Determination device and program

Abstract

The present invention provides a determination device that can appropriately determine the user's condition based on the user's vibrations while they are in bed. [Solution] A determination device comprising an acquisition unit for acquiring vibrations of a user on a bed and a control unit, the acquisition unit for acquiring vibration data from the acquisition unit, a calculation unit for calculating determination data by deconvolving the vibration data, and a determination unit for determining the user's sleeping position or sleeping posture on the bed by inputting the determination data calculated from the user's vibration data on the bed into a trained model generated with the determination data as input data and the user's sleeping position or sleeping posture as output.

Description

Technical Field

[0001] The present invention relates to a determination device and the like.

Background Art

[0002] For example, Patent Document 1 discloses an invention related to a bed device including a vibration detection unit that is disposed on a bed and separately detects a component along the longitudinal direction of the bed of vibrations applied to the bed and a component along the vertical direction with respect to the bed, a signal extraction unit that extracts a heartbeat vibration signal derived from the heartbeat of a living body existing on the bed from the vibrations detected by the vibration detection unit, and a state estimation unit that estimates the posture of the living body on the bed based on the heartbeat vibration signal extracted by the signal extraction unit.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] An object of the present disclosure is to provide a determination device and the like that can appropriately determine the state of a user based on the vibrations of the user on a bed.

Means for Solving the Problems

[0005] The determination device of the present disclosure is a determination device including an acquisition unit that acquires vibrations of a user on a bed and a control unit, the acquisition unit that acquires vibration data from the acquisition unit, a calculation unit that calculates determination data obtained by performing deconvolution on the vibration data, and a determination unit that determines the sleeping position or sleeping posture of the user on the bed by inputting the determination data into a learned model generated with the determination data as input data and the sleeping position or sleeping posture of the user as output.

[0006] The program disclosed herein provides a computer capable of communicating with a determination device comprising an acquisition unit for acquiring vibrations of a user on a bed and a control unit, and includes an acquisition function for acquiring vibration data from the acquisition unit, a calculation function for calculating determination data by deconvolving the vibration data, and a determination function for determining the user's sleeping position or posture on the bed by inputting the determination data calculated from the user's vibration data on the bed into a trained model generated with the determination data as input data and the user's sleeping position or posture as output. [Effects of the Invention]

[0007] According to this disclosure, it is possible to provide a determination device, etc., that can appropriately determine the state of a user on a bed device based on the vibrations of the user on the bed device. [Brief explanation of the drawing]

[0008] [Figure 1] This is a diagram illustrating the overall structure of this embodiment. [Figure 2] This diagram illustrates the functional configuration of the hardware in this embodiment. [Figure 3] This is a diagram illustrating the configuration of the sensor in this embodiment. [Figure 4] This is a diagram illustrating the functional configuration of the software in this embodiment. [Figure 5] This is an operation flow diagram illustrating the main processing in this embodiment. [Figure 6] This is an operation flow illustrating the reference model generation process in this embodiment. [Figure 7] This is an operation flow illustrating the judgment data acquisition process in this embodiment. [Figure 8] This figure illustrates an example of data processing in this embodiment. [Figure 9]This figure illustrates an example of data processing in this embodiment. [Figure 10] This figure illustrates an example of data processing in this embodiment. [Figure 11] This is an operation flow illustrating the sleeping position determination process in this embodiment. [Figure 12] This is an operation flow diagram illustrating the process for determining sleeping posture in this embodiment. [Figure 13] This diagram illustrates the effects of applying this embodiment. [Figure 14] This is an operation flow illustrating a different embodiment of the judgment data acquisition process of this embodiment. [Modes for carrying out the invention]

[0009] Hereinafter, one embodiment for carrying out the present invention will be described with reference to the drawings. Specifically, the case in which the determination device of the present invention is applied will be described, but the scope to which the present invention is applicable is not limited to this embodiment.

[0010] Generally, a known technique involves using multiple load sensors to calculate the user's center of gravity and determine their sleeping position. However, this method is not very user-friendly, as it requires multiple load sensors to be pre-installed in the bed or to be placed beneath the bed.

[0011] Furthermore, while some devices attempt to determine the user's sleeping position based on vibrations, it has been difficult to accurately determine the user's sleeping position and posture in such cases.

[0012] To address these challenges, a determination device that can accurately determine a user's sleeping position and posture using a simple method will be described using the following embodiment.

[0013] Here, the user refers to a person who actually uses the bed system 1, such as the bed 3 and the determination device 10. For example, the user refers to a patient admitted to a hospital or a facility, a person requiring care, or a person lying on the bed 3 (mattress 5) at home.

[0014] Also, in the present embodiment, when referring to staff, it means a person who supports the user. For example, the staff includes medical staff such as doctors and nurses in a hospital, care staff in a facility, and the family members of the user at home.

[0015] Also, in the present embodiment, when referring to an operator, it means a person who operates the bed system 1 (bed 3, determination device 10). The operator is mainly staff, but when the user operates, the user is included in the operator.

[0016] [1. Overall System] FIG. 1 is a diagram for explaining the overall outline of the bed system 1 to which the determination device of the present invention is applied. As shown in FIG. 1, the bed system 1 has a determination device 10 for determining, for example, the position and posture of a user P on the bed 3 (mattress 5).

[0017] The determination device 10 may be configured to include a detection device 12 placed between the floor portion of the bed 3 and the mattress 5, and a processing device 14 for processing the value output from the detection device 12. Also, the determination device 10 may be constituted by a single detection device 12 having the function of the processing device 14, for example.

[0018] When the user P is in bed on the mattress 5, the detection device 12 detects body vibration (vibration emitted from the human body) as the biological signal of the user P. Then, based on the detected vibration, the posture and biological information value of the user P can be calculated. The detection device 12 may output and display the calculated biological information value as the biological information value of the user P. Here, the detection device 12 may calculate the activity amount of the user P and, in addition, calculate the heart rate and respiratory rate as the biological information value.

[0019] Furthermore, since the processing unit 14 can be a general-purpose device, it is not limited to information processing devices such as computers, but may be composed of devices such as tablets or smartphones.

[0020] Here, the detection device 12 is configured in a sheet-like shape to be thin. As a result, even when placed between the bed 3 and the mattress 5, it can be used without causing discomfort to the user P, and thus biometric information values, including activity levels in bed, can be measured over a long period of time. In other words, biometric information values, etc., are acquired as part of the user's condition when the user is lying down and at rest.

[0021] The detection device 12 only needs to be able to acquire the user P's biosignals (such as body movements, respiratory movements, and heart rate). In this embodiment, the user's sleeping position and posture are determined based on body vibrations, and activity level, heart rate, respiratory rate, etc., are calculated, but for example, an actuator with a strain gauge may be used. Alternatively, this may be implemented using a smartphone or tablet placed on the bed 3 (or mattress 5) by utilizing a built-in acceleration sensor, etc.

[0022] Furthermore, bed 3 can be installed in various locations. For example, bed 3 is typically installed in the hospital where user P is hospitalized or in the facility where they reside, but in the case of home care, bed 3 may also be installed in the home.

[0023] Furthermore, the determination device 10 can communicate with other devices via the network NW. Of the determination device 10, the detection device 12 may be connected to the network NW via, for example, the processing device 14, or the detection device 12 may be directly connected to the network NW via the access point 30 using a wireless LAN or the like. Alternatively, the detection device 12 may be directly connected to the network NW by incorporating a communication module that can communicate with, for example, a mobile communication network (LTE / 4G / 5G / 6G, etc.).

[0024] The network NW can be connected to, for example, a server device 40 and a terminal device 50. The server device 40 may store, for example, biological information values ​​acquired by the judgment device 10, or it may store the user's sleeping position and sleeping posture determined by the judgment device 10. The server device 40 may also be, for example, an electronic medical record server that stores the user's disease information.

[0025] The terminal device 50 may be, for example, an information processing device such as a smartphone, tablet, or laptop computer used by medical staff such as doctors and nurses. Alternatively, the terminal device 50 may be an information processing device used by facility staff or family members.

[0026] [2. Functional Configuration] Next, the functional configuration of the determination device 10 in the bed system 1 will be explained using Figures 2 and 3. In this embodiment, the determination device 10 includes a detection device 12 and a processing device 14, and each functional unit (process) except for the vibration data acquisition unit 400 may be implemented in either unit. In other words, by combining these devices, the determination device 10 functions.

[0027] [2.1 Hardware Configuration] As shown in Figure 2, the determination device 10 is configured to include, as necessary, one or more of the following: a control unit 100, a storage unit 200 (storage 210, ROM 220, and RAM 230), a vibration data acquisition unit 400, an input unit 600, an output unit 700, a notification unit 800, and a communication unit 900.

[0028] In the case of Figure 1, the control unit 100, vibration data acquisition unit 400, and storage unit 200 are provided in the detection device 12, while the other components may be provided in the processing device 14.

[0029] The control unit 100 controls the entire determination device 10. The control unit 100 realizes various functions by reading and executing various programs stored in the memory device, the storage unit 200 (for example, storage 210 or ROM 220). The control unit 100 may be implemented by one or more control devices / arithmetic units (CPU (Central Processing Unit), SoC (System on a Chip)). Alternatively, the control unit 100 may be composed of a control circuit.

[0030] The memory unit 200 stores various types of information as data. The memory unit 200 is generally a device that includes one or more storage 210, ROM 220, and RAM 230, and stores data in any of them as needed.

[0031] Storage 210 is a non-volatile storage device capable of storing programs and data. For example, it may consist of storage devices such as HDDs (Hard Disk Drives) or SSDs (Solid State Drives). Alternatively, Storage 210 may be configured as an externally connectable USB memory stick or memory card. Furthermore, Storage 210 may be, for example, a storage area located in the cloud.

[0032] ROM220 is a non-volatile memory that can retain programs and data even when the power is turned off.

[0033] RAM230 is the main memory primarily used by the control unit 100 during processing. RAM230 is a rewritable memory that temporarily holds data including programs read from storage 210 and ROM220, as well as execution results.

[0034] The vibration data acquisition unit 400 acquires vibration data. In this embodiment, as an example, body vibration is acquired using a sensor that detects pressure changes. The vibration data acquisition unit 400 may acquire the vibration data as, for example, numerical values ​​or as waveforms.

[0035] Here, an example of the vibration data acquisition unit 400 will be described with reference to Figure 3. The vibration data acquisition unit 400 is, for example, a vibration sensor that can detect the user's vibrations (body vibrations). At least two vibration sensors are provided; for example, in Figure 3, two (vibration sensor 410 and vibration sensor 420) are provided on the left and right sides of the detection device 12. The control unit 100 can determine the user's sleeping position and sleeping posture based on the vibration data acquired from the vibration data acquisition unit 400. The control unit 100 may also acquire and store the changes in the user's sleeping position and sleeping posture in chronological order.

[0036] Furthermore, the vibration sensors 410 and 420 are installed at a predetermined distance apart in the shorter direction of the bed 3. The distance between the two vibration sensors can be, for example, approximately the width of the user in the lateral direction, and preferably the distance between the two sensors is about 15 cm to 60 cm.

[0037] The vibration data acquisition unit 400 may also acquire vibration data from the user using load sensors provided on the bed 3. For example, a load cell may be provided on the frame of the bed 3, and vibration data may be acquired from the strain of the load cell. Alternatively, a load sensor may be provided on the actuator of the bed 3, and vibration data may be acquired from the load sensor of the actuator.

[0038] The input unit 600 is used by users or operators to input various conditions or to input commands to start processing. The input unit 600 may be implemented by any input means, such as a hardware key or a software key.

[0039] The output unit 700 outputs information such as the user's status (getting out of bed, being in bed), the user's sleeping position and posture while in bed, as well as biometric information values ​​such as heart rate and respiratory rate, and notifies of abnormalities. The output unit 700 may be a display device such as a display, or a notification device (sound output device) that notifies alarms, etc. It may also be an external storage device that stores data, or a transmission device that transmits data over a communication channel, etc. It may also be a communication device for notifying other devices.

[0040] Furthermore, the input unit 600 and output unit 700 may be implemented by other devices. For example, they may be implemented using a terminal device connected via the communication unit 900 (for example, a smartphone or tablet used by a user). In this case, the terminal device may be capable of executing a program that enables the control unit 100, which will be described later, to perform the processing.

[0041] The notification unit 800 provides notifications to users, etc. For example, the notification unit 800 may be a speaker that outputs sound or an LED that is a light-emitting device. The notification unit 800 may also provide notifications to other devices (for example, terminal devices such as the user's smartphone, or a nurse call system).

[0042] The communication unit 900 communicates with other devices. For example, if the device is nearby, the communication unit 900 provides communication using methods such as wireless LAN (or wired LAN) or Bluetooth®. The communication unit 900 may also be a device that provides short-range wireless communication such as NFC. The communication unit 900 may also provide communication using methods that enable mobile communication such as 4G / LTE / 5G / 6G. The communication unit 900 may also be an interface (e.g., USB) for communicating with other devices.

[0043] [2.2 Software Configuration] The software configuration will be explained with reference to Figure 4. For example, the control unit 100 implements each function by executing programs and applications stored in the memory unit 200 (e.g., storage 210, ROM 220, RAM 230).

[0044] The biometric information acquisition unit 110 acquires the biometric information of user P. The biometric information acquisition unit 110 calculates the user's activity level from vibration data, for example. The biometric information acquisition unit 110 may also detect body vibrations per sampling unit time from the vibration data acquisition unit 400 and calculate the activity level based on the number of detected body vibrations. Alternatively, the biometric information acquisition unit 110 may calculate the activity level from changes in the user's sleeping posture and movements.

[0045] In this embodiment, the biological information acquisition unit 110 may extract respiratory and heart rate components from the vibration data acquired from the vibration data acquisition unit 400 and determine the respiratory rate and heart rate based on the respiratory interval and heart rate interval. Alternatively, the biological information acquisition unit 110 may analyze the periodicity of body movement from the vibration data (e.g., using Fourier transform) and calculate the respiratory rate and heart rate from the peak frequency.

[0046] The sleep state determination unit 120 determines the user's sleep state. For example, the sleep state determination unit 120 determines the user's sleep state based on vibration data acquired by the vibration data acquisition unit 400. The sleep state determination unit 120 may determine two sleep states: "wakeful" and "sleeping." Furthermore, the sleep state determination unit 120 may further determine the "sleeping" state as "REM sleep" and "non-REM sleep," or it may further determine multiple levels (depth of sleep) of the "sleeping" state.

[0047] Furthermore, the sleep state determination unit 120 may determine whether a person is in a sleep state or a wakeful state based on the magnitude of their activity level and how the activity level changes over time. For example, the sleep state determination unit 120 does not need to determine that a person is in a wakeful state even if there is temporary body movement. The sleep state determination unit 120 may determine that a person is in a wakeful state if the user P's body movement continues for a certain period of time.

[0048] The user status determination unit 130 determines the user's status. In this embodiment, the user's status can be determined as the sleeping position, which is the position the user is in when lying down on the bed 3, and the sleeping posture, which is the posture the user is in when lying down. The user status determination unit 130 may determine either the user's sleeping position or the user's sleeping posture, or it may determine both.

[0049] Furthermore, the user status determination unit 130 may determine that the user's posture is one other than lying down. For example, the user status determination unit 130 may determine that the user is in a seated position on the edge of the bed. The user status determination unit 130 may also determine whether the user is out of bed or in bed.

[0050] The bed state acquisition unit 140 acquires the state of the bed. The bed state can include, for example, the angle of the back bottom and the angle of the knee bottom (foot bottom). By acquiring the bottom angles, the bed state acquisition unit 140 can determine, for example, whether the bottom of bed 3 is flat or whether the back (foot) is raised. The bed state acquisition unit 140 may also acquire the state of mattress 5. For example, if mattress 5 is an air mattress, the state of the air mattress (such as the pressure state or whether the rolling function is being performed) may be acquired.

[0051] Storage 210 reserves storage areas for biometric information storage area 202, reference model storage area 204, judgment data storage area 206, and user state data storage area 208.

[0052] The biological information storage area 202 may store information about biological information values ​​calculated by the control unit 100 from the acquired vibration data (biological signals), such as activity level. In addition, information about biological information values ​​such as respiratory rate and heart rate may also be stored. Furthermore, it is preferable that the biological information is stored in a time series at predetermined intervals.

[0053] The reference model storage area 204 is a trained model used by the user state determination unit 130 when determining the user's sleeping position and sleeping posture, and stores a reference model used when determining the user's sleeping position and sleeping posture. Preferably, the reference model is a trained model that the control unit 100 has trained using machine learning with parameters calculated based on vibration data as explanatory variables and the user's sleeping position and / or sleeping posture as the objective variable. A trained model is prepared for each user. Alternatively, the reference model may be a table that stores reference vibration data values ​​and information on sleeping posture. The reference model will be described in detail later.

[0054] The judgment data storage area 206 stores judgment data used by the user status determination unit 130 when determining the user's status. The judgment data storage area 206 contains values ​​calculated based on vibration data. The control unit 100 may store the judgment data in the judgment data storage area 206 in chronological order at all times, or it may store it as needed.

[0055] The user status data storage area 208 stores the user status determined by the user status determination unit 130. In this embodiment, the user status data storage area 208 stores the user's sleeping position and the user's sleeping posture as part of the user's status. The user status data storage area 208 may store either the user's sleeping position or the user's sleeping posture, or it may store both.

[0056] Furthermore, the user status data storage area 208 may also store other user status information, such as whether the user is out of bed or in bed. Additionally, the user status data storage area 208 may store data related to sleep (for example, whether the user is sleeping or awake) as part of the user's status. Furthermore, the user status data storage area 208 may also store user-related data (for example, the bottom state of bed 3) as part of the user's status.

[0057] [3. Processing Flow] The process by which the user status determination unit 130 determines the user's status in this embodiment will be explained with reference to the figures. Note that the following explanation describes processes (for example, the user status determination unit 130) that are appropriately executed by the functional units shown in Figure 4, but for the sake of explanation, it will be described as being executed by the control unit 100.

[0058] [3.1 Main Processing] Figure 5 is an operation flow illustrating the main processing in this embodiment. The control unit 100 selects the user occupying bed 3 (S102). For example, in a hospital, the staff selects the user when a user is admitted and begins using bed 3. Similarly, in a facility, the staff selects the user when a user moves in and begins using bed 3. In other words, the staff only needs to select the user when a user is using bed 3 for the first time or when a different user is using it than the previous time.

[0059] The control unit 100 determines whether a reference model corresponding to the selected user is already stored in the reference model storage area 204 (S104). If a reference model is not stored, the control unit 100 executes a reference model generation process to generate a reference model (S106). As a result, a reference model corresponding to the user is stored in the reference model storage area 204. The detailed operation of the reference model generation process will be described later.

[0060] Next, when a user is in bed (S108; Yes), the control unit 100 executes a sleeping position determination process (S110). Also, when a user is in bed (S108; Yes), the control unit 100 executes a sleeping posture determination process (S112). The control unit 100 may execute either the sleeping position determination process or the sleeping posture determination process, or it may execute both. Furthermore, the control unit 100 may execute the two processes in parallel, or it may execute the two processes alternately by switching between them at regular intervals.

[0061] Here, the determination of whether a user is in bed can be made, for example, by the control unit 100 when it receives vibration data exceeding a threshold from the vibration data acquisition unit 400. Alternatively, the control unit 100 may determine that a user is in bed when it detects a load exceeding a threshold on the bed 3 from a load sensor or the like. In addition, the control unit 100 may use any known method to determine whether or not a user is in bed.

[0062] The control unit 100 then performs the sleeping position determination process and the sleeping posture determination process until the user leaves the bed (S114; No). The control unit 100 also terminates this process when the user leaves the bed (S114; Yes). Alternatively, when the user leaves the bed, the control unit 100 may return to S108 and perform the sleeping position determination process and / or sleeping posture determination process again when the user is in bed.

[0063] [3.2 Baseline Model Generation Process] Figure 6 is a flowchart showing an example of the process for generating a reference model. The control unit 100 generates a trained model corresponding to the user selected in S102 by executing the reference model generation process. If a trained model has already been generated, the control unit 100 performs retraining based on newly acquired vibration data.

[0064] First, the user specifies their sleeping position (S202). The specified sleeping position becomes the so-called training data (dependent variable). The sleeping position specified by the user may be, for example, three locations such as "right edge," "center," and "left edge." Alternatively, the sleeping position specified by the user may be two locations such as "center" and "edge," or "center" and "outside the center."

[0065] Next, the control unit 100 determines whether it is also necessary to acquire the user's sleeping posture (S204). Here, in order to be able to determine the user's sleeping posture as well, when generating a trained model, the control unit 100 asks the user to indicate their sleeping posture (S204; Yes → S206). In other words, in addition to the user's sleeping position, the user's sleeping posture can also be used as training data (target variable).

[0066] Here, the sleeping position specified by the user may be one of four positions, such as "supine," "right lateral," "left lateral," or "prone." Alternatively, the sleeping position may simply be one of two positions, such as "supine" and "lateral," or "supine" and "prone."

[0067] Next, the control unit 100 acquires judgment data by executing a judgment data acquisition process (S208). The judgment data is output data that the control unit 100 has processed based on the vibration data of the occupant. The judgment data acquisition process will be explained below with reference to Figure 7.

[0068] The control unit 100 acquires vibration data (S252). Subsequently, the control unit 100 performs a Fourier transform on the control unit 100 (S254). Figure 8(a) is an example of a vibration waveform acquired from the vibration sensor 410 as vibration data. Figure 8(a) is a diagram showing the vibration data of one sensor (the vibration sensor 410 located on the left) as a waveform. In Figure 8(a), the vertical axis shows the input value and the horizontal axis shows the time transition.

[0069] Here, the control unit 100 performs a Fourier transform to convert the signal expressed in the time domain into the frequency domain. Figure 8(b) is a graph showing the state after the control unit 100 has converted the vibration data into a logarithmic amplitude spectrum. The vertical axis represents the value obtained by the Fourier transform of the input signal (for example, power expressed in logarithmic notation), and the horizontal axis represents the frequency. Although Figure 8 only explains the vibration data acquired from the vibration sensor 410 located on the left, the same processing is applied to the vibration data acquired from the vibration sensor 420 located on the right.

[0070] Next, the control unit 100 performs a low-frequency cut-off process because the low-frequency components have a large dispersion (S256). Here, it is preferable to cut low-frequency components below 0.5 Hz, or below 2 Hz, and more preferably below 1 Hz. Also, the control unit 100 does not necessarily have to cut the low-frequency components.

[0071] Furthermore, the control unit 100 performs mirroring processing on the data after low-frequency cut-off processing to combine the logarithmic amplitude spectra based on the vibration data from the left and right sides, and obtains a combined logarithmic amplitude spectrum (joint log amp).

[0072] Figure 9(a) shows the logarithmic amplitude spectrum based on the vibration sensor 410 located on the left. Figure 9(b) shows the logarithmic amplitude spectrum based on the vibration sensor 420 located on the right.

[0073] Here, for example, we cut out the region R102, which represents the low-frequency component, from the logarithmic amplitude spectrum in Figure 9(a). Also, since the data in the logarithmic amplitude spectrum of Figure 9(a) is folded back at the Nyquist frequency, we cut out the mirroring region R104.

[0074] Similarly, the region R106, which represents the low-frequency components, is cut from the logarithmic amplitude spectrum in Figure 9(b). Also, since the data in the logarithmic amplitude spectrum of Figure 9(b) is folded back at the Nyquist frequency, the region R108, which is the mirroring portion, is cut.

[0075] The control unit 100 then combines the remaining data. The combined logarithmic amplitude spectrum is shown in Figure 9(c). Note that in Figure 9(c), the vertical and horizontal axes are the same as in Figures 9(a) and (b), with the left side of Figure 9(c) being a combination of part of Figure 9(a) and the right side being a combination of part of Figure 9(b).

[0076] Next, the control unit 100 performs deconvolution on the combined logarithmic amplitude spectrum (S260). This allows the control unit 100 to deconvolve the combined logarithmic amplitude spectrum, reconstruct the user's motion component, and output it as judgment data.

[0077] For example, the control unit 100 derives an approximation curve approximated by a quartic function. Then, the control unit 100 deconvolves by subtracting only the approximation curve from the combined logarithmic amplitude spectrum, thereby restoring the user's body movement component. This is because the acquired vibration data (logarithmic amplitude spectrum) is affected by the mattress filter, and deconvolution makes it possible to exclude this effect from the logarithmic amplitude spectrum.

[0078] Figure 10(a) shows an example of the derived quartic function. Here, the graph in Figure 10(b) is obtained by subtracting the approximation curve in Figure 10(a) from the combined logarithmic amplitude spectrum in Figure 9(c). Figure 10(b) is a graph showing the deconvoluted judgment data (reconstructed motion component).

[0079] Thus, in step S208 of Figure 6, the control unit 100 outputs the judgment data by executing the judgment data acquisition process shown in Figure 7.

[0080] Next, the control unit 100 needs to acquire the sleeping position (for example, if the answer was Yes in S204), and if there is a sleeping position that has not yet been acquired, it changes the sleeping position and acquires the judgment data.

[0081] Furthermore, the control unit 100 determines whether there are any sleeping positions that have not been acquired if all sleeping positions have been acquired at the current sleeping position (S212; No → S214). If there are any sleeping positions for which judgment data has not yet been acquired, the sleeping position is changed and judgment data is acquired (S214; Yes → S202).

[0082] Furthermore, if the control unit 100 has acquired judgment data for all sleeping positions and postures, it generates a reference model, which is a trained model, using the specified sleeping position and posture as training data (target variable) and the judgment data as explanatory variables (S214; No → S216).

[0083] Here, the trained model used as the reference model is preferably a supervised learning-based machine learning model. While there are no particular restrictions on the supervised learning-based machine learning model, examples include Random Forest, Support Vector Machine (SVM), Light Gradient Boost Machine (LGBM), XGBoost, and neural networks.

[0084] Here, the control unit 100 performs the following specific actions as a method for learning the sleeping position (a method for acquiring judgment data):

[0085] (1) Learning based on sleeping position The control unit 100 notifies the user to lie down in the center position on bed 3. The user lies down in the center position on bed 3 (preferably in a supine position). Then, judgment data is acquired for a predetermined time (for example, 10 seconds).

[0086] Next, the control unit 100 notifies the user to lie down on bed 3 in the right-hand position (right end). The user lies down on bed 3 in the right-hand position (right end) (preferably in a supine position). Then, a predetermined time (for example, 10 seconds) is measured, and judgment data is acquired. Similarly, the control unit 100 notifies the user to lie down on bed 3 in the left-hand position (left end). The user lies down on bed 3 in the left-hand position (left end) (preferably in a supine position). Then, a predetermined time (for example, 10 seconds) is measured, and judgment data is acquired.

[0087] This allows the control unit 100 to learn based on the judgment data corresponding to the user's sleeping position, generate a basic model, or retrain it.

[0088] (2) Learning based on sleeping position The control unit 100 notifies the user to lie down in a supine position on bed 3. The user lies down in a supine position on bed 3 (preferably in the center position). Then, a predetermined time (for example, 10 seconds) is measured and judgment data is acquired.

[0089] Next, the control unit 100 notifies the user to lie face down in bed 3. The user lies face down in bed 3 (preferably in the center position). Then, a predetermined time (for example, 10 seconds) is measured and judgment data is acquired.

[0090] Next, the control unit 100 notifies the user to lie on their right side in bed 3. The user lies on their right side in bed 3 (preferably in the center position). Then, a predetermined time (for example, 10 seconds) is measured, and judgment data is acquired. Similarly, the control unit 100 notifies the user to lie on their left side in bed 3. The user lies on their left side in bed 3 (preferably in the center position). Then, a predetermined time (for example, 10 seconds) is measured, and judgment data is acquired.

[0091] This allows the control unit 100 to learn based on the judgment data corresponding to the user's sleeping posture, generate a basic model, or retrain itself.

[0092] In Figure 6, the sleeping position is instructed first, followed by the sleeping posture. However, the control unit 100 may skip the instruction regarding the sleeping position. That is, by not executing S202 and S214, the control unit 100 can learn based on judgment data corresponding only to the user's sleeping posture, generate a basic model, or retrain it.

[0093] (3) Learning based on sleeping position and posture Furthermore, the above-mentioned (1) and (2) may be combined. For example, for the user's sleeping position, data may be obtained to determine the sleeping posture as supine, right lateral, and left lateral, respectively, depending on whether the user is sleeping in the center, slightly to the right, or slightly to the left. This allows the system to learn based on the judgment data corresponding to the user's sleeping position and sleeping posture, generate a basic model, and retrain it.

[0094] (4) Enter the user's posture. (1) to (3) involve the control unit 100 notifying the user of their sleeping position and posture in order to generate a basic model. The system learns by having the user sleep in that position or assume that sleeping posture. Alternatively, the operator may input the user's initial sleeping position and posture when they first arrive in bed. For example, if the user is lying supine in the center of bed 3, the operator may input that they are lying supine in the center using a control device or the like. As a result, the control unit 100 learns the reference model using the acquired judgment data as training data, indicating that the user is lying in the center and in a supine position.

[0095] (5) Determined by the system Figure 6 shows instructions for the sleeping position (S202) and sleeping posture (S206), but for example, the control unit 100 may determine the sleeping position and sleeping posture. For example, when the reference model generation process is executed, the control unit 100 uses an imaging device to capture images of the user's condition. The control unit 100 then determines the user's sleeping position and sleeping posture from the captured images and uses these as training data, which are the target variables. Alternatively, the control unit 100 may acquire the user's body pressure distribution on the bed 3 (mattress 5). The control unit 100 may then determine the user's sleeping position and sleeping posture based on the acquired body pressure distribution and use these as training data, which are the target variables.

[0096] Here, the control unit 100 may obtain the user's body pressure distribution from sensors provided on the mattress, or from a sheet-shaped detection device 12 or other pressure sensors.

[0097] [3.3 Sleeping position determination process, sleeping posture determination process] By using a reference model, the control unit 100 can determine the user's sleeping position and / or sleeping posture.

[0098] For example, Figure 11 is a diagram illustrating the sleeping position determination process. The control unit 100 executes the process of acquiring determination data for users who are in bed (for example, Figure 7) (S302). Then, by inputting the user's determination data into the reference model, if the sleeping position can be determined (S304; Yes), the sleeping position is output (S306).

[0099] Figure 12 illustrates the process for determining the sleeping posture. The control unit 100 performs the process of acquiring determination data for the user in bed (for example, Figure 7) (S402). Then, by inputting the user's determination data into the reference model, if the sleeping posture can be determined (S404; Yes), the sleeping posture is output (S406).

[0100] [4. Effects] Here, we will explain the effects of applying this embodiment. Figure 13 is a table showing the accuracy of determining the sleeping position when a simple logarithmic amplitude spectrum is used as the judgment data as the reference model, and when data obtained by deconvolution processing the logarithmic amplitude spectrum described in this embodiment is used.

[0101] Here, the accuracy rate for a standard thick urethane mattress is shown as the first mattress, the accuracy rate for a thin urethane mattress is shown as the second mattress, and the accuracy rate for the Emma mattress is shown as the third mattress.

[0102] In this experiment, eight people were used on the first mattress, ten on the second mattress, and eight on the third mattress, and the experiment was conducted while the participants were awake. After each participant initially generated a basic model, they took four sleeping positions (supine, left lateral, right lateral, and prone) at each sleeping position, and 10 sets of test data were used for each position. The sleeping position determined by the sleeping position determination process was then compared with the actual sleeping position to calculate the accuracy rate.

[0103] It can be seen that the average accuracy rate for determining sleeping position using a basic model generated based on deconvolution-processed data is significantly higher than the average accuracy rate for determining sleeping position using a basic model generated using only logarithmic amplitude spectra, i.e., when only logarithmic amplitude spectra are used as the judgment data.

[0104] On average across the three datasets, the average accuracy rate for determining sleeping position using only logarithmic amplitude spectra as judgment data was 82.3%, while the average accuracy rate for determining sleeping position based on deconvolution-processed judgment data was 98.7%, demonstrating extremely accurate determination of sleeping position.

[0105] [5. Other Embodiments] The following describes other embodiments different from those described above. As a first alternative embodiment, after performing deconvolution processing in the determination data processing shown in Figure 7, a resampling process may be performed further. That is, as shown in Figure 14, the control unit 100 may perform a resampling process after performing deconvolution processing in S262 (S262).

[0106] Furthermore, when resampled data was used as the judgment data, the accuracy rate shown in Figure 14 improved even further. The accuracy rate for the first mattress remained the same at "96.1%", but the accuracy rates for the second and third mattresses were "100%".

[0107] In a second embodiment, as shown in Figure 5, the control unit 100 performs a sleeping position determination process and / or a sleeping posture determination process when the user is in bed. The control unit 100 may also perform a sleeping position determination process and / or a sleeping posture determination process under different conditions.

[0108] For example, the control unit 100 may perform a sleeping position determination process and / or a sleeping posture determination process when the following conditions are met. (1) While the user is asleep (from falling asleep to waking up) (2) When the bed 3 is in a flat or nearly flat position (for example, when the backrest angle is 0 degrees or less than 3 degrees) (3) While the user's heart rate and respiratory rate are different from normal. For example, while they are 20% or more higher than the user's average daily heart rate. (4) While the time period falls within the specified time period (for example, between 0:00 and 5:00) (5) While the user is receiving the prescribed treatment (for example, after being administered an anesthetic or after taking a drug that causes drowsiness)

[0109] The control unit 100 may combine one or more of the above-mentioned conditions. For example, the control unit 100 may determine the user's sleeping position only when the user is asleep and the time period falls within a predetermined time period.

[0110] In a third embodiment, a reference model with different criteria may be applied instead of a pre-trained model that has been trained using machine learning. For example, in the reference model generation process, the image of the deconvolution-processed waveform data described in Figure 10(c) is stored in the reference model as judgment data. When the control unit 100 determines the user's sleeping position and posture, it acquires the user's judgment data on the bed 3 as waveform data. The control unit 100 may then determine the user's sleeping position and / or posture by comparing (pattern matching) this with the waveform data stored in the reference model. Alternatively, the control unit 100 may determine the position and posture numerically instead of using waveform data. In this case, the control unit 100 can determine the user's sleeping position and posture without machine learning.

[0111] In a fourth embodiment, some or more functions may be implemented by a server device. For example, the reference model generation process described in Figure 6 may be executed by the server device, and the reference model may be stored in the server device. Also, the main process described in Figure 5, the sleeping position determination process described in Figure 11, and the sleeping posture determination process described in Figure 12 may be executed by the server device. In this way, for example, the determination device 10 can be implemented by providing only a simple sensor device and a communication device to realize the above-described embodiment. In this case, the determination results of the user's sleeping position and sleeping posture may be displayed on a terminal device of a staff member or the like.

[0112] In a fifth embodiment, an alert may be output based on the sleeping position determination process and the sleeping posture determination process. For example, if the control unit 100 determines, by performing the sleeping position determination process, that the user's sleeping position has moved towards the edge of the bed 3, it may send a bed exit notification to the staff's terminal device. Alternatively, if the control unit 100 determines, by performing the sleeping posture determination process, that the user has not changed their sleeping posture for a predetermined time (for example, not turning over), it may send a notification for pressure ulcer prevention to the staff's terminal device.

[0113] In the sixth embodiment, the judgment data acquisition process may be performed in a different order as appropriate. For example, in the embodiment described above, the control unit 100 performed a mirroring process to obtain a combined logarithmic amplitude spectrum, and then performed a deconvolution process. For example, the control unit 100 may perform a deconvolution process on the logarithmic amplitude spectrum, then perform a mirroring process, and then obtain a combined logarithmic amplitude spectrum. That is, the control unit 100 may execute S260 first, and then execute S258.

[0114] [6. Variant] This disclosure is not limited to the embodiments described above, and various modifications are possible. In other words, embodiments obtained by combining technical means that are appropriately modified within the scope of this disclosure are also included in the technical scope.

[0115] Furthermore, we intend to acquire rights to any of the technologies described in the specification through amendments or divisional applications, etc.

[0116] Furthermore, in each embodiment, the program that operates in each device is a program that controls the CPU and other components (a program that makes the computer function) in order to realize the functions of the embodiments described above. The information handled by these devices is temporarily stored in a temporary storage device (for example, RAM) during processing, and then stored in various ROMs or HDDs, and read, modified, and written by the CPU as needed.

[0117] Here, the recording medium for storing the program may be any of the following: semiconductor media (e.g., ROM or non-volatile memory card), optical recording medium or magneto-optical recording medium (e.g., DVD (Digital Versatile Disc), CD (Compact Disc), BD (Blu-ray® Disc)), magnetic recording medium (e.g., magnetic tape, flexible disk), etc.

[0118] Furthermore, when distributing the program to the market, it can be stored on a portable recording medium and distributed, or transferred to a server computer connected via a network such as the Internet. In this case, the storage device of the server device is, of course, also included in this disclosure.

[0119] Furthermore, the data mentioned above may not be stored within the device itself, but rather stored on an external device and retrieved as needed. For example, the data may be stored on a NAS (Network Attached Storage) or on the cloud.

[0120] Furthermore, the scope of this disclosure is not limited to the configurations explicitly described in the specification, but also includes combinations of the technologies disclosed herein. While the configurations for which patent protection is sought are described in the attached claims, there is no intention to exclude them from the technical scope simply because they are not described in the claims.

[0121] Furthermore, the phrases "in the case of..." and "when..." in the above-mentioned specification are explained as examples only, and do not represent a configuration limited to those described. Even for configurations other than those described, we disclose information that would be obvious to a person skilled in the art, and we intend to acquire rights to such information.

[0122] Furthermore, the descriptions of the processes and data flows described in the specification are not limited to the order in which they are described. For example, configurations in which parts of the process are deleted or the order is rearranged are also disclosed, and the company intends to acquire rights to them.

[0123] Furthermore, although the functions described in the embodiments are explained as being performed by each device, they may also be implemented by a single device or by utilizing an external server.

[0124] Furthermore, each functional block or feature of the apparatus used in the embodiments described above may be implemented or executed by an electrical circuit, such as an integrated circuit or a plurality of integrated circuits. An electrical circuit designed to perform the functions described herein may include a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic device, discrete gates or transistor logic, discrete hardware components, or a combination thereof. The general-purpose processor may be a microprocessor, a conventional processor, controller, microcontroller, or state machine. The aforementioned electrical circuit may consist of digital circuits or analog circuits. Also, if advances in semiconductor technology lead to the emergence of integrated circuit technologies that replace current integrated circuits, one or more aspects of this disclosure may use new integrated circuits based on such technologies.

[0125] Furthermore, in this embodiment, the processing unit 14 outputs biological information based on the results output by the detection device 12, but the detection device 12 may calculate everything itself. Also, this can be implemented not only by installing an application on a terminal device (e.g., a smartphone, tablet, or computer), but also by, for example, processing on the server side and returning the processing results to the terminal device.

[0126] For example, the detection device 12 may upload biometric information to the server, thereby enabling the server to perform the processing described above. This detection device 12 may be implemented as a device such as a smartphone with a built-in acceleration sensor and vibration sensor. [Explanation of Symbols]

[0127] 1 System 10 Judgment device 12 Detection device 14 Processing Unit 100 Control Unit 200 Storage section 210 storage 220 ROM 230 RAM 400 Vibration data acquisition unit 600 Input Section 700 Output section 800 News Department 900 Communications Department 3 beds 5 Mattress 40 Server Devices 50 Terminal devices

Claims

1. A determination device comprising an acquisition unit for acquiring vibrations of a user on a bed and a control unit, An acquisition unit that acquires vibration data from the aforementioned acquisition unit, A calculation unit that calculates judgment data by deconvolving the aforementioned vibration data, A determination unit determines the sleeping position or sleeping posture of a user on a bed by inputting the determination data calculated from the vibration data of the user on the bed into a trained model that is generated using the aforementioned determination data as input data and the user's sleeping position or sleeping posture as output. A determination device equipped with this device.

2. The calculation unit performs deconvolution on the vibration data, then resamples it to calculate the determination data. The determination device according to claim 1.

3. The determination device according to claim 1, wherein the calculation unit converts the vibration data into a logarithmic amplitude spectrum by performing a Fourier transform.

4. On the bed, two sensor devices are installed on the left and right sides. The determination device according to claim 1, wherein the acquisition unit acquires the vibration data from the two sensor devices provided on the left and right sides.

5. The determination device according to claim 4, wherein the calculation unit combines vibration data acquired from the two sensor devices provided on the left and right sides into a single data set and then performs deconvolution.

6. A computer capable of communicating with a determination device that includes an acquisition unit for acquiring vibrations of a user on a bed and a control unit, The acquisition function acquires vibration data from the aforementioned acquisition unit, A calculation function that calculates judgment data by deconvolving the aforementioned vibration data, A determination function that determines the user's sleeping position or posture on the bed by inputting the aforementioned determination data as input data and the user's sleeping position or posture as output to a trained model, and inputting the determination data calculated from the user's vibration data on the bed, A program that achieves this.

Images