Sleep-wake determination system, sleep-wake determination method, and program

Abstract

[Problem] To provide a sleep / waking determination system and the like that determine sleep and waking with substantially high accuracy using few worn instruments. [Solution] One aspect of the present invention provides a sleep / waking determination system. The sleep / waking determination system comprises at least one processor that can execute a program to perform the following steps. An acquisition step for acquiring a signal that indicates the acceleration of at least a portion of the body of a user. A band limiting step for limiting a frequency component of the signal that indicates the acceleration to a specific frequency component. A transformation step for performing a Fourier transformation on a signal that consists of the specific frequency component to generate determination data. A determination step for determining the sleep and waking of the user on the basis of the determination data and preset reference data.

Description

[Technical field]

[0001] The present invention relates to a sleep / awake determination system, a sleep / awake determination method, and a program. [Background technology]

[0002] To maintain good health, people need to get healthy sleep, and it is known that sleep disorders such as insomnia, sleep-related breathing disorders, and hypersomnia can cause health problems. To understand a person's sleep state, it is necessary to examine how that person actually sleeps over a night or several days.

[0003] As a method for examining a person's sleep state, an overnight polysomnography (PSG) has been developed, as proposed in Patent Document 1 and the like. In PSG, a large number of electrodes and sensors are attached to the body of the subject, and the electrodes and sensors are connected to dedicated measuring instruments to measure basic data such as electroencephalograms, electrocardiograms, electromyograms, and respiratory states, and the sleep and wakefulness states are examined based on the basic data. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] JP 2013-99507 A Summary of the Invention [Problem to be solved by the invention]

[0005] In sleep tests using PSG, such as those in Patent Document 1, many measuring instruments are used, and the test locations are limited to hospitals and laboratories. Therefore, many people cannot easily perform long-term tests for more than several days. In addition, attaching many electrodes and sensors to the body in an environment different from home can cause stress and make it difficult to sleep. Therefore, it is difficult to properly test normal sleep conditions.

[0006] In view of the above circumstances, the present invention provides a sleep / wake determination system and the like that determines sleep / wake with sufficiently high accuracy using a small number of wearing devices. [Means for solving the problem]

[0007] According to one aspect of the present invention, there is provided a sleep / wake determination system. The sleep / wake determination system includes at least one processor capable of executing a program to perform the following steps: In the acquisition step, a signal indicating the acceleration of at least a part of a user's body is acquired. In the band limiting step, frequency components included in the signal indicating the acceleration are limited to specific frequency components. In the conversion step, a signal composed of the specific frequency components is Fourier transformed to generate determination data. In the determination step, the user's sleep / wake state is determined based on the determination data and preset reference information.

[0008] According to this aspect, it is possible to determine whether the person is asleep or awake with sufficiently high accuracy using a small number of wearing devices. [Brief description of the drawings]

[0009] [Figure 1] 1 is a diagram showing an outline of the configuration of a sleep / wake determination system 1 according to the present embodiment. [Diagram 2] 2 is a block diagram showing a hardware configuration of the sleep / wake determination system 1 shown in FIG. [Diagram 3] 1 is a photograph showing an example of a wearable device 2. [Figure 4] 2 is a functional block diagram of a control unit 33 in the information processing device 3. FIG. [Diagram 5] 2 is an activity diagram showing the flow of information processing in the sleep / wake determination system 1. FIG. [Figure 6] FIG. 13 is a conceptual diagram for explaining epoch E. [Figure 7] FIG. 13 is a conceptual diagram for explaining epoch E. [Figure 8]FIG. 8A shows a graph showing the L2 norm of acceleration per epoch (FIG. 8A) and a hypnogram showing actual sleep (FIG. 8B). [Figure 9] 13 is a graph showing the accuracy of the trained model IF1 for each sampling frequency. [Figure 10] 11 is a graph showing the accuracy of sleep / wake determination for each bit rate in analog-to-digital conversion. [Figure 11] 13 is a schematic diagram showing the standardization process performed by a correction unit 334. FIG. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0010] Hereinafter, embodiments of the present invention will be described with reference to the drawings. Various features shown in the following embodiments can be combined with each other.

[0011] Incidentally, the program for realizing the software appearing in this embodiment may be provided as a non-transitory computer-readable recording medium, or may be provided so as to be downloadable from an external server, or may be provided so that the program is started on an external computer and its functions are realized on a client terminal (so-called cloud computing).

[0012] In addition, in this embodiment, the term "unit" may include, for example, a combination of hardware resources implemented by a circuit in the broad sense and software information processing that can be specifically realized by these hardware resources. In addition, in this embodiment, various information is handled, and this information is represented, for example, by physical values ​​of signal values ​​representing voltage and current, high and low signal values ​​as a binary bit collection consisting of 0 or 1, or quantum superposition (so-called quantum bits), and communication and calculation can be performed on the circuit in the broad sense.

[0013] In addition, a circuit in the broad sense is a circuit realized by at least appropriately combining a circuit, circuitry, a processor, a memory, etc. In other words, it includes an application specific integrated circuit (ASIC), a programmable logic device (e.g., a simple programmable logic device (SPLD), a complex programmable logic device (CPLD), and a field programmable gate array (FPGA)), etc.

[0014] 1. Hardware Configuration In this section, the overall configuration of the sleep / wake determination system 1 will be described.

[0015] 1.1 Sleep-wake determination system 1 Fig. 1 is a diagram showing an outline of the configuration of a sleep / awake determination system 1 according to this embodiment. The sleep / awake determination system 1 is a system including a wearable device 2 and an information processing device 3, which can exchange information with each other via electrical communication means. Fig. 2 is a block diagram showing the hardware configuration of the sleep / awake determination system 1 shown in Fig. 1.

[0016] 1.2 Wearable devices2 Fig. 3 is a photograph showing an example of the wearable device 2. As shown in Fig. 3, the wearable device 2 is a small device that can be worn on the arm of a user, for example. As shown in Fig. 2, the wearable device 2 has a communication unit 21, a storage unit 22, a control unit 23, and an acceleration sensor 24, and these components are electrically connected via a communication bus 20 inside the wearable device 2. Each component will be further described below.

[0017] The communication unit 21 is preferably implemented to have wired communication means such as USB, IEEE1394, Thunderbolt (registered trademark), wired LAN network communication, etc., wireless LAN network communication, mobile communication such as 3G / LTE / 5G, Bluetooth (registered trademark) communication, etc. as necessary. In particular, in this embodiment, the communication unit 21 is preferably configured to be able to write information including a time-series three-dimensional acceleration vector v(x, y, z) measured by an acceleration sensor 24 described later to an external storage medium M. The type and form of the storage medium M are not particularly limited, and for example, a flash memory, a card-type memory, an optical disk, etc. may be appropriately adopted.

[0018] The memory unit 22 stores various information defined by the above description. This can be implemented as a storage device such as a solid state drive (SSD), or as a memory such as a random access memory (RAM) that stores temporarily required information (arguments, arrays, etc.) related to program calculations. It may also be a combination of these. In particular, it can store information including a time-series three-dimensional acceleration vector v(x, y, z) measured by an acceleration sensor 24 described later. Note that it may also be implemented so that the information is stored directly in the above-mentioned storage medium M without going through the memory unit 22.

[0019] The control unit 23 processes and controls the overall operation related to the wearable device 2. The control unit 23 is, for example, a central processing unit (CPU) not shown. The control unit 23 realizes various functions related to the wearable device 2 by reading out a predetermined program stored in the storage unit 22. That is, information processing by software stored in the storage unit 22 is specifically realized by the control unit 23, which is an example of hardware, and can be executed as each functional unit included in the control unit 23. Note that the control unit 33 is not limited to being single, and may be implemented to have multiple control units 23 for each function. Also, a combination of these may be used.

[0020] The acceleration sensor 24 is configured to be able to measure the acceleration of a part of the user's body, for example, an arm, as three-dimensional vector information. In other words, information including a time-series three-dimensional acceleration vector v(x, y, z) can be obtained from the user.

[0021] 1.3 Information processing device 3 2, the information processing device 3 has a communication unit 31, a storage unit 32, and a control unit 33, and these components are electrically connected via a communication bus 30 inside the information processing device 3. Each component will be further described below.

[0022] The communication unit 31 is preferably implemented to have wired communication means such as USB, IEEE1394, Thunderbolt (registered trademark), wired LAN network communication, wireless LAN network communication, mobile communication such as 3G / LTE / 5G, Bluetooth (registered trademark) communication, etc. as necessary. In particular, in this embodiment, the communication unit 31 is preferably implemented as a storage medium reading unit capable of reading information stored in an external storage medium M. The storage medium M stores information including a time-series three-dimensional acceleration vector v(x, y, z) acquired from the user by the wearable device 2. Thus, the communication unit 31, which is a storage medium reading unit, is configured to be able to read the three-dimensional acceleration vector v(x, y, z) stored in the storage medium M.

[0023] The storage unit 32 stores various pieces of information defined by the above description. This can be implemented as a storage device such as a solid state drive (SSD), or as a memory such as a random access memory (RAM) that stores temporarily required information (arguments, arrays, etc.) related to the calculation of a program. It may also be a combination of these. In particular, the storage unit 32 stores various programs and the like related to the information processing device 3 executed by the control unit 33.

[0024] Furthermore, the storage unit 32 stores a trained model that has learned correlations between the feature value f(N) of a desired epoch, the feature value f(N±δ) of a recent epoch, and the user's sleep and wakefulness. As the machine learning algorithm, it is preferable to appropriately adopt a conventional algorithm. Examples include logistic regression, random forest, XGBoost, and multilayer perceptron (MLP). Furthermore, every time the information processing device 3 is used, machine learning can be further performed using the training data to update the trained model.

[0025] The control unit 33 (an example of a processor) processes and controls the overall operation related to the information processing device 3. The control unit 33 is, for example, a central processing unit (CPU) not shown. The control unit 33 realizes various functions related to the information processing device 3 by reading out a predetermined program stored in the storage unit 32. That is, information processing by software (stored in the storage unit 32) is specifically realized by hardware (the control unit 33), and can be executed as each functional unit described below. Note that, although a single control unit 33 is shown in FIG. 2, in reality, this is not limited to this, and multiple control units 33 may be provided for each function. Also, a combination of these may be used.

[0026] 2. Functional configuration In this section, the functional configuration of this embodiment will be described. As described above, information processing by software stored in the storage unit 32 is specifically realized by the control unit 33, which is an example of hardware, and can be executed as each functional unit included in the control unit 33. Fig. 4 is a functional block diagram of the control unit 33 in the information processing device 3. That is, the control unit 33 includes the following units.

[0027] The acquisition unit 331 is configured to acquire various information from the outside via a network, a storage medium M, or the like. For example, the acquisition unit 331 may acquire a signal indicating the acceleration of at least a part of the user's body. This will be described in further detail later.

[0028] The band limiting unit 332 is configured to perform band limiting processing on the signal acquired by the acquisition unit 331. For example, the band limiting unit 332 may limit the frequency components included in the signal indicating acceleration to a specific frequency component C. This will be described in further detail later.

[0029] The conversion unit 333 is configured to perform a Fourier transform process on the signal consisting of the specific frequency component C limited by the band limiting unit 332. For example, the signal consisting of the specific frequency component C may be subjected to a Fourier transform to generate data for determination. This will be described in further detail later.

[0030] The correction unit 334 is configured to execute a correction process. For example, the correction unit 334 may standardize and correct the distribution of frequency components of the determination data generated by the conversion unit 333. This will be described in more detail later.

[0031] The determination unit 335 is configured to determine whether the user is asleep or awake. For example, the determination unit 335 may determine whether the user is asleep or awake based on the specific frequency component C and preset reference information IF. This will be described in more detail later.

[0032] The display control unit 336 is configured to generate various display information and control the display contents viewable by the user. The display information may be information itself generated in a form viewable by the user, such as a screen, an image, an icon, text, etc., or may be rendering information for displaying the screen, an image, an icon, text, etc.

[0033] 3. Information processing method In this section, an information processing method of the sleep / wake determination system 1 described above will be described.

[0034] (Overview of the process) FIG. 5 is an activity diagram showing the flow of information processing in the sleep / wake determination system 1. The flow of information processing will be outlined below along each activity in FIG. 5. The user here is assumed to be a person who wishes to be determined whether he / she is asleep or awake using the service provided by the sleep / wake determination system 1. Note that the determination of sleep / wake here may include determining whether the user is in a sleeping state or an awake state. The user wears the wearable device 2 on, for example, his / her arm.

[0035] First, a user is lying down and resting with the wearable device 2 attached to at least a part of the user's body, for example, one arm. During this time, the acceleration sensor 24 in the wearable device 2 sequentially detects and measures the acceleration of the user's one arm (activity A101). Log data of signals indicating the measured acceleration is sequentially stored in the storage medium M inserted in the wearable device 2.

[0036] Next, after sufficient log data of the signal indicating the acceleration has been acquired, the storage medium M is replaced from the wearable device 2 to the information processing device 3, thereby transferring the signal indicating the acceleration stored in the storage medium M to the information processing device 3. In other words, the acquisition unit 331 acquires a signal indicating the acceleration of at least a part of the user's body (activity A102). The acquired signal indicating the acceleration may be read into a temporary storage area of ​​the storage unit 32.

[0037] Next, the control unit 33 reads out a predetermined program stored in the storage unit 32, thereby limiting the frequency components included in the acceleration signal to the specific frequency components C. In other words, the band limiting unit 332 limits the frequency components included in the signal indicating the acceleration to the specific frequency components C, and particularly preferably, the band limiting unit 332 limits the frequency components to the specific frequency components C using a high-pass filter HF (activity A103). According to this embodiment, the specific frequency components C can be limited to high-frequency components equal to or higher than the cutoff frequency, and the user's sleep / wake state can be determined more robustly.

[0038] Next, the control unit 33 reads out a predetermined program stored in the storage unit 32, thereby executing a Fourier transform on the signal composed of the specific frequency component C. It is preferable that the Fourier transform uses an FFT (Fast Fourier Transformation) algorithm. In other words, the conversion unit 333 performs a Fourier transform on the signal composed of the specific frequency component C to generate data for determination (activity A104).

[0039] Next, the control unit 33 standardizes the frequency components of the generated determination data by reading out a predetermined program stored in the storage unit 32. In other words, the correction unit 334 standardizes and corrects the frequency components of the determination data (distribution of the specific frequency component C) (activity A105).

[0040] Next, the standardized determination data is treated as an epoch E defined in units of a predetermined time t. Specifically, the control unit 33 reads out a predetermined program stored in the storage unit 32, and calculates a feature value f(N) for each epoch E (activity A106). Here, N is the serial number of the epoch E, and will be described in more detail later.

[0041] Thereafter, the control unit 33 reads out the reference information IF stored in the storage unit 32 (activity A107), and provides the feature amount f of the desired epoch E and the feature amount f of the epoch E defined around it to the reference information IF, thereby displaying the judgment result regarding the sleep and wakefulness of the user. Particularly preferably, the reference information IF is a trained model IF1 that has learned the correlation between the specific frequency component C and sleep and wakefulness. According to this embodiment, the sleep and wakefulness of the user can be judged with high accuracy using machine learning.

[0042] In other words, the determination unit 335 determines whether the user is asleep or awake based on the determination data including the specific frequency component C and the preset reference information IF. Preferably, the determination unit 335 specifies a feature value f for each epoch E defined by a predetermined time t based on the specific frequency component C. The determination unit 335 also determines whether the user is asleep or awake based on the feature value f of a desired epoch E among the epochs E, the feature value f of a recent epoch E that is earlier than the desired epoch E in the time series, and a trained model IF1 that is an example of the reference information IF. Then, the display control unit 336 controls to display the determination result so as to be presentable to the user (activity A108). According to this aspect, the user's sleep or wakefulness can be determined with higher accuracy.

[0043] To summarize the above, this sleep / wake determination method includes each step in the sleep / wake determination system 1. In the acquisition step, a signal indicating the acceleration of at least a part of the user's body is acquired. In the band limiting step, frequency components included in the signal indicating the acceleration are limited to specific frequency components C. In the conversion step, a signal consisting of specific frequency components C is Fourier transformed to generate determination data. In the determination step, the user's sleep / wake state is determined based on the determination data and preset reference information.

[0044] According to this embodiment, it is possible to determine whether the user is asleep or awake with sufficiently high accuracy using a small number of wearing devices. In particular, by first limiting the band and then performing a Fourier transform, it is possible to make the feature value f in the determination data prominent, which contributes to the determination of whether the user is asleep or awake. In addition, since the device does not depend on the sampling frequency at which the device acquires the signal, it is possible to more robustly determine whether the user is asleep or awake.

[0045] (Details of information processing) Here, the details of the information processing outlined in FIG. 5 will be described. FIGS. 6 and 7 are conceptual diagrams for explaining epoch E. FIG. 8 is a diagram showing a graph (FIG. 8A) showing the L2 norm of acceleration for each epoch, and a hypnogram (FIG. 8B) showing actual sleep. FIG. 9 is a graph showing the accuracy of the trained model IF1 for each sampling frequency. FIG. 10 is a graph showing the accuracy of sleep / wake determination for each bit rate in analog-to-digital conversion. FIG. 11 is a schematic diagram showing the standardization process by the correction unit 334.

[0046] 6, an epoch E is signal log data of acceleration defined for a predetermined time t. This predetermined time t is, for example, 5 to 600 seconds, preferably 10 to 120 seconds, and more preferably 20 to 60 seconds. Specifically, for example, 10, 20, 30, 40, 50, 60, 70, 80, 90, 100, 110, 120, 130, 140, 150, 160, 170, 180, 190, 200, 210, 220, 230, 240, 250, 260, 270 , 280, 290, 300, 310, 320, 330, 340, 350, 360, 370, 380, 390, 400, 410, 420, 430, 440, 450, 460, 470, 480, 490, 500, 510, 520, 530, 540, 550, 560, 570, 580, 590, 600 seconds, and may be within a range between any two of the numbers listed here.

[0047] In this embodiment, a signal indicating acceleration is treated as a plurality of epochs E. The Nth epoch E in the time series is referred to as a desired epoch E, and epochs E slightly older than that, for example, N-1 to N-4th epochs E, are referred to as recent epochs E. In this embodiment, the feature value f(N) of the desired epoch E and the feature value f(N-δ) of the recent epoch E are input, and the user's sleep / wake state is determined based on the above-mentioned trained model stored in the storage unit 32.

[0048] Specifically, the value of δ is, for example, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, and may be within a range between any two of the values ​​exemplified here. Of course, the number of epochs is merely an example and is not limited to this. Also, instead of the feature f(N-δ), a feature f(N±δ) may be adopted as shown in FIG. 7. In such a case, for example, the N-4th to N+4th epochs E may be referred to as surrounding epochs E.

[0049] In addition, the feature value f of the epoch E is not particularly limited. For example, the feature value f(N) may be extracted from the L2 norm of a 3D acceleration vector v(x, y, z) expressing acceleration, which is subjected to a high-pass filter HF, FFT, and standardization processing, and used to determine whether the subject is asleep or awake. Specifically, for example, the feature value f(N) may be a histogram generated by classifying a scalar value such as the L2 norm or its logarithm using a plurality of thresholds, or may be a power spectrum based on a product value obtained by multiplying a scalar value by a window function. For example, FIG. 8A shows a logarithmic power spectrum of the time difference of acceleration. And FIG. 8B shows a hypnogram correlated to FIG. 8A.

[0050] Next, as shown in Fig. 9, it was found that the accuracy of judgment changes depending on the sampling frequency of the training data when generating the trained model IF1. Specifically, the trained model IF1 was generated with the training data sampling frequencies of 50 Hz, 25 Hz, 10 Hz, and 5 Hz in the order of Fig. 8A to Fig. 8D.

[0051] Overall, there was a tendency for the accuracy score to be higher when the sampling frequency of the training data was matched to the sampling frequency of the input sample signal. Furthermore, in particular, if the sampling frequency of the training data when generating the trained model IF1 was 25 Hz or higher, relatively high accuracy could be achieved even if the sampling frequency of the input sample signal was low. In addition, there was no significant difference between the cases of resampling and thinning when changing the sampling frequency.

[0052] From the above, it is preferable that the trained model IF1 is a trained model IF1 that has learned the correlation between the specific frequency component C and sleep and wakefulness by acquiring a signal at a sampling frequency of 5 Hz or more. In addition, it is preferable that the sampling frequency of the sample signal input to the trained model IF1 is also 5 Hz or more. When the sampling frequency of the input sample is matched to that during training for such a trained model, highly accurate judgment can be realized.

[0053] More preferably, the trained model IF1 is a trained model IF1 that has learned the correlation between the specific frequency component C and sleep and wakefulness by acquiring a signal at a sampling frequency of 25 Hz or more. According to this embodiment, even if the sampling frequency of the input sample is lower than that during training, highly accurate determination can be achieved.

[0054] That is, the sampling frequencies of the signals used in learning the learned model IF1 are, for example, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98, 99, 100 Hz, and may be within a range between any two of the values ​​exemplified here.

[0055] Furthermore, as shown in FIG. 10, it was confirmed that the bit rate in analog-digital conversion affects the accuracy of the determination. According to FIG. 10, it was found that when resampling processing was performed, a difference was observed between 4 bits and 6 bits, and when thinning processing was performed, a difference was observed between 6 bits and 8 bits. That is, preferably, the acquisition unit 331 performs analog-digital conversion of the acceleration at a bit rate of 8 bits or more to acquire a signal. Specifically, for example, the bit rate may be 8, 10, 12, 14, 16, 32, 48, or 64 bits, and may be within a range between any two of the numerical values ​​exemplified here. According to such an embodiment, the accuracy of the digitally converted signal can be maintained high, so that the sleep and wakefulness of the user can be determined more robustly. Note that the analog-digital conversion itself may be performed at 8 bits or more, or may be performed so that the acquired signal is ultimately at 8 bits or more.

[0056] Furthermore, as shown in FIG. 11, it has been confirmed that the frequency sensitivity of the wearable device 2 varies depending on the product. Here, the cases of the wearable device 2a and the wearable device 2b are illustrated. In this embodiment, in order to mitigate errors due to the variance in the frequency sensitivity of each product, a standardization process by the correction unit 334 is introduced. By carrying out the standardization process, the distribution of the frequency sensitivity of both the wearable devices 2a and 2b becomes substantially uniform. That is, in this embodiment, by including the standardization process by the correction unit 334, the influence of the frequency sensitivity is reduced regardless of the product, and the user's sleep / wake state can be determined more robustly.

[0057] 4.Other The sleep / wake determination system 1 according to this embodiment may be configured as follows.

[0058] In the above embodiment, the configuration of the sleep / wake determination system 1 has been described, but at least one computer may be provided with a program for executing each step in the sleep / wake determination system 1. According to such an embodiment, it is possible to determine sleep / wake with sufficiently high accuracy using a small number of wearing devices. In addition, since the device does not depend on the sampling frequency at which the signal is acquired, it is possible to more robustly determine the sleep / wake state of the user.

[0059] Log data of signals indicating acceleration may be transferred from the wearable device 2 to the information processing device 3 via a communication network such as the Internet, an intranet, or dedicated wireless communication, without going through the storage medium M. Furthermore, the measurement of acceleration by the wearable device 2 and the determination of sleep and wakefulness by the information processing device 3 may be performed online in approximately real time with a certain latency.

[0060] In this embodiment, the acquisition unit 331, the band limiting unit 332, the conversion unit 333, the correction unit 334, the determination unit 335 and the display control unit 336 are described as functional units realized by the control unit 33 of the information processing device 3, but at least a part of them may be implemented as functional units realized by the control unit 23 of the wearable device 2.

[0061] Furthermore, the wearable device 2 and the information processing device 3 may be integrated together. In other words, the information processing device 3 is the wearable device 2 that can be worn by a user on a part of the body, and further includes an acceleration sensor 24, and the acceleration sensor 24 may be configured to be capable of measuring a three-dimensional acceleration vector v(x, y, z).

[0062] The band limiting unit 332 may be configured to use a band pass filter instead of the high pass filter HF to limit the specific frequency component C. According to such an embodiment, the specific frequency component can be limited to a preferable component, and the sleep / wake state of the user can be determined more robustly.

[0063] 5 may be reversed. In other words, the correction unit 334 may standardize and correct the distribution of frequency components of the acquired signal, and the band limiting unit 332 may limit the standardized frequency components to the specific frequency components C. According to such an embodiment, it is possible to reduce the influence of frequency sensitivity that differs from device to device, and to more robustly determine whether the user is asleep or awake.

[0064] In addition to the acceleration sensor 24 of the present embodiment, other sensors may be added as appropriate. For example, an SpO2 sensor, an ambient light sensor, a heart rate sensor, etc. may be added. The SpO2 sensor may measure transcutaneous arterial blood oxygen saturation, and the result may be additionally used to determine whether the user is asleep or awake. The ambient light sensor may measure the intensity of the user's ambient light, and the result may be additionally used to determine whether the user is asleep or awake. The heart rate sensor may measure the user's heart rate, and the result may be additionally used to determine whether the user is asleep or awake.

[0065] Furthermore, it may be provided in the following aspects:

[0066] (1) A sleep-wake determination system comprising at least one processor capable of executing a program to perform each of the following steps: an acquisition step acquires a signal indicating the acceleration of at least a part of a user's body; a band limiting step limits frequency components contained in the signal indicating the acceleration to specific frequency components; a conversion step performs a Fourier transform on the signal composed of the specific frequency components to generate determination data; and a determination step determines whether the user is asleep or awake based on the determination data and preset reference information.

[0067] (2) In the sleep / wake determination system described in (1) above, the band limiting step limits the specific frequency components using a high-pass filter or a band-pass filter.

[0068] (3) In the sleep / wake determination system according to (1) or (2) above, the correction step further comprises standardizing and correcting a distribution of frequency components of the determination data.

[0069] (4) In the sleep / wake determination system described in (1) or (2) above, the correction step further comprises standardizing and correcting the distribution of frequency components of the acquired signal, and the band limiting step limits the standardized frequency components to the specific frequency components.

[0070] (5) In the sleep / wake determination system according to any one of (1) to (4) above, in the acquiring step, the acceleration is analog-to-digital converted at a bit rate of 8 bits or more to acquire the signal.

[0071] (6) In the sleep / wake determination system described in any one of (1) to (5) above, the reference information is a trained model that has learned the correlation between the specific frequency component and the sleep and wakefulness.

[0072] (7) In the sleep-wake determination system described in (6) above, the trained model is a trained model that has learned the correlation between the specific frequency component and the sleep and wakefulness by acquiring the signal at a sampling frequency of 5 Hz or more.

[0073] (8) In the sleep-wake determination system described in (7) above, the trained model is a trained model that has learned the correlation between the specific frequency component and the sleep and wakefulness by acquiring the signal at a sampling frequency of 25 Hz or more.

[0074] (9) In the sleep / wake determination system described in any one of (1) to (8) above, in the determination step, features are identified for each epoch defined by a predetermined time based on the determination data, and the user's sleep / wake is determined based on the features of a desired epoch among the epochs, the features of a recent epoch that is earlier than the desired epoch in the time series, and the reference information.

[0075] (10) A sleep / awake determination method, comprising the steps of the sleep / awake determination system according to any one of (1) to (9) above.

[0076] (11) A program for causing at least one computer to execute each step of the sleep / wake determination system according to any one of (1) to (9) above. Of course, this is not the case.

[0077] Finally, although various embodiments of the present invention have been described, these are presented as examples and are not intended to limit the scope of the invention. The novel embodiments can be implemented in various other forms, and various omissions, substitutions, and modifications can be made without departing from the spirit of the invention. The embodiments and their modifications are included within the scope and spirit of the invention, and are included in the scope of the invention and its equivalents described in the claims. [Explanation of symbols]

[0078] 1: Sleep / wake detection system 2: Wearable devices 2a: Wearable devices 2b: Wearable devices 3: Information processing equipment 20: Communication bus 21: Communications Department 22: Storage section 23: Control section 24: Acceleration sensor 30: Communication bus 31: Communications Department 32: Storage section 33: Control section 331: Acquisition Department 332: Bandwidth limiting section 333: Conversion section 334: Correction section 335: Judgment section 336: Display control unit C: Specific frequency component E: Epoch HF: High pass filter IF : Reference information IF1: Trained model M:Storage medium f : feature t: predetermined time v : 3D acceleration vector

Claims

1. A sleep / wake determination system, comprising: At least one processor capable of executing a program to perform the following steps: The acquiring step includes acquiring a signal indicative of an acceleration of at least a part of a body of the user; In the decision step, Identifying a feature amount for each epoch defined by a predetermined time based on a specific frequency component included in the signal indicating the acceleration; The device determines whether the user is asleep or awake based on the feature amount of a desired epoch among the epochs, the feature amount of a recent epoch that is earlier than the desired epoch in the time series, and preset reference information.

2. The sleep / wake determination system according to claim 1, Furthermore, in the band limiting step, a frequency component included in the signal indicating the acceleration is limited to the specific frequency component, Furthermore, in the conversion step, the signal composed of the specific frequency components is Fourier transformed to generate data for determination; In the determining step, whether the user is asleep or awake is determined based on the determination data and preset reference information.

3. The sleep / wake determination system according to claim 2, The band limiting step uses a high-pass filter or a band-pass filter to limit the specific frequency components.

4. The sleep / wake determination system according to claim 2, Furthermore, in the correction step, the distribution of frequency components of the judgment data is standardized and corrected.

5. The sleep / wake determination system according to claim 2, Furthermore, in the correction step, a distribution of frequency components of the acquired signal is standardized and corrected; The band limiting step limits the standardized frequency components to the specific frequency components.

6. The sleep / wake determination system according to claim 1, In the acquiring step, the acceleration is subjected to analog-to-digital conversion at a bit rate of 8 bits or more, and the signal is acquired.

7. The sleep / wake determination system according to claim 1, The reference information is a trained model that has learned the correlation between the specific frequency component and the sleep and wakefulness.

8. The sleep / wake determination system according to claim 7, The trained model is a trained model that has learned the correlation between the specific frequency component and the sleep and wakefulness by acquiring the signal at a sampling frequency of 5 Hz or more.

9. The sleep / wake determination system according to claim 8, The trained model is a trained model that has learned the correlation between the specific frequency component and the sleep and wakefulness by acquiring the signal at a sampling frequency of 25 Hz or higher.

10. A sleep / wake determination method, comprising: A method comprising the steps of the sleep / wake determination system according to any one of claims 1 to 9.

11. A program, A system for causing at least one computer to execute each step in the sleep / wake determination system according to any one of claims 1 to 9.

Images