JP7126651B2 - Sleep depth measurement device, sleep depth measurement method, computer program and trained model - Google Patents

Description

The present invention relates to a sleep depth measurement device, a sleep depth measurement method, a computer program, and a trained model.

Patent Document 1 discloses a sleep depth determination unit that determines sleep depth using biological information of a sleeping user, a muscle electrical stimulation intensity determination unit that determines the intensity of muscle electrical stimulation based on the sleep depth, an electrical muscle stimulation output unit that outputs electrical muscle stimulation at the intensity determined by the electrical muscle stimulation intensity determination unit using electrodes placed on the user's skin, so as not to disturb sleep of the user. Meanwhile, an electrical stimulation system is disclosed that produces a training effect through electrical muscle stimulation.

In Patent Document 2, based on sleep information including time-series data indicating changes in sleep depth of a subject and wake-up time information indicating the time at which an alarm sounds, at the time indicated by the wake-up time information, wake up well. A sleep management device is disclosed that improves a sense of wakefulness upon waking by calculating a recommended sleep onset time that is estimated to be possible and notifying the subject of the calculated recommended sleep onset time.

In Patent Document 3, gain control is performed on the user's heartbeat signal to control the peak value to be constant, and normalization processing is performed on the strength of the heartbeat signal calculated using the gain value at that time. sleep stage by determining the sleep stage based on the normalized heart rate intensity obtained by applying and the variance value indicating the variation of the data for a predetermined time calculated for the data of the obtained normalized heart rate strength. A sleep stage determination device capable of determining with accuracy is disclosed.

JP 2017-64390 A JP 2017-46960 A JP 2016-202463 A

On the other hand, in the electrical stimulation system of Patent Document 1 and the sleep stage determination device of Patent Document 3, sleep depth is determined based on only one type of biological information, and the reliability of such determination results is poor.

In addition, in the sleep management device of Patent Document 2, although it is described that the depth of sleep is determined based on the autonomic nerve activity index and the pulse wave deviation, the method for determining the depth of sleep is not specified. The reliability of judgment results cannot be guaranteed.

However, in order to increase the reliability of sleep depth, it is appropriate to make determinations based on multiple types of biological data. In addition, it is also required to improve the versatility of such a device in preparation for an unspecified number of users. However, Patent Documents 1 to 3 are not devised at all regarding this matter.

The present invention has been made in view of such circumstances, and the purpose thereof is to provide a sleep depth measurement device, a sleep depth measurement method, and a versatile sleep depth measurement device capable of measuring sleep depth with higher reliability. It is to provide a computer program and a trained model.

A sleep depth measurement device according to the present embodiment is a sleep depth measurement device that measures the depth of sleep of a sleeper. a calculation unit that calculates a feature value based on biometric data of a species; a determination unit that performs a determination that the feature value is equal to or greater than a threshold value or a determination that the feature value is less than the threshold value; and the determination in time series. and a determination unit that determines the depth of sleep based on the determination result of the unit.

The sleep depth measurement device according to the present embodiment is characterized in that the plurality of types of biological data includes data related to body movements of the sleeper during sleep.

The depth-of-sleep measurement apparatus according to the present embodiment is characterized in that the plurality of types of biological data further include data relating to the sleeper's heartbeat and respiration during sleep.

The sleep depth measurement apparatus according to the present embodiment is characterized in that the determining unit determines the depth of sleep based on a ratio of the above determination or the below determination within a predetermined period.

The sleep depth measurement device according to the present embodiment is characterized in that the learning model uses parameters to output the feature value within a predetermined range.

The depth-of-sleep measurement apparatus according to the present embodiment is characterized by comprising an adjustment unit that adjusts the parameter based on the calculation result of the feature value.

The sleep depth measurement device according to the present embodiment is characterized in that the parameters include weights.

The sleep depth measurement device according to the present embodiment is characterized by comprising an output unit that outputs data representing the sleep depth determined by the determination unit.

The sleep depth measurement method according to the present embodiment is a sleep depth measurement method for measuring sleep depth with a sleep depth measurement device, in which multiple types of biological data related to a sleeper are acquired, and a plurality of acquired biological data are acquired using a learning model. calculating a feature value related to sleep depth based on the biological data of the species, determining whether or not the feature value is equal to or greater than a threshold value, and determining the sleep depth based on the result of the determination in time series; including.

A computer program according to the present embodiment uses a learning model in a computer to calculate a feature value based on a plurality of types of biometric data relating to an acquired sleeper, and determines whether the feature value is equal to or greater than a threshold. It is determined whether or not, and processing for determining the depth of sleep is executed based on the result of the determination in time series.

The trained model according to the present embodiment includes an input layer into which body motion data, heartbeat data and respiration data per unit time are input, and body motion data, heartbeat data and respiration data and parameters input from the input layer. and an output layer that outputs a feature value indicating the degree of sleep based on the output from the intermediate layer. Body movement data per unit time obtained from a bed sensor. , heart rate data and respiration data are input to the input layer, and used in the process of specifying the feature value obtained from the output layer through the calculation by the intermediate layer.

The learned model according to the present embodiment is characterized in that the specified feature value is used for greater than or equal to a threshold value or greater than or less than a threshold value determination.

According to the present invention, by measuring the depth of sleep of a sleeper based on a plurality of types of biological data, it is possible to measure the depth of sleep with higher reliability and increase the versatility of such a device.

1 is a block diagram showing a configuration of a main part of a sleep depth measurement device according to Embodiment 1; FIG. FIG. 2 is a functional block diagram showing the essential configuration of a control section of the depth-of-sleep measuring device according to the present embodiment; FIG. 4 is a flow chart showing processing of sleep depth measurement in the sleep depth measurement device according to the present embodiment. FIG. 2 is a conceptual diagram conceptually showing processing of sleep depth measurement in the sleep depth measurement device according to the present embodiment; FIG. 10 is a block diagram showing the main configuration of a sleep depth measurement device according to Embodiment 2;

The depth-of-sleep measurement device, the depth-of-sleep measurement method, the computer program, and the learned model according to the embodiments of the present invention will be described below in detail with reference to the drawings.

(Embodiment 1)
FIG. 1 is a block diagram showing a main configuration of a sleep depth measurement device according to Embodiment 1. FIG. In the figure, reference numeral 20 denotes a sleep depth measurement device according to this embodiment. A sleep sensor 10 is connected to the sleep depth measurement device 20 .

The sleep sensor 10 is, for example, a sheet-shaped sensor and is installed between the bed mattress and the sheets. The sleep sensor 10 detects biological data such as the pulse, respiration, body movement, and body temperature of the person lying on the bed from the time the person enters the bed to the time the person leaves the bed. The sleep sensor 10 outputs the detected biological data to the sleep depth measurement device 20 .

Note that the sleep sensor 10 and the sleep depth measurement device 20 may be integrated. In addition, the sleep sensor 10 is installed at the base of the bed or on the wall of the room, and uses microwaves or infrared rays to detect the pulse, respiration, body movement, body temperature, etc. of the person lying on the bed. good.

The sleep depth measuring device 20 acquires biological data of a sleeping person lying on a bed (hereinafter referred to as a sleeper) from the sleep sensor 10, and measures the depth of sleep of the sleeper based on the acquired biological data. Measure.

The sleep depth measurement device 20 includes a control unit 21, a main memory unit 22, an auxiliary memory unit 23, a sensor I/F (InterFace) 24, a timer unit 25, a display unit 26, and an instruction reception unit 27. I have. These components are connected to each other by a bus. Information devices such as general-purpose personal computers, tablets, and smartphones may also be used as the sleep depth measurement device 20 .

The main storage unit 22 is an SRAM (Static RAM), a DRAM (Dynamic RAM), a flash memory, a semiconductor memory disk, or the like. The main storage unit 22 temporarily stores necessary information during the processing performed by the control unit 21 and the program being executed by the control unit 21 . In addition, the main storage unit 22 stores a threshold, which will be described later.

The auxiliary storage unit 23 is SRAM, DRAM, flash memory, semiconductor memory disk, hard disk, or the like. The auxiliary storage unit 23 stores programs to be executed by the control unit 21 and various information necessary for executing the programs.

For example, the auxiliary storage unit 23 stores computer programs related to processing in a calculation unit 212, a determination unit 213, a determination unit 214, and an adjustment unit 215, which will be described later.
The sleep depth measurement device 20 according to the present embodiment may be configured such that the computer program is provided in a portable recording medium such as a CD-ROM via an I/F unit (not shown). . Furthermore, the sleep depth measurement device 20 according to the present embodiment may be configured such that the computer program can be downloaded from an external device via a communication unit (not shown).

The sensor I/F 24 (acquisition unit) is an interface that acquires biological data from the sleep sensor 10 . The sensor I/F 24 sends the acquired biometric data to the control unit 21 via the bus.
The display unit 26 (output unit) is composed of, for example, an LCD or an EL (Electroluminescence) panel, and displays the depth of sleep determined by the control unit 21 as described later.

The instruction receiving unit 27 is a so-called operation unit or the like. The instruction receiving unit 27 receives from the user an instruction to display the depth of sleep of the predetermined sleeper, an instruction to transmit the depth of sleep to an external device (for example, the user's smart phone), and the like.

The control unit 21 controls the sleep depth measurement device 20 . One or a plurality of CPUs, GPUs (Graphics Processing Units), multi-core CPUs, or the like is used for the control unit 21 . The control unit 21 measures the depth of sleep of the sleeper based on biological data acquired from the sleep sensor 10 via the sensor I/F 24 .

Here, the depth of sleep is indicated by six stages of wakefulness, REM sleep, stage 1, stage 2, stage 3 and stage 4, for example. Arousal is the state of being awake. REM sleep is a state of light sleep in which the body is asleep but the brain is active. Stages 1 to 4 are states of so-called non-REM sleep, of which sleep stages 3 and 4 indicate a state of deep sleep. In each stage, the larger the number, the deeper the state of sleep.

FIG. 2 is a functional block diagram showing the main configuration of the controller 21 of the sleep depth measurement device 20 according to this embodiment. The control unit 21 includes a CPU 211 , a calculation unit 212 , a determination unit 213 , a determination unit 214 and an adjustment unit 215 .

The calculation unit 212 calculates a score (characteristic value) relating to the depth of sleep of the sleeper based on biological data acquired from the sleep sensor 10 via the sensor I/F 24 . For example, the biological data includes multiple types of biological data. The plurality of types of biological data includes, for example, at least body motion data (data relating to body motion) representing body motion of the sleeper during sleep. In addition, the plurality of types of biological data include heartbeat data (data related to heartbeat) representing the heart rate of the sleeper during sleep, breathing data (data related to breathing) representing the number of breaths (exhalation or inhalation) of the sleeper during sleep. related data). A case where the biometric data consists of the body motion data, the heartbeat data, and the respiration data will be described below as an example. However, this embodiment is not limited to this. The body motion data may be intervals between motions of the sleeper, the heart rate data may be intervals between pulses, and the respiration data may be intervals between respirations. It may be an interval.

The calculator 212 uses a learning model. The learning model has excellent versatility, uses predetermined parameters, and outputs the score so that it falls within a predetermined range. The calculation unit 212 calculates the score by inputting the biometric data into the learning model. A detailed description will be given below.

The learning model includes an input layer to which body motion data, heartbeat data and respiration data per unit time are input, and an intermediate layer that performs calculations based on the body motion data, heartbeat data and respiration data input from the input layer and parameters. A learned model comprising a layer and an output layer that outputs a feature value indicating the degree of sleep based on the output from the intermediate layer, body movement data per unit time obtained from the sleep sensor 10 (bed sensor), Heart rate data and respiration data are input to the input layer and used in the process of identifying feature values obtained from the output layer through computation by the intermediate layer.

The learning model uses, for example, a so-called neural network. For convenience of explanation, it is assumed that the learning model has an input layer, an intermediate layer, and an output layer (see FIG. 4). Also, the input layer, the intermediate layer, and the output layer each have a plurality of nodes.
For example, the biometric data is input to the input layer and output from each node of the input layer to the nodes of the intermediate layer, where the parameters are multiplied. Then, the biometric data multiplied by the parameter is output from each node of the intermediate layer to a node of the output layer, and is further multiplied by the parameter. The parameters include, for example, weights (transformation matrices) or biases. The learning model is configured to perform the above-described processing when the biometric data is input, and to calculate a number (score) between 0 and 1, which is a characteristic value indicating the degree of sleep.

The output layer has 11 output layers with increments of 0.1 such as 0, 0.1, 0.2, 0.3, . . . , 0.9, 1.0. Each probability may be output.

For example, the learning model is generated by repeating (learning) processing from the input layer to the output layer while adjusting the parameters. That is, the learning model performs so-called supervised learning in advance using reference biometric data for learning (supervised data). Input the reference biometric data for learning into the learning model, take the error between the output score and the correct answer (for example, "0.5"), propagate the error to each layer so as to approach the correct answer Adjust the parameters of each layer (node). This series of processes is repeated to reduce the error with the correct answer. Through such learning, the learning model according to the present embodiment is finally completed.

The score (feature value) specified by the learning model is used to make an above-determination that the score is equal to or greater than the threshold or a less-than-determination that the score is less than the threshold, as described later.

The calculation unit 212 calculates the score using the learning model from the biological data (body movement data, heartbeat data, and respiration data) per unit time (one second) acquired from the sleep sensor 10 . The calculation unit 212 sends the calculated score to the determination unit 213 every second.
The sleep depth measurement device 20 according to this embodiment is not limited to the above description. For example, the calculation unit 212 may calculate the score every 10 seconds and send the calculated score to the determination unit 213, or the calculation unit 212 may calculate the score at time intervals longer than 10 seconds. Alternatively, the calculation unit 212 may calculate the score using the average value of biometric data within a predetermined period of time.

Alternatively, a plurality of pieces of time-series biometric data may be subjected to convolution processing and then input to the neural network. Also, a recurrent neural network (RNN) may be used for time-series biometric data.

The determination unit 213 determines whether or not the score calculated by the calculation unit 212 is greater than or equal to the threshold value stored in the main storage unit 22 . For example, the threshold is "0.75", and the determination unit 213 performs an above determination that the score is equal to or greater than the threshold, or a less than determination that the score is less than the threshold. The judging unit 213 outputs "1" when the above judgment is made, and outputs "0" when the below judgment is made. Hereinafter, "0" output with the above determination and "1" output with the less than determination are collectively referred to as an output result. The determination by the determination unit 213 is performed, for example, every second, and the output result associated with such determination is stored in the main storage unit 22 .

The determination unit 214 determines the depth of sleep of the sleeper at predetermined time intervals (for example, 10 minutes) based on the determination result of the determination unit 213 . The determination unit 214 determines the depth of sleep based on the output result of the determination unit 213 stored in the main storage unit 22 for a predetermined period of time. The determining unit 214 determines the depth of sleep based on the ratio of "0" or "1" within the predetermined time. For example, the determination unit 214 compares the number of “0”s or “1s” within the predetermined time with a plurality of thresholds stored in advance in the main storage unit 22 to determine which one of the above six stages determine the sleep depth of Specifically, the determining unit 214 determines the depth of sleep in order of wakefulness, REM sleep, stage 1, stage 2, stage 3, and stage 4, assuming that the greater the number of “0” or “1”, the deeper the sleep. decide. A result of determination by the determination unit 214 is stored in the main storage unit 22 .

The sleep depth measurement device 20 according to this embodiment is not limited to the above description. for example,
The determining unit 214 determines the depth of sleep by comparing the number of consecutive "0"s or "1s" within the predetermined time with a plurality of thresholds stored in advance in the main storage unit 22. may be configured.

The adjuster 215 mitigates individual differences among sleepers by adjusting the parameters used in the learning model. In other words, people have different sleeping habits, and some people turn over during their sleep. In measuring the depth of sleep of such a person (hereinafter referred to as a singular person), if a learning model similar to that of a person belonging to the so-called standard group is used, it cannot be said that the depth of sleep is accurately measured. Therefore, it is necessary to change the learning model for such a peculiar person according to the person. In such a case, the adjustment unit 215 adjusts the parameters to change the learning model.

The adjustment unit 215 needs to read out the determination result of the determination unit 214 within a predetermined period (for example, one month) from the main storage unit 22 and change the parameters based on the determination result of the determination unit 214. Determine whether or not

For example, in the determination result of the determination unit 214 within the predetermined period, if the ratio or number of wakefulness or REM sleep state is equal to or greater than a predetermined threshold value, or equal to or greater than the average value of people belonging to a standard group, The adjuster 215 determines that the parameters need to be adjusted.
If it is determined that the parameters need to be adjusted, the adjustment unit 215 adjusts the parameters of the learning model so that the score output from the calculation unit 212 approaches the average value (for example, 0.5) of people belonging to the standard group. to adjust. That is, learning is performed by repeating the processing from the input layer to the output layer of the learning model while adjusting the parameters so that the score output by the calculation unit 212 approaches 0.5.

In this way, the adjusting unit 215 adjusts the parameters of the learning model for outputting the score, so that the individual differences of sleepers are reflected in the learning model, so that it is possible to accurately measure (determine) the depth of sleep. .

The sleep depth measurement device 20 according to this embodiment is not limited to the above description. For example, a plurality of learning models with different weights or biases are stored in the auxiliary storage unit 23, and one of the plurality of learning models is selected according to the determination result of the determining unit 214 within a predetermined period. You can change it. That is, within the predetermined period, if the number of wakefulness or REM sleep states is greater than a predetermined threshold value, if it is greater than the average value of people belonging to the standard group, etc., in any of the plurality of learning models, calculation It is configured to vary the learning model used by unit 212 . The calculation unit 212 can select the optimal learning model for the sleeping person by repeating the exchange of learning models (learning).

The CPU 211 executes the control program stored in the auxiliary storage unit 23 to control each component described above, and operates the entire device as the sleep depth measurement device 20 according to the present embodiment.

FIG. 3 is a flowchart showing the processing of sleep depth measurement in the sleep depth measurement device 20 according to the present embodiment, and FIG. It is a conceptual diagram schematically shown. Hereinafter, processing of sleep depth measurement in the sleep depth measurement device 20 according to the present embodiment will be described based on FIGS. 3 and 4. FIG.

A person (sleeping person) is included in sleep after lying down in bed for sleep, the sleep sensor 10 detects the biological data of the sleeping person during sleep, and the detected biological data is sent to the sleep depth measuring device 20 output to Detection of biological data by the sleep sensor 10 is performed until the sleeper wakes up and leaves the bed.

At this time, the sleep depth measurement device 20 acquires biological data from the sleep sensor 10 via the sensor I/F 24 (step S101). The sensor I/F 24 sends the acquired biometric data to the control unit 21 .

The biometric data is input to the calculation unit 212, and body motion data, heartbeat data, and respiration data are input to the input layer of the calculation unit 212, as shown in FIG. The calculation unit 212 performs the above-described processing using the learning model M on the body motion data, the heartbeat data, and the respiration data, and calculates a predetermined score (step S102).

For example, as illustrated in FIG. 4, in such a learning model M, weights W11 to W19 (or biases) are applied to values related to body motion data, heartbeat data, and respiration data at each node of the input layer. It is multiplied and output to each corresponding node in the intermediate layer. Further, each node in the intermediate layer is further multiplied by weights W21 to W29 (or bias) and output to each corresponding node in the output layer. As a result, the learning model M finally outputs a score within the range of 0-1. The score calculated by the calculation unit 212 in this manner is input to the determination unit 213 .

The determination unit 213 determines whether or not the score input from the calculation unit 212 is equal to or greater than the threshold (step S103). The determination unit 213 outputs "1" when the score is equal to or greater than the threshold value, i.e., the above determination, and outputs "0" when the score is less than the threshold value, i.e., the less than determination. to output The output result accompanying the determination by the determination unit 213 is stored in the main storage unit 22 (step S104). That is, the main storage unit 22 stores "0" or "1" in chronological order.

At this time, the CPU 211 instructs the timing unit 25 to start timing, and the timing unit 25 starts timing according to the instruction from the CPU 211 . The CPU 211 determines whether or not a predetermined period of time (for example, 10 minutes) has elapsed based on the time measurement result of the time measurement unit 25 (step S105).

When the CPU 211 determines that the predetermined time has not elapsed (step S105: NO), the process proceeds to step S108.
When the CPU 211 determines that the predetermined time has passed (step S105: YES), the determination unit 214 determines the depth of sleep of the sleeper (step S106). The determination unit 214 determines the depth of sleep based on the output result (ratio of “0” or “1”) from the determination unit 213 stored in the main storage unit 22 for a predetermined period of time. The method of determining the depth of sleep by the determination unit 214 has already been described, and detailed description thereof will be omitted. The determination result of the determination unit 213 is stored in the main storage unit 22 (step S107). That is, the main storage unit 22 stores one of the six levels of sleep depth in chronological order.

When the process in step S107 is finished in this way, or when it is determined in step S105 that the predetermined time has not elapsed, the CPU 211 determines whether biological data has been acquired from the sleep sensor 10. (Step S108). Such determination is made by the CPU 211 monitoring the sensor I/F 24 .
When the CPU 211 determines that biometric data has been acquired from the sleep sensor 10 (step S108: YES), the process returns to step S101. Further, when the CPU 211 determines that the biometric data is not acquired from the sleep sensor 10 (step S108: NO), for example, when the sleeper wakes up, the process ends.

The sleep depth measurement device 20 according to this embodiment is not limited to the above description. For example, data representing the sleep depth determined by the determination unit 214 may be output to the display unit 26 and the sleep depth may be displayed on the display unit 26 .

Moreover, the sleep depth measurement device 20 according to the present embodiment is not limited to the above description. For example, the sleep depth measurement device 20 according to the present embodiment includes a communication unit (output unit), and the data representing the sleep depth determined by the determination unit 214 is transmitted to an external portable information device (for example, a smartphone, a mobile phone, tablet, etc.).

Furthermore, the sleep depth measurement device 20 according to the present embodiment is not limited to the above description. If it is determined in step S105 of FIG. 3 that the predetermined time has not elapsed, and if it is determined in step S108 that the biometric data has not been obtained, the processing of steps S106 and S107 is performed without ending the process. can be configured to For example, the determination unit 214 may be configured to determine the depth of sleep based on the ratio of "0" or "1" stored in the main storage unit 22 up to that time.

Further, the adjustment unit 215 adjusts the parameters used in the learning model M based on the determination result of the determination unit 214 within the predetermined period stored in the main storage unit 22, thereby to change the learning model M. The parameter adjustment by the adjustment unit 215 has already been described, and detailed description thereof will be omitted.

As described above, in the sleep depth measurement device 20 according to the present embodiment, the depth of sleep of the sleeper is measured using a plurality of types of biological data, so it is possible to obtain the depth of sleep with a higher degree of reliability. be.

Further, in the sleep depth measurement device 20 according to the present embodiment, the sleep depth is measured (determined) based on the score calculated by the calculator 212 using the learning model M having a neural network configuration. In addition, neural networks have a high expressive ability to approximate functions. Therefore, it is possible to improve the degree of approximation of the functions for the multiple types of input biometric data and the output score. Therefore, it is possible to obtain a sleep depth with higher accuracy, and the versatility of the sleep depth measurement device 20 can be improved.

(Embodiment 2)
In sleep depth measurement device 20 according to the present embodiment, determination processing by determination unit 213 and determination processing of sleep depth by determination unit 214 are performed by a server, which will be described later.

FIG. 5 is a block diagram showing the main configuration of the sleep depth measurement device 20 according to Embodiment 2. As shown in FIG. A sleep depth measurement device 20 according to the present embodiment includes a server 40 . For example, server 40 is connected via network 31 . Note that a plurality of sleep depth measurement devices 20 and portable information devices may be connected to the network 31 .

In the sleep depth measurement device 20 according to the present embodiment, as in Embodiment 1, the control unit 21 has the CPU 211, the calculation unit 212, and the adjustment unit 215, but does not have the determination unit 213 and the determination unit 214 ( not shown). In sleep depth measurement device 20 according to the present embodiment, server 40 is configured to perform the functions of determination unit 213 and determination unit 214 in the first embodiment.

Sleep depth measurement device 20 according to the present embodiment includes communication section 28 (output section) that transmits the score calculated by control section 21 (calculation section 212 ) to server 40 . The communication unit 28 transmits the calculated score to the server 40 via the network 31 .

The server 40 is, for example, a large computer installed in a data center, a general-purpose personal computer, or an information processing device such as a tablet.
The server 40 includes a server control section 45 and a server communication section 46, which are connected to each other via a bus. The server communication section 46 receives the calculated score from the communication section 28 via the network 31 .

The server control unit 45 includes a server CPU 41 , a server determination unit 42 , a server determination unit 43 and a storage unit 44 .

The server determination unit 42 determines whether or not the received score is equal to or greater than the threshold value stored in the storage unit 44, like the determination unit 213 does. Further, the server determination unit 43 determines the sleep depth of the sleeper at predetermined time intervals based on the determination result of the server determination unit 42, similarly to the determination unit 214. FIG. Since the server determination unit 42 and the server determination unit 43 correspond to the determination unit 213 and the determination unit 214 in Embodiment 1, respectively, detailed description of the server determination unit 42 and the server determination unit 43 is omitted.

The storage unit 44 is an SRAM (Static RAM), a DRAM (Dynamic RAM), a flash memory, a semiconductor memory disk, or the like. The storage unit 44 temporarily stores necessary information during the process performed by the server CPU 41 and the program being executed by the server CPU 41 . The storage unit 44 also stores the score received by the server communication unit 46 from the communication unit 28, the output result accompanying the determination by the server determination unit 42, and the determination result by the server determination unit 43. FIG.

The sleep depth measurement device 20 according to this embodiment is not limited to the above description. For example, the sleep depth measurement device 20 may include a plurality of information processing devices, and each information processing device may perform different processing. For example, the sleep depth measurement device 20 may separately have an information processing device for learning model generation and an information processing device for sleep depth determination.

The same reference numerals are assigned to the same parts as in the first embodiment, and detailed description thereof will be omitted.

Note that the calculation unit 212, the determination unit 213, the determination unit 214, and the adjustment unit 215 described above may be configured by hardware logic, or may be configured in software by the CPU 211 executing a predetermined program. good.
Moreover, the server determination unit 42 and the server determination unit 43 described above may be configured by hardware logic, or may be constructed by software when the server CPU 41 executes a predetermined program.

In the above description, the case where the learning model M having the neural network has three layers has been described as an example. can be done.

Furthermore, in the above description, a case where the biometric data is pulse, respiration, body movement, body temperature, etc. has been described as an example, but the sleep depth measurement device 20 according to the present embodiment is not limited to this. The biometric data may further include information related to sleep breathing or snoring.
As for breathing, the sleep sensor 10 measures the periodicity of the sleeper's breathing and the like, and such measurement results can be used as biometric data. As for snoring, the sleep sensor 10 measures the presence or absence or degree of snoring of the sleeper, and such measurement results can be used as biometric data.

The sleep depth measurement device 20 according to the present embodiment may measure sleep depth using all biological data related to pulse, respiration, body movement, body temperature, sleep breathing, and snoring. may be used selectively to measure sleep depth.

20 sleep depth measuring device 24 sensor I/F
42 server determination unit 43 server determination unit 212 calculation unit 213 determination unit 214 determination unit 215 adjustment unit 26 display unit 28 communication unit M learning model W weight

Claims (10)

In a sleep depth measurement device that measures the depth of sleep of a sleeper,
an acquisition unit that acquires multiple types of biometric data relating to a sleeping person;
Based on multiple types of biological data acquired using a learning modelindicate the degree of sleepa calculation unit that calculates a feature value;
A determination unit that performs an above determination that the feature value is equal to or greater than a threshold or a less than determination that it is less than a threshold;
within a specified period in time seriespluralBased on the determination result of the determination unit,every predetermined perioda determination unit that determines sleep depth;
A sleep depth measurement device comprising: 2. The depth-of-sleep measurement device according to claim 1, wherein the plurality of types of biological data includes data relating to body movements of the sleeper during sleep. 3. The depth-of-sleep measurement apparatus according to claim 2, wherein the plurality of types of biological data further includes data relating to heartbeat and respiration of the sleeper during sleep. 4. The depth-of-sleep measuring apparatus according to claim 1, wherein the determination unit determines the depth of sleep based on the ratio of the above determination or the below determination within the predetermined period. 5. The depth-of-sleep measuring device according to claim 1, wherein the learning model uses parameters to output the feature value within a predetermined range. 6. The depth-of-sleep measurement apparatus according to claim 5, further comprising an adjustment unit that adjusts the parameter based on the calculation result of the feature value. 7. The sleep depth measuring device according to claim 5, wherein said parameters include weights. 8. The depth-of-sleep measurement device according to any one of claims 1 to 7, further comprising an output unit that outputs data representing the depth of sleep determined by the determination unit. sleep depth measuring deviceButIn the sleep depth measurement method for measuring sleep depth,
Acquiring multiple types of biometric data related to sleepers,
Sleep based on multiple types of biological data acquired using a learning modelindicates the degree ofCalculate the feature value,
The feature value is, judgment that it is equal to or greater than the threshold, or judgment that it is less than the thresholdand
within a specified period in time seriespluralBased on the result of the determination,every predetermined periodA sleep depth measurement method comprising determining sleep depth. on the computer,
Using a learning model, based on multiple types of biometric data related to sleepers acquiredindicate the degree of sleepCalculate the feature value,
The feature value is, judgment that it is equal to or greater than the threshold, or judgment that it is less than the thresholdand
within a specified period in time seriespluralBased on the result of the determination,every predetermined periodA computer program that performs the process of determining sleep depth.

Images