB 107 I/F unit 110 Biological information measurement unit 120 Outlier extraction unit 130 Feature amount acquisition unit 140 Learning unit 150 Sleep state determination unit 160 ... determination result output unit

Claims (11)

measuring means for measuring the user's biometric information including the heartbeat interval ;
Acquisition means for acquiring a body motion feature amount , which is an index indicating body motion of the user, based on the biometric information not measured from a sleeping person, which is included in the biometric information measured by the measurement means;
determination means for determining a sleep state of the user based on the body motion feature amount acquired by the acquisition means;
A sleep state determination device comprising: further comprising extraction means for extracting outliers, which are biological information that does not meet a predetermined criterion from among the biological information measured by the measuring means, as biological information that is not measured from the sleeping person ,
The acquisition means acquires the body motion feature amount based on the outliers extracted by the extraction means.
The sleep state determination device according to claim 1, characterized by: The extracting means extracts the outlier from the biometric information measured in the predetermined period for each predetermined period,
The acquiring means acquires the body motion feature amount based on the extracted outliers for each predetermined period of time,
The determining means determines the sleep state of the user in the predetermined period based on the body motion feature amount for each predetermined period.
The sleep state determination device according to claim 2, characterized in that: Based on the ratio between the number of outliers extracted by the extracting means and the number of inliers that are biological information from which the outliers have been deleted from the biological information measured by the measuring means, the acquisition means Get the body motion feature quantity,
4. The sleep state determination device according to claim 2 or 3, characterized in that: The acquiring means acquires the body motion feature quantity based on the number of the outliers extracted by the extracting means.
4. The sleep state determination device according to claim 2 or 3, characterized in that: The acquisition means further acquires a sleep feature amount that is an index indicating the sleep characteristics of the user based on the inlier, which is the biometric information obtained by removing the outlier from the biometric information measured by the measurement means,
The determination means determines a sleep state based on the body movement feature amount and the sleep feature amount,
The sleep state determination device according to any one of claims 2 to 5, characterized in that: The acquisition means acquires the sleep feature amount based on the inliers and the biometric information obtained by replacing the outliers with biometric information that matches a predetermined criterion,
The sleep state determination device according to claim 6, characterized by: The acquisition means acquires the body motion feature amount and the sleep feature amount, which is an index indicating the sleep feature of the user, from the biological information measured by the measurement means,
The determination means determines a sleep state based on the body movement feature amount and the sleep feature amount,
The sleep state determination device according to any one of claims 1 to 7, characterized in that: Further comprising learning means for learning a machine learning model for determining a sleep state based on the body motion feature amount in a predetermined period and information indicating the sleep state in the predetermined period,
The determination means determines the sleep state using the machine learning model learned by the learning means.
The sleep state determination device according to any one of claims 1 to 8, characterized in that: A sleep state determination method executed by a sleep state determination device,
Acquiring biological information including heartbeat intervals of a user measured by an external sensor, and indicating the body movement of the user based on the biological information not measured from a sleeping person included in the acquired biological information an acquisition step of acquiring a body motion feature quantity that is an index;
a determination step of determining the sleep state of the user based on the body motion feature amount acquired in the acquisition step;
A sleep state determination method comprising: the computer,
Acquiring biological information including heartbeat intervals of a user measured by an external sensor, and indicating the body movement of the user based on the biological information not measured from a sleeping person included in the acquired biological information Acquisition means for acquiring a body motion feature value as an index;
determination means for determining a sleep state of the user based on the body motion feature amount acquired by the acquisition means;
A program characterized by functioning as

Images