Method of individual sleep inducing control based on vital signs and system therefor

The biosignal-based system addresses the lack of customization in conventional sleep technologies by using sensors to identify users and adapt content to their sleep stages and exercise needs, enhancing sleep quality and exercise management.

KR103045718B1Active Publication Date: 2026-11-10KOREA ELECTRONICS TECH INST
View PDF 11 Cites 0 Cited by

Patent Information

Application Number
KR1020240120131
Authority / Receiving Office
KR · KR
Patent Type
Patents
Current Assignee / Owner
Filing Date
2024-09-04
Publication Date
2026-11-10
Estimated Expiration
2044-09-04

AI Technical Summary

Technical Problem

Conventional sleep management technologies lack customization for individuals, requiring manual input of activity types and are uncomfortable to wear during sleep and exercise, failing to distinguish between users and provide tailored content.

Method used

A biosignal-based system using sensors to measure ballistocardiogram signals for user identification, classifying activities into sleep or exercise, and providing customized content through acoustic and video signals, with noise cancellation and sleep stage monitoring, and personalized exercise programs.

Benefits of technology

Provides comfortable, personalized sleep induction and exercise management by automatically recognizing users and adapting content to their sleep stages and exercise needs, improving sleep quality and inducing a refreshing wake-up.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 112024097270241-PAT00001_ABST
    Figure 112024097270241-PAT00001_ABST
Patent Text Reader

Abstract

A biosignal-based individual sleep induction management system is disclosed. The biosignal-based individual sleep induction management system according to the present invention comprises a sensor unit that measures biosignals in real time, an output unit that outputs at least one of an acoustic signal and a video signal, and a processor that identifies a user based on the measured biosignals, classifies the user's activity into one of a group including exercise and sleep, classifies a sleep stage if classified as sleep activity, and provides sleep care content through the output unit based on the identified user information and the classified sleep stage.
Need to check novelty before this filing date? Find Prior Art

Description

Technology Field

[0001] The present invention relates to a biosignal-based individual sleep induction management method and such system, and more specifically, to an individual sleep induction management method and system capable of providing customized content to a recognized user by measuring biosignals through a sensor. Background Technology

[0002] Numerous monitoring and sleep analysis technologies have been proposed to improve users' sleep quality. In most cases, these technologies provide consistent content based on monitored sleep stages derived from statistical criteria. Furthermore, monitoring devices provide consistent services without distinguishing between individuals.

[0003] Users must directly input their activity status so that wearable devices, such as smartwatches, provide content suitable for exercise or sleep.

[0004] Wearable devices such as smartwatches must be worn continuously while in close contact with the body during activities to provide measurements and related services.

[0005] Conventional sleep management technologies as described above have not been provided in a customized manner for each individual, involve the inconvenience of having to manually input the types of activities, and are uncomfortable to wear on the body during sleep and exercise. Prior art literature

[65535] Published Patent Application No. 10-2019-0049442 (May 9, 2019): 'Auditory stimulation-based sleep induction system and method thereof' The problem to be solved

[0006] Thus, the present invention was conceived in recognition of the above-mentioned problem, and the objective of the present invention is to provide a biosignal-based individual sleep induction management system and a biosignal-based individual sleep induction management method capable of identifying an individual using a non-restraining sensor, automatically recognizing activities without separate input, and providing customized content for sleep induction management optimized for each individual. means of solving the problem

[0007] A biosignal-based individual sleep induction management system according to the present invention for achieving the above objectives comprises a sensor unit for measuring biosignals in real time, an output unit for outputting at least one of an acoustic signal and a video signal, and a processor that identifies a user based on the measured biosignals, classifies the user's activity into a group including exercise and sleep, classifies a sleep stage if classified as sleep activity, and provides sleep care content through the output unit based on the identified user information and the classified sleep stage.

[0008] Here, the sensor unit includes a plurality of sensors arranged inside the sleeping mat, and the processor measures the user's ballistocardiogram signal (BCG) through the sensor unit and can identify the user based on the characteristic points of the measured ballistocardiogram signal.

[0009] Here, the processor can determine the posture of the user of the sleeping mat from the measured ballistocardiogram signal, classify the user's activities according to the posture, and classify the sleep stage based on the ballistocardiogram signal.

[0010] Here, the sensor unit measures ambient noise, and the output unit outputs opposite noise to cancel out the noise, and the processor can control the output unit to output the opposite noise having an inverse phase noise pattern opposite to the measured noise pattern, by comparing the noise measured through the sensor unit with a threshold point during the sleep activity and, if the measured noise is greater than or equal to the threshold point, determining the pattern of the measured noise.

[0011] Meanwhile, the biosignal-based individual sleep induction management system further includes an airable device that determines the sleep stage by measuring the user's breathing sounds during sleep, or a wearable device that is attached to the user's body to determine the sleep stage by measuring brain waves or photocircumference, and the processor can use the sleep stage determined by the airable device or the wearable device as label data to label the biosignal data measured by the sensor unit, and use the labeled data as training data to extract a deep learning-based learning model.

[0012] Meanwhile, the processor updates the sleep care content according to the change in the sleep stage, and if the sleep stage is classified as a sleep stage deeper than a preset standard, it can disable the output of the sleep care content.

[0013] Here, the processor monitors the sleep stage from 30 minutes before the set wake-up time when the wake-up time is set, initiates a wake-up notification function that induces the user into the REM sleep stage upon waking, and if the user does not reach the REM sleep stage by the wake-up time, provides content of binaural beats that induce a change in the sleep stage, and when the wake-up time arrives, provides a wake-up environment by controlling light and an alarm through the output unit, and when the user's re-sleep is detected from the biosignal measured in real time, the wake-up notification function can be resumed.

[0014] Meanwhile, the processor can establish an agent that performs reinforcement learning by monitoring the sleep quality of a user who receives the sleep care content through the sensor unit, creating clusters based on at least one characteristic among the improvement or decline of the monitored sleep quality, the playback time of the content provided according to the sleep stage, and the change in total sleep time, and providing rewards that improve the characteristics belonging to each cluster according to which content is provided by the output unit.

[0015] Here, the processor classifies multiple users who have experienced the content into promoters, passives, and detractors based on their user-specific recommendation scores for the provided content, and can evaluate the reinforcement-learned agent through the value obtained by subtracting the detractor ratio from the promoter ratio (NPS).

[0016] Meanwhile, if the user's activity is classified as exercise activity, the processor may provide exercise content in which at least one of the rest time and exercise intensity changes based on the identified user information and the measured biosignal.

[0017] Meanwhile, a biosignal-based individual sleep induction management method according to one embodiment of the present invention includes the steps of: measuring a biosignal in real time; identifying a user based on the measured biosignal; classifying the user's activity into one of a group including exercise and sleep; classifying a sleep stage if classified as sleep activity; and providing sleep care content based on the identified user information and the classified sleep stage.

[0018] In addition, according to one embodiment of the present invention, a computer-readable recording medium is provided that records a program for executing the biosignal-based individual sleep induction management method on a computer. Effects of the invention

[0019] The biosignal-based individual sleep induction management system and biosignal-based individual sleep induction management method according to the present invention can help the user get a good night's sleep by providing sleep induction content tailored to the individual that is automatically recognized in a comfortable sleep environment without restraint, and improve the quality of sleep by inducing a refreshing feeling upon waking, and can provide a personalized exercise program by automatically identifying the user's exercise activity. Brief explanation of the drawing

[0020] FIG. 1 is a block diagram showing a biosignal-based individual sleep induction management system according to one embodiment of the present invention; FIG. 2 is a block diagram showing the detailed configuration of a processor of a biosignal-based individual sleep induction management system according to an embodiment of the present invention; FIG. 3 is a block diagram conceptually showing a software module stored in the storage unit of the processor of FIG. 2; FIG. 4 is a diagram showing a sensor unit of a biosignal-based individual sleep induction management system according to an embodiment of the present invention; FIG. 5 is a schematic diagram illustrating embodiments of the sensor unit, processor, and output unit of a biosignal-based individual sleep induction management system; FIG. 6 is a diagram illustrating the configuration of an embodiment for managing exercise activity of a biosignal-based individual sleep induction management system; FIG. 7 is a diagram illustrating a method for reinforcement learning of recommended content for sleep induction according to an embodiment of the present invention; FIG. 8 is a flowchart illustrating a biosignal-based individual sleep induction management method according to an embodiment of the present invention; and, FIG. 9 is a flowchart illustrating a specific embodiment of a biosignal-based individual sleep induction management method according to an embodiment of the present invention. Specific details for implementing the invention

[0021] The present invention will be described in more detail below with reference to the drawings. Furthermore, in describing the present invention, detailed descriptions of related known functions or configurations are omitted if it is determined that such detailed descriptions would unnecessarily obscure the essence of the invention. Additionally, the terms described below are defined considering their functions in the present invention, and these may vary depending on the intentions or relationships of the user or operator. Therefore, their definitions should be based on the content throughout this specification.

[0023] FIG. 1 is a block diagram showing a biosignal-based individual sleep induction management system according to one embodiment of the present invention.

[0024] Referring to FIG. 1, a biosignal-based individual sleep induction management system (1000) according to one embodiment of the present invention includes a sensor unit (100), a processor (200), and an output unit (300).

[0025] The sensor unit (100) measures the biosignal in real time. Specifically, the sensor unit (100) may be composed of a sensor capable of measuring the user's biosignal or may include a sensor.

[0026] The sensor unit (100) may include a plurality of sensors. The plurality of sensors may measure different types of biosignals or measure the same type of biosignal at different locations.

[0027] The sensor unit (100) can transmit the measured biosignal to the processor (200) in the form of an analog signal or as digitized biosignal data according to the communication method.

[0028] In one embodiment, the biosignal measured by the sensor unit (100) is a ballistocardiogram signal (BCG).

[0029] The ballistics signal is a signal that measures the vibration (ballistics) resulting from changes in blood flow in the heart and blood vessels associated with the moment blood ejected from the ventricles during the cardiac cycle passes through the aorta and transmits a rebound to our body.

[0030] As mentioned above, ballistocardiography is known to serve as an indicator of the heart's activity status similar to electrocardiography, containing information regarding cardiac output, regurgitation due to myocardial dysfunction, and abnormal blood flow phenomena. Therefore, this biosignal has the potential for clinical application, such as evaluating cardiac function, diagnosing heart diseases (including myocardial disorders), verifying treatment effects, and observing the degree of recovery. Ballistocardiographic signals can be measured using accelerometers, load cells, PDVF film sensors, and EMFi sensors, in addition to piezoelectric sensors. Since these sensors eliminate the need to attach electrodes to the body, signals can be measured in an unrestrained or unconscious state, making them useful for long-term health monitoring or during daily life.

[0031] The processor (200) controls various components of the system (1000). Specifically, the processor (200) can control the output unit (300) based on information received from the sensor unit (100). The processor (200) can identify a user based on the measured biosignals of the sensor unit (100). And, the processor (200) can classify the activity of the identified user into one of the groups including exercise and sleep.

[0032] First, biosignals have unique characteristics for each individual. In particular, the feature points of the biosignals differ to the extent that they can distinguish individuals. The feature points may be patterns analyzed from the biosignals. For example, the biosignals of a user being measured include ballistocardiogram (BCG) signals, heart rate (BPM) signals, electrocardiogram (ECG) signals, heart rate variability (HRV), and photoplethysmography (PPG) signals. From these collected biosignals, biosignal patterns unique to the user can be analyzed. The user's biosignal patterns include repeating cycles, signal magnitude or average energy per unit time, maximum / minimum amplitudes, etc. Additionally, signal processing can be performed based on the periodic signal forms of the biosignals. For example, the processor (200) can compute the Fourier transform of the biosignals, correlation analysis, convolution for autocorrelation analysis, etc.

[0033] Feature points of a user's pre-analyzed biosignals or unique user profile information based thereon can be stored and kept in a repository as user personal information and used to identify and authenticate the user.

[0034] The processor (200) determines the user's activities. The user's activities are classified into several types according to criteria. According to the present invention, the user's activities are classified into at least sleep activities and exercise activities. Additionally, daily or static activities, eating activities, etc., may be added according to criteria. Sleep activities can be viewed as activities from the preparatory actions for going to bed until fully waking up. Exercise activities can be viewed as activities that apply an exercise functional load to the body. Biosignals show different patterns depending on whether it is sleep activity or exercise activity. Generally, biosignals show slower and longer cycles during sleep activities than during daily activities, and faster and shorter cycles during exercise activities.

[0035] The processor (200) classifies the sleep stage when classified as a sleep activity.

[0036] Here, sleep stages are classified into Wake, REM, and NREM, and NREM can be further subdivided into Stage 1, Stage 2, Stage 3, and Stage 4. The definition of sleep stages is based on Rechtschaffen and Kales. Wakefulness (wake, or the stage of being fully awake, known as wakefulness), REM (REM sleep stage), and NREM (non-REM sleep stage) are distinguished by muscle tone and eye movements, along with waveforms appearing on an electroencephalogram (EEG). Eye movements can be observed via EOG.

[0037] In the fully awake stage (W), human brainwaves show alpha waves (8-13 Hz) and beta waves (13-35 Hz). In the sleep onset stage and light non-REM sleep stage 1 (NREM Stage 1 or N1), theta waves (4-7 Hz) are measured and slow eye movements are observed. In non-REM sleep stage 2 (NREM Stage 2 or N2), spindle waves (12-14 Hz) and K-complex waves are observed. In non-REM sleep stage 3 (NREM Stage 3 or N3), delta waves (2-4 Hz, 75 mv) are observed in the 20-50% (epoch) range, and muscles are in a relaxed state. In non-REM sleep stage 4 (NREM Stage 4 or N4), delta waves account for more than 50% (epoch). In the REM sleep stage, beta and theta waves are measured and rapid eye movements are observed.

[0038] The above sleep stages cycle. After entering the first N1 stage, the REM sleep stage repeats periodically at intervals of approximately 90 minutes ± 20 minutes. REM sleep is nearly equivalent to the wake state.

[0039] In the present invention, the deep sleep stage is defined as stages 3 and 4 of non-REM sleep.

[0040] The processor (200) provides sleep care content through the output unit (300) based on identified user information and classified sleep stages. Sleep care content refers to any control for managing the user's sleep. Sleep care content may be an expression or operation of various information, such as video, sound, light, or movement of equipment.

[0041] A detailed description of the operation and function of the processor (200) will be given later with reference to FIGS. 2 and FIGS. 3.

[0042] The output unit (300) outputs sleep care content under the control of the processor (200). The output unit (300) can be configured in various forms and can be various devices that can particularly aid in sleep. The output unit (300) can be a display, speaker, lighting, air conditioner, electric curtain, motion pillow, motion bed, etc.

[0044] FIG. 2 is a block diagram showing the detailed configuration of a processor of a biosignal-based individual sleep induction management system according to one embodiment of the present invention.

[0045] Referring to FIG. 2, the processor (200) includes a communication unit (210), a processing unit (220), an input unit (230), an output unit (240), a storage unit (250), and a display unit (260).

[0046] The communication unit (210) communicates with one or more other units or devices. The communication unit (210) can receive a measured biosignal from the sensor unit (100) and transmit content to the output unit (300).

[0047] The processing unit (220) controls the components of the processor (200) overall.

[0048] The processing unit (220) includes RAM (221), ROM (222), main CPU (223), graphics processing unit (224), first to n interfaces (225-1 to 225-n), and a bus (226).

[0049] RAM (221), ROM (222), main CPU (223), graphics processing unit (224), first to n interfaces (225-1 to 225-n), etc. can be connected to each other via a bus (226).

[0050] The first to n interfaces (225-1 to 225-n) are connected to various components (210, 230, 240, 250, 260). One of the interfaces may be a network interface connected to an external device through a network.

[0051] The main CPU (223) accesses the storage unit (250) and performs booting using the O / S stored in the storage unit (250). Then, it performs various operations using various programs, content, data, etc. stored in the storage unit (250).

[0052] In particular, the main CPU (223) can perform operations for biosignal processing, user recognition, activity classification, sleep stage classification, and selection and provision of sleep care content.

[0053] A set of instructions for booting the system is stored in the ROM (222). When a turn-on command is input and power is supplied, the main CPU (223) copies the O / S stored in the storage unit (250) to the RAM (221) according to the instructions stored in the ROM (222), and executes the O / S to boot the system. When booting is complete, the main CPU (223) copies various application programs stored in the storage unit (250) to the RAM (221), and executes the application programs copied to the RAM (221) to perform various operations.

[0054] The graphics processing unit (224) generates a screen containing various objects such as icons, images, and text using a calculation unit (not shown) and a rendering unit (not shown). The calculation unit (not shown) calculates attribute values ​​such as coordinate values, shape, size, and color for each object to be displayed according to the layout of the screen based on a received control command. The rendering unit (not shown) generates a screen of various layouts containing objects based on the attribute values ​​calculated by the calculation unit (not shown).

[0055] In particular, the graphics processing unit (224) can implement objects generated by the main CPU (223) into a GUI (Graphic User Interface), icon, user interface screen, etc.

[0056] The display unit (260) can display a screen generated by the graphics processing unit (224).

[0057] The input unit (230) receives input from the user. The input unit (230) can receive user information or dedicated account input. Additionally, the input unit (230) can receive input such as the user's desired bedtime, wake-up time, preferred sleep care content, unwanted sleep care content, and sleep environment control on / off settings.

[0058] The output unit (240) can output any information or state. The output unit (140) can output sound, light, vibration, etc., indicating a power state, sleep induction operation state, wake-up induction operation state, etc.

[0059] The storage unit (250) stores various data such as firmware or O / S (Operating System) software modules for the operation of the processor (200) and various multimedia content.

[0060] In particular, the storage unit (250) stores modules for various functions of the processor (200).

[0062] FIG. 3 is a block diagram conceptually showing a software module stored in the storage unit of the processor of FIG. 2.

[0063] Referring to FIG. 3, the storage unit (250) stores a signal processing module (251), a user recognition module (252), a user activity classification module (253), a sleep stage classification module (254), a sleep induction module (255), a sleep induction content learning module (256), a sleep environment control module (257), a wake-up notification module (258), and an exercise management module (259).

[0064] The signal processing module (251) includes an operation for processing a biosignal measured by the sensor unit (100), which is received through the communication unit (210) or input through the input unit (230). The biosignal processing may include a Fourier transform for frequency analysis capable of extracting feature points for user recognition, a convolution operation for autocorrelation analysis, a filtering operation for noise removal, a signal amplification operation, etc.

[0065] In particular, in one embodiment of the present invention, the sensor unit (100) includes a plurality of sensors (110), and some of the signals measured by each sensor (110) can be processed to include valid biological signals.

[0066] FIG. 4 is a diagram showing the sensor part of a biosignal-based individual sleep induction management system according to one embodiment of the present invention.

[0067] Referring to FIG. 4(a), the sensor unit (100) is a mat containing a plurality of sensors (110). Referring to FIG. 4(c), the sensor (110) is a multi-sensor containing two types of sensors. The sensor (110) includes a piezoelectric sensor (113) and a Force Sensitive Resistor (FSR) sensor (112). The piezoelectric sensor (113) is sensitive to changes in relative pressure and can measure ballistocardiogram signals. The FSR sensor (112) can measure absolute pressure. The measurement signals of the FSR sensor (112) and the piezoelectric sensor (113) included in the multi-sensor (110) are transmitted to an independent channel by being connected via a communication port (114).

[0068] The signal processing module (251) enables the identification of the location of the sensor (110) received in each channel, and also enables the processing of the signal of the sensor (110) into a valid biosignal only for the signal of the channel in which a certain pressure or higher due to the user's body weight is measured by the FSR sensor (112). The signal processing module (251) enables digital bandpass filtering, J-peak recognition of the ballistocardiogram signal, etc. The signal processing module (251) enables the calculation of heart rate variability (HRV) from secondary information processed from the ballistocardiogram signal, such as the interval time between J-peaks.

[0069] In another embodiment, the signal processing module (251) can store a data set for a certain time interval for a high-quality signal having a sufficiently large SNR, and estimate heart rate and respiration through digital signal processing and frequency analysis to determine the user's apnea sleep state during sleep.

[0070] The user recognition module (252) enables the recognition of a user based on a biosignal. To this end, the user recognition module (252) of the storage unit (250) has profile information containing feature points of the user's biosignal or information that can identify multiple users including feature points stored in advance. The user recognition module (252) can compare feature points or profiles and determine that a user is the currently active user if they show a match rate above a certain confidence level, even if they do not match exactly. The user recognition module (252) may include an AI module that learns the biosignal or its feature points and classifies a specific user. The AI ​​module may be updated to identify the user by learning data accumulated a certain number of times or over a certain period for the initially authenticated user. The AI ​​module may be a deep learning-based learning model such as a CNN.

[0071] The user activity classification module (253) enables classification of what the user's activity is. The user activity may be one of several groups including various activities. The groups include at least sleep activities and exercise activities. Sleep activities and exercise activities can be distinguished based on biosignals measured by the sensor unit (100). Furthermore, the user activity classification module (253) can determine the user's posture. The sensor unit (100) according to the embodiment of FIG. 4 includes a plurality of sensors (110), and a plurality of FSR sensors (112) enable distinguishing locations on the mat where pressure is applied by the user's body. From this, the user's posture can be determined. The user's posture may be a lying position, a sitting position, a standing position with only two feet, a prone position with palms touching, etc. At least the lying position allows for more accurate determination of sleep activities, and other postures help determine what kind of exercise is performed during exercise.

[0072] The sleep stage classification module (254) enables the classification of a user's sleep stage during sleep activity. Sleep stages can be classified based on ballistocardiogram signals or other biosignals. Sleep stage classes such as Wake, REM, NREM Stage 1, 2, 3, and 4 can be classified by learning biosignals through an AI model. The training data may be biosignals such as raw signals from sensors or signal-processed ballistocardiograms, or biosignals such as processed heart rate variability. Data labeling for training may be performed by other commercial equipment capable of classifying sleep stages. The commercial equipment may be an airable device that determines the sleep stage by measuring the user's breathing sounds during sleep, or a wearable device that is attached to the user's body to measure brainwaves or photocircumference to determine the sleep stage. The training model may be an RNN-based LSTM (long-short term memory) model. For example, an AI model that learns ballistocardiogram signals by receiving guidance on the sleep stages of a wearable device can classify sleep stages from a specific user's ballistocardiogram signals.

[0073] The sleep induction module (255) enables the user to fall asleep during sleep activity by inducing sleep. The sleep induction module (255) changes or updates the sleep care content to be output according to the sleep stage. The sleep induction module (255) may provide speaker control that outputs white noise sound during the sleep activity initiation stage. The sleep induction module (255) may provide control that gradually reduces the sound as the sleep stage progresses and disables the output of content when entering the deep sleep stage.

[0074] According to the present invention, binaural beats for inducing sleep in a user are provided. Specifically, the sleep induction module (255) can provide content in the form of binaural beats.

[0075] According to one embodiment, the sleep induction module (255) may measure the user's NREM time ratio and REM time ratio to create brainwaves optimized for sleep and provide brainwave-inducing binaural beats according to a threshold. Specifically, if the ratio of REM time to NREM time is relatively higher than a preset value, REM brainwave-inducing binaural beats are provided at the output unit (300), and if the ratio of REM time to NREM time is relatively lower than a preset value, NREM brainwave-inducing binaural beats are provided at the output unit (300).

[0076] Binaural beats are specific sounds that regulate brainwaves and can be implemented by playing sounds with different frequencies to both ears of a subject. For example, if a sound of 300 Hz (e.g., carrier wave) is played to one ear and a sound of 310 Hz (sleep wave + carrier wave) is played to the other ear through the output unit (300), the brain receives it as a 10 Hz wave, and the brainwaves can be regulated to 10 Hz.

[0077] In one example, the sleep induction module (255) measures the user's NREM time ratio and REM time ratio, and if the ratio of REM time to NREM time is relatively higher than a preset value, it outputs a REM brainwave-inducing binaural beat (e.g., a binaural beat of 3.5 to 7 Hz), and conversely, if the ratio of REM time to NREM time is relatively lower than a preset value (e.g., 4:1), it outputs an NREM brainwave-inducing binaural beat (a binaural beat of 3.5 Hz).

[0078] The sleep-inducing content learning module (256) is intended to provide sleep-inducing content that is more suitable for the user. The sleep-inducing content provided by default in the initial state of the system (1000) may not be helpful or satisfactory to a specific user. The sleep-inducing content learning module (256) enables the analysis of sleep care effects while continuously monitoring the user's sleep quality.

[0079] Sleep quality can be evaluated by regular sleep duration, time to enter the deep sleep stage, total sleep duration, number or frequency of irregular awakenings during sleep, blood oxygen saturation during sleep, etc.

[0080] The selection of appropriate sleep content can be performed by a reinforcement learning AI model. Sleep care content provided during the REM sleep stage and NREM sleep stage (specifically, Stage 1 and Stage 2) can be learned by the AI ​​model based on at least one characteristic among several evaluation indicators, such as content usage time, improvement or decline in sleep quality, and change in total sleep time. For reinforcement learning, clusters are created by grouping at least one characteristic of the content provided for each sleep stage. Then, a reward system and an agent can be established for each cluster to perform reinforcement learning.

[0081] Referring to FIG. 7, reinforcement learning is a method in which an agent or system learns to converge to an optimal state by providing positive or negative rewards when it takes an action corresponding to a state of the environment or an observation of the environment. The environment may be the aforementioned characteristics, such as a user's sleep state, biosignals, and sleep quality; the agent may be a biosignal-based individual sleep induction management system (1000) or processor (200) including an AI module that is a sleep induction management service provider; the action is sleep care content provided; the state is information being monitored; and the reward may be given positively if the user's sleep quality improves, or negatively if the sleep quality deteriorates. Thus, when the reinforcement-learned agent or system (1000) identifies a user from biosignals, it becomes possible to provide the best sleep care content for that user.

[0082] The sleep-inducing content learning module (256) can perform learning that incorporates feedback from multiple users. Specifically, the sleep-inducing content learning module (256) can perform respondent classification to calculate the Net Promoter Score (NPS). Through the sleep-inducing content learning module (256), the processor (200) classifies multiple users who have experienced the content into promoters, passive users, and detractors based on a user-specific recommendation score representing the user's satisfaction with the provided content, and evaluates the reinforcement-learned agent through the value obtained by subtracting the detractor ratio from the promoter ratio (NPS). For example, if users rate sleep care content or services on a scale of 10, users who give a very good rating of 9 to 10, who are expected to actively recommend the product or service, can be classified as recommenders; users who give a rating of 7 to 8, who had a satisfactory experience but are expected not to actively recommend it, can be classified as neutrals; and users who give a rating of 0 to 6, who are dissatisfied with the product or service and are likely not to recommend it to others or even not recommend it, can be classified as non-recommenders.

[0083] NPS is calculated by subtracting the non-recommender ratio from the recommender ratio.

[0084] NPS = (Promoter Ratio) - (Detractor Ratio)

[0085] For example, if 60% of the 100 responding users are recommenders and 20% are non-recommenders, the NPS is 60 - 20 = 40.

[0086] Through this, the performance of the Agent that has undergone user-specific reinforcement learning is evaluated, and if the NPS score is low, reinforcement learning can be performed to create an upward trend in the NPS score, or the established Agent can be discarded and a new Agent provided.

[0087] The sleep environment control module (257) is intended to control peripheral devices to provide an environment suitable for sleep. For example, the sleep environment control module (257) may include commands for turning off or dimming lighting devices to make the room dark or dark, controlling curtain rail motors to close curtains, and controlling the temperature of air conditioners or heaters.

[0088] Additionally, the sleep environment control module (257) can provide a noise canceling function. Specifically, the sensor unit (100) can measure ambient noise. The noise measured by the sensor unit (100) is transmitted to the processor (200) to perform signal processing for analyzing the ambient noise. Through the sleep environment control module (257), the processor (200) can determine noise of a certain magnitude or greater and noise having a periodic pattern, and generate data capable of outputting sound waves to cancel it out.

[0089] Specifically, the processor (200) can compare the noise measured through the sensor unit (100) during sleep activity with a threshold point, and if the measured noise is greater than or equal to the threshold point, determine the pattern of the measured noise, and control the output unit (300) to output the opposite noise having an inverse phase noise pattern opposite to the measured noise pattern.

[0090] The wake-up notification module (258) provides a function to wake up a sleeping user. The wake-up notification module (258) may have a wake-up time set by the user. When the wake-up time is set by the wake-up notification module (258), the processor (200) monitors the sleep stage from a predetermined time before the set wake-up time, for example, 30 minutes before. By monitoring the sleep stage in advance, the processor (200) initiates a wake-up notification function that induces the REM sleep stage, which allows the user to feel less fatigue upon waking. Since the REM sleep stage is close to a fully awake state during the sleep cycle, one can wake up quickly and refreshed.

[0091] The processor (200) can provide binaural beat content that induces a change in the sleep stage if the REM sleep stage is not reached by the wake-up time via the wake-up notification module (258). The binaural beat can be updated to gradually change brain waves from the current sleep stage to transition to a wake state.

[0092] The processor (200) can resume the wake-up notification function when the user's relapse is detected from the biosignal measured in real time by the wake-up notification module (258). Immediate measures to wake the user who has fallen asleep again can be performed.

[0093] The wake-up notification module (258) may have a control method for waking the user through means such as sound, light, or vibration at the optimal sleep stage for waking up. In one example, the wake-up notification module (258) may create a wake-up environment by using an eye mask equipped with a display and a speaker to provide a sound source to output an alarm sound and to provide a display image to emit light. When the wake-up time arrives via the wake-up notification module (258), the processor (200) may provide a wake-up environment by controlling the light and alarm through the eye mask. The eye mask may gradually illuminate the light and emit light at an illuminance that is not too blinding to the eyes, and may output natural sounds that allow one to feel a good morning mood or binaural beat sounds of a specific frequency that clear or activate the brain.

[0094] The exercise management module (259) enables the provision of a personalized fitness program to the user when the user's activity is classified as an exercise activity. The exercise management module (259) can provide exercise content. The processor (200) can provide exercise content in which at least one of the rest time and exercise intensity varies based on the biosignal measured by the exercise management module (259).

[0095] Specifically, biometric data such as the user's heart rate, respiratory rate, and heart rate variability, measured or calculated by the sensor unit (100), can serve as a basis for setting a rest time suitable for the user's physical ability or adjusting the exercise intensity. The exercise intensity can be adjusted by the type of exercise, number of repetitions, weight, etc. The rest time between the end of one set and the start of the next set can be set by the user's biometric data measured during the exercise. Meanwhile, the exercise management module (259) can generate an exercise program suitable for the user based on identified user information and lasting for the exercise time set by the user. The exercise management module (259) can sort the types of exercise that reflect the user's preferences in order. And, the exercise intensity can be set from the user's exercise ability information, which is identified user information.

[0096] FIG. 6 is a diagram illustrating the configuration of an embodiment for managing exercise activities of a biosignal-based individual sleep induction management system.

[0097] Referring to FIG. 6, the system (1000) is composed of a sensor unit (100) including a plurality of sensors (110), and a smart mirror having a processor (200) and an output unit (300). A user sits on a mat (100) facing the smart mirror and exercises. The mat (100) measures the user's ballistocardiogram, etc. The measured ballistocardiogram data is transmitted to the smart mirror. The processor (200) analyzes the ballistocardiogram data to calculate the user's heart rate, respiratory rate, etc. The calculated heart rate, respiratory rate, etc. are used to set an efficient recovery time and to adjust the intensity of the next set of exercise.

[0098] The smart mirror captures the user's exercise posture through a camera. The captured exercise posture can be used to provide posture correction feedback. The output unit (300) can provide exercise content that guides the exercise posture. The processor (200) can calculate an exercise score based on the user's exercise performance and output it to the output unit (300).

[0100] FIG. 5 is a schematic diagram illustrating embodiments of the sensor unit, processor, and output unit of a biosignal-based individual sleep induction management system.

[0101] Referring to FIG. 5, the sensor unit (100) may be composed of at least one of a microphone (100-1), a motion bed (100-2), a smartwatch (100-3), and an electroencephalogram (EEG) band (100-4). Additionally, a smart ring and a smart belly band may be included. The motion bed (100-2), which includes a piezoelectric sensor (110), can measure ballistocardiogram (BCG) signals. The smartwatch (100-3) can measure electrocardiogram (ECG) signals. The EEG band (100-4) can measure electroencephalogram (EEG) signals. The smart ring can measure photoplethysmography (PPG) signals. The smart belly band can measure uterine electromyography (EMG) signals. The microphone (100-1) can measure ambient noise or snoring sounds during the user's sleep.

[0102] The processor (200) can establish a communication connection with the sensor unit (100) and the output unit (300) via a wireless communication router (200-2). In one embodiment, the processor (200) may include a server (200-1) connected via a broadband network. The processor (200) may be composed of a user's smartphone (200-3), a set-top box (200-4) for sleep induction management services, and / or a PC (200-5).

[0103] The output unit (300) may be composed of an eye mask (300-1) and / or a speaker (300-2). The eye mask (300-1) may include a display (310) or a light source. The output unit (300) may include the smart mirror of FIG. 6, a TV, a moving part of a motion bed, a curtain rail motor, a heating and cooling air conditioner, etc.

[0105] FIG. 8 is a flowchart illustrating a biosignal-based individual sleep induction management method according to one embodiment of the present invention.

[0106] Referring to FIG. 8, a biosignal-based individual sleep induction management method can be started (S110) by configuring the system (1000) and turning on the power.

[0107] First, a step (S120) of measuring a biosignal in real time is included. The biosignal to be measured may be a ballistocardiogram signal from a piezoelectric sensor.

[0108] A step (S130) of identifying a user based on a measured biosignal and classifying user activity into one of the groups including exercise and sleep is included. The user may be identified based on feature points of the ballistocardiogram signal. The user's activity may be determined based on the user's posture determined by the ballistocardiogram signal.

[0109] If classified as sleep activity, a step (S140) for classifying sleep stages is included. Sleep stages are classified at least into Wake / REM / NREM stages, and sleep stages can be classified by an AI model based on ballistocardiogram signals.

[0110] In addition, the step (S150) of providing sleep care content based on identified user information and classified sleep stages is included. The sleep care content may include binaural beats that induce sleep or binaural beats that induce waking. The sleep care content may be customized for the user. The sleep care content may be selected or adjusted by a reinforcement learning AI model based on the user's preferences or the user's sleep quality.

[0111] The biosignal-based individual sleep induction management method system (1000) can be terminated by controlling the user to turn off the power (S160).

[0113] FIG. 9 is a flowchart illustrating a specific embodiment of a biosignal-based individual sleep induction management method according to an embodiment of the present invention.

[0114] Referring to FIG. 9, the biosignal-based individual sleep induction management method is initiated by a user's start command or power-on control (S205).

[0115] A biosignal measured by one or more sensors of the sensor unit (100) is collected through an independent channel, and the collected signal may undergo preprocessing processes such as amplification, filtering, noise removal, SNR comparison, and valid signal extraction (S210).

[0116] Personal authentication (user identification) of the user is performed based on biometric signals (S215).

[0117] And, the user's activity is classified (S220).

[0118] If the user activity is classified as exercise, the process branches to step S225. During the exercise activity, appropriate exercise content is recommended to the user (S230). In the initial state, the exercise content may be a default recommended program.

[0119] The heart rate and / or respiratory rate of a user exercising are measured (S235). While they may be measured directly by the user's wearable device, if a ballistics signal is measured by a piezoelectric sensor, the heart rate and respiratory rate can be calculated from the ballistics signal.

[0120] After one set of exercises is finished, a rest time is set between the next set of repetitions (S240). The rest time can be set based on heart rate and breathing rate.

[0121] The exercise is repeated until the total number of sets is completed (S245: no). The exercise is terminated when the exercise time or number of repetitions is finished (S245: yes).

[0122] When the exercise is finished, an exercise report is generated (S250). The exercise report may include information such as exercise intensity, exercise time, and calories burned, which can evaluate the user's exercise performance.

[0123] The exercise process can be automatically terminated upon the end of the exercise and can also be terminated by a user's termination command (S295).

[0124] Meanwhile, as a result of classifying user activity (S220), if the user is classified as sleep activity, branch to step S255.

[0125] Sleep-inducing content is recommended to the user who is starting to sleep (S260). The recommended content is fixed in the initial state but can be adjusted to suit the user through learning. The recommended sleep-inducing content is provided while being updated according to the sleep stage (S260).

[0126] While providing sleep care content, the sleep of the user who is sleeping is analyzed (S265). The sleep analysis can determine the sleep effect of the sleep care content on the user.

[0127] If the user does not fall asleep, the content is updated or continued and repeated until the user falls asleep (S270: no). When the user falls asleep (S270: yes), the sleep environment, including ambient noise analysis, is controlled (S275). Sleep environment control may include noise canceling.

[0128] When a wake-up time is set, the wake-up notification function (S280) is activated. The wake-up notification function may include providing binaural beats to induce the user into the REM sleep stage at the wake-up time. When sufficient wake-up conditions are met, wake-up alarm content using an alarm sound and / or light may be provided.

[0129] If the user falls asleep again (S285: yes), the wake-up alarm function is performed again (S280).

[0130] When the user wakes up (S285: no), a sleep report is generated (S290). The sleep report may include information such as the playback time of sleep care content, total sleep time, and sleep quality.

[0131] The process can be automatically terminated with the output of the sleep report, and the process can be terminated by a user's control command (S295).

[0133] Meanwhile, a non-transitory computer-readable medium may be provided, which stores a program for executing a biosignal-based individual sleep induction management method according to the present invention by a computer.

[0134] A non-transient readable medium refers to a medium that stores data semi-permanently and can be read by a device, rather than a medium that stores data for a short moment, such as a register, cache, or memory. Specifically, the various applications or programs described above may be stored and provided on non-transient readable media such as CDs, DVDs, hard disks, Blu-ray discs, USBs, memory cards, and ROMs.

[0135] Furthermore, although preferred embodiments of the present invention have been illustrated and described above, the present invention is not limited to the specific embodiments described above. It is understood that various modifications can be made by those skilled in the art without departing from the essence of the invention as claimed in the claims, and such modifications should not be understood individually from the technical spirit or perspective of the present invention. Explanation of the symbols

[0136] 100: Sensor unit 200: Processor 300: Output section

Claims

Claim 1 A sensor unit for measuring a biosignal in real time, and an output unit for outputting at least one of an acoustic signal and a video signal; A biosignal-based individual sleep induction management system comprising: a processor that identifies a user based on the measured biosignal, classifies the user's activity into one of groups including exercise and sleep, classifies the sleep stage if classified as sleep activity, and provides sleep care content through the output unit based on the identified user information and the classified sleep stage; wherein the processor updates the sleep care content according to changes in the sleep stage, disables the output of the sleep care content if the sleep stage is classified as a deeper sleep stage than a preset standard, and, if a wake-up time is set, monitors the sleep stage from 30 minutes prior to the set wake-up time to initiate a wake-up notification function that induces the user into the REM sleep stage upon waking, provides binaural beat content that induces a change in the sleep stage if the user fails to reach the REM sleep stage by the wake-up time, controls light and an alarm through the output unit to provide a wake-up environment when the wake-up time arrives, and resumes the wake-up notification function if the user's re-sleep is detected from the biosignal measured in real time. Claim 2 A biosignal-based individual sleep induction management system according to claim 1, wherein the sensor unit comprises a plurality of sensors arranged inside a sleep mat, and the processor measures the user's ballistocardiogram signal (BCG) through the sensor unit and identifies the user based on the characteristic points of the measured ballistocardiogram signal. Claim 3 In paragraph 2, the processor determines the posture of the user of the sleep mat from the measured ballistocardiogram signal, classifies the user's activities according to the posture, and classifies the sleep stages based on the ballistocardiogram signal, a biosignal-based individual sleep induction management system. Claim 4 A biosignal-based individual sleep induction management system according to claim 3, wherein the sensor unit measures ambient noise, the output unit outputs opposite noise to cancel out the noise, and the processor, during sleep activity, compares the noise measured through the sensor unit with a threshold point, and if the measured noise is greater than or equal to the threshold point, determines the pattern of the measured noise, and controls the output unit to output the opposite noise having an inverse phase noise pattern opposite to the measured noise pattern. Claim 5 A biosignal-based individual sleep induction management system according to claim 1, further comprising: an airable device that determines a sleep stage by measuring the breathing sound of a user during sleep; or a wearable device attached to the user's body that determines a sleep stage by measuring brain waves or photocircumference; wherein the processor labels the biosignal data measured by the sensor unit using the sleep stage determined by the airable device or the wearable device as label data, and extracts a deep learning-based learning model by utilizing the labeled data as learning data. Claim 6 delete Claim 7 delete Claim 8 A biosignal-based individual sleep induction management system according to claim 1, wherein the processor monitors the sleep quality of a user who receives the sleep care content through the sensor unit, generates a cluster based on at least one characteristic among the improvement or decline of the monitored sleep quality, the playback time of the content provided according to the sleep stage, and the change in total sleep time, and establishes an agent that performs reinforcement learning through rewards in which the characteristics belonging to the cluster are improved for each generated cluster according to which content is provided by the output unit. Claim 9 A biosignal-based individual sleep induction management system, wherein, in claim 8, the processor classifies multiple users who have experienced the content into promoters, passives, and detractors according to user-specific recommendation scores for the provided content, and evaluates the reinforcement-learned agent through the value obtained by subtracting the ratio of detractors from the ratio of promoters (NPS). Claim 10 A biosignal-based individual sleep induction management system according to claim 1, wherein the processor provides exercise content in which at least one of rest time and exercise intensity changes based on the identified user information and the measured biosignal when the user's activity is classified as exercise activity. Claim 11 A method for managing individual sleep induction based on biosignals comprises: a step of measuring biosignals in real time; a step of identifying a user based on the measured biosignals, classifying the user's activity into one of a group including exercise and sleep, classifying a sleep stage if classified as sleep activity, and providing sleep care content based on the identified user information and the classified sleep stage; the step of providing sleep care content includes updating the sleep care content according to changes in the sleep stage, and if the sleep stage is classified as a sleep stage deeper than a preset standard, deactivating the output of the sleep care content; the step of providing sleep care content includes, if a wake-up time is set, monitoring the sleep stage from 30 minutes before the set wake-up time to initiate a wake-up notification function that induces the user into the REM sleep stage upon waking, and if the user fails to reach the REM sleep stage by the wake-up time, providing binaural beat content that induces a change in the sleep stage, and when the wake-up time arrives, controlling light and an alarm through an output unit to provide a wake-up environment, and if the user's re-sleep is detected from the biosignals measured in real time, reactivating the wake-up notification function. A biosignal-based personalized sleep induction management method. Claim 12 A computer-readable recording medium having a program for executing the biosignal-based individual sleep induction management method described in paragraph 11 on a computer.

Citation Information

Patent Citations

  • Apparatus and Method for assisting sound sleep

    KR1020160092889A

  • Method and device for estimating physical state of a user

    KR1020200011818A

  • Sleep monitoring and sleep induction device

    KR1020200064288A

  • Methods and system for customized sleep management

    KR1020210027033A

  • Sleep stage analysis method based on EEG data

    KR1020220074069A