Sleep monitoring method and system using same
By integrating multiple sensors into the student's sleep monitoring belt, using machine learning and hidden Markov models to analyze sleep state, the problem of students not being able to monitor sleep during school is solved, and accurate sleep monitoring and abnormal display without wearing is achieved.
Patent Information
- Application Number
- CN202510692198.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-05-27
- Publication Date
- 2025-08-12
AI Technical Summary
During school, students cannot use their smartphones to monitor their sleep due to various restrictions, and wearing smart bracelets or watches is easy to forget or lose, which makes teachers unable to understand students' sleep in a timely manner.
The sleep monitoring belt is adopted, and a variety of sensors are integrated to collect real-time signals and historical signals. The sleep analysis model is constructed through machine learning algorithms and hidden Markov models, analyzing users' sleep state and transfer patterns, generating health reports, and storing and displaying them through servers and terminals.
It realizes sleep monitoring without wearing, improves the accuracy and real-time nature of sleep results, avoids the problems of forgetting to wear and losing, and provides convenient sleep quality assessment and visual display of abnormal situations.
Smart Images
Figure CN120458515A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of sleep monitoring, and in particular relates to a sleep monitoring method and a system using the method. Background Art
[0002] Existing sleep monitoring can generally be carried out using smartphones, smart bracelets or watches. The principle of sleep monitoring through smartphones is mainly to use the built-in accelerometer of the phone to sense movements such as turning over during sleep and judge the sleep state. The principle of monitoring through smart bracelets or watches is mainly to use the various sensors built into the smart bracelets or watches to judge the user's sleep state, heart rate, respiratory rate, blood oxygen saturation and other indicators.
[0003] Due to various restrictions in schools, most students are unable to use smartphones while at school, resulting in the inability to monitor their sleep. If students choose to wear smart bracelets or watches for sleep monitoring, they are likely to forget to wear or lose the smart bracelets or watches, making it impossible for teachers to intuitively understand the students' sleep conditions and unable to complete the need to trace their sleep conditions.
[0004] Therefore, there is an urgent need for an effective solution to solve the problem that students easily forget to wear or lose their smart bracelets or watches and cannot understand their sleeping status in time. Summary of the Invention
[0005] The purpose of the present invention is to provide a sleep monitoring method and a system using the method to solve the above-mentioned problems existing in the prior art.
[0006] In order to achieve the above object, the present invention adopts the following technical solutions: In a first aspect, the present invention provides a sleep monitoring method, comprising: Acquire real-time signals collected by sensors and historical signals stored in cloud storage or a local database, the real-time signals including real-time pressure, real-time blood oxygen saturation, real-time heart rate, real-time respiratory rate, and real-time human activity signals; the historical signals including historical pressure, historical blood oxygen saturation, historical heart rate, historical respiratory rate, and historical human activity signals, and preprocess the real-time and historical signals; Training a pre-built sleep analysis model based on the pre-processed historical signal to obtain a trained sleep analysis model; Inputting the pre-processed real-time signal into the trained sleep analysis model for processing to obtain the user's sleep results, wherein the user's sleep results include the user's sleep state and the user's sleep transition pattern, and generating a health report based on the user's sleep results; Upload sleep results and health reports to cloud storage or local database for storage.
[0007] In one possible design, the real-time signal and the historical signal are pre-processed, including: Perform data cleaning operations on real-time signals and historical signals to obtain cleaned real-time signals and historical signals; Performing denoising on the cleaned real-time signal and historical signal to obtain denoised real-time signal and historical signal; The denoised real-time signal and the historical signal are normalized to obtain the preprocessed real-time signal and the historical signal.
[0008] In one possible design, the sleep analysis model is constructed based on a machine learning algorithm and a hidden Markov model; and the pre-constructed sleep analysis model is trained based on the pre-processed historical signal to obtain a trained sleep analysis model, including: Label the preprocessed historical signals to obtain labeled historical signals; Use machine learning algorithms to extract features from labeled historical signals to obtain historical sequence features; Taking the historical sequence features as observation values, and obtaining corresponding sleep states based on the historical sequence features, the sleep states include awake state, light sleep, deep sleep, and rapid eye movement (REM) sleep; Initializing the parameters of the hidden Markov model, including a state transition probability matrix, an observation probability distribution, and an initial state probability vector, and training the hidden Markov model using historical sequence features and their corresponding sleep states to obtain a trained sleep analysis model.
[0009] In one possible design, a machine learning algorithm is used to extract features from labeled historical signals to obtain historical sequence features, including: A convolutional neural network is used to extract features from labeled historical signals to obtain historical sleep feature sequences. The RNN model is used to process the historical sleep feature sequence to obtain the historical sequence features.
[0010] In one possible design, a convolutional neural network is used to extract features from labeled historical signals to obtain a historical sleep feature sequence, including: The convolution layer is used to extract features from the labeled historical signals to obtain local historical features; The pooling layer reduces the dimension of local historical features to obtain key sleep features; The key features are combined through the fully connected layer to obtain the historical sleep feature sequence.
[0011] In a second aspect, the present invention provides a sleep monitoring system, including a sleep monitoring belt, a server, and a terminal; The sleep monitoring belt includes a sensor unit and a communication module; the sensor unit is used to detect real-time signals; The communication module is communicatively connected to the sensor unit and is used to upload real-time signals to the terminal; The terminal includes a student terminal and a teacher terminal, and the sleep monitoring belt, the student terminal and the teacher terminal are all connected to the server for communication; The server is configured to execute the sleep monitoring method according to the first aspect; The student terminal is used to check personal sleep status and generate a personal sleep report based on the personal sleep status; The teacher terminal is used to view the individual sleep status of all student terminals and generate a class sleep report based on the individual sleep status of all student terminals.
[0012] In one possible design, the sensor unit includes a pressure sensor, a health monitoring sensor, a blood oxygen sensor, a heart rate sensor, a respiratory rate sensor, and a human presence sensor. The pressure sensor is used to detect pressure; the health monitoring sensor is used to detect blood glucose concentration; the blood oxygen sensor is used to detect real-time blood oxygen saturation; the heart rate sensor is used to detect heart rate changes; the respiratory rate sensor is used to record the number of breaths per minute; and the human presence sensor is used to detect whether the target is human activity.
[0013] In a possible design, the user's sleep result also includes the user's sleep score. The server is further configured to determine whether the user's sleep score exceeds a preset score threshold. If so, a warning message is generated and uploaded to the terminal.
[0014] In one possible design, the server is also used to determine whether the user has fallen asleep at the preset bedtime based on real-time signals, and if not, trigger an intelligent sleep reminder; it is also used to determine whether the user is in a sleeping state at the preset wake-up time based on real-time signals, and if in a sleeping state, trigger a wake-up reminder.
[0015] The beneficial effects of the present invention are as follows: (1) The present invention discloses a sleep monitoring method and a system using the method, which trains a sleep analysis model and uses the trained sleep analysis model to process data collected by sensors in real time, analyzes the user's sleep results, and uses the sleep analysis model to analyze the data during sleep, thereby improving the accuracy and real-time performance of the sleep results.
[0016] (2) The present invention uses a sleep monitoring belt with multiple sensors to collect physiological indicators, body conditions and behaviors during sleep, thereby evaluating the quality of the sleep process. It is not easy to forget to wear or lose the belt, and it is easy to use. The sleep results are displayed in the form of visual charts, making it easy to observe abnormal conditions during sleep. BRIEF DESCRIPTION OF THE DRAWINGS
[0017] Figure 1 This is a flowchart of the sleep monitoring method provided in Example 1. DETAILED DESCRIPTION
[0018] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the present invention will be briefly introduced below in conjunction with the drawings and the description of the embodiments or the prior art. Obviously, the following description of the structure of the drawings is only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative work. It should be noted that the description of these embodiments is used to help understand the present invention, but does not constitute a limitation of the present invention.
[0019] Example 1: like Figure 1 As shown, this embodiment provides a sleep monitoring method, including but not limited to the following steps: S1. Acquire real-time signals collected by sensors and historical signals stored in cloud storage or a local database, the real-time signals including real-time pressure, real-time blood oxygen saturation, real-time heart rate, real-time respiratory rate, and real-time human activity signals; the historical signals including historical pressure, historical blood oxygen saturation, historical heart rate, historical respiratory rate, and historical human activity signals, and preprocess the real-time and historical signals; Specifically, in step S1, the real-time signal and the historical signal are pre-processed, including: S101 performs data cleaning operations on real-time signals and historical signals to obtain cleaned real-time signals and historical signals; S102. Perform denoising on the cleaned real-time signal and the historical signal to obtain the denoised real-time signal and the historical signal; Preferably, wavelet transform denoising and sliding average filtering denoising can be selected for denoising. The principle of wavelet transform denoising is to separate the signal and noise through multi-scale decomposition, suppress the noise coefficient by threshold processing, and finally reconstruct the signal; the principle of sliding average filtering denoising is to reduce the noise in the signal by calculating the average value of the data in a window in the data sequence. Denoising can improve the usability of the signal and facilitate subsequent model training and application.
[0020] S103. Perform normalization processing on the denoised real-time signal and the historical signal to obtain pre-processed real-time signal and historical signal.
[0021] S2. training a pre-built sleep analysis model based on the preprocessed historical signals to obtain a trained sleep analysis model; Furthermore, the sleep analysis model is constructed based on a machine learning algorithm and a Hidden Markov Model (HMM). In this embodiment, preferably, the machine learning algorithm adopts a Convolutional Neural Network (CNN) and an RNN model (Recurrent Neural Network).
[0022] Specifically, in step S2, the pre-built sleep analysis model is trained based on the pre-processed historical signal to obtain the trained sleep analysis model, including: S201. Label the preprocessed historical signal to obtain a labeled historical signal; In this embodiment, the pre-processed historical signals are labeled, and the labeled historical signals are used as a training set to train the sleep analysis model.
[0023] S202. Using a machine learning algorithm to extract features from the labeled historical signals to obtain historical sequence features; Specifically, in step S202, a machine learning algorithm is used to extract features from the labeled historical signals to obtain historical sequence features, including: S2021. Use a convolutional neural network to extract features from labeled historical signals to obtain a historical sleep feature sequence; Specifically, in step S2021, a convolutional neural network is used to extract features from the labeled historical signals to obtain a historical sleep feature sequence, including: S20211. Perform feature extraction on the labeled historical signal through a convolutional layer to obtain local historical features; wherein the local historical features are features that can reflect the sleep state.
[0024] S20212. Reduce the dimension of local historical features through pooling layers to obtain key sleep features; S20213. Combine key features through a fully connected layer to obtain a historical sleep feature sequence.
[0025] S2022. Use the RNN model to process the historical sleep feature sequence to obtain historical sequence features.
[0026] Among them, the principle of the RNN model is to capture the dynamic characteristics between sequences through a cyclic structure. The historical sleep feature sequence contains timestamps. The RNN model is used to capture the temporal dependency of the historical sleep feature sequence to obtain the historical sequence characteristics.
[0027] S203. Using the historical sequence features as observation values, and obtaining corresponding sleep states based on the historical sequence features, the sleep states including awake state, light sleep, deep sleep, and rapid eye movement (REM) sleep; S204. Initialize the parameters of the hidden Markov model, which include a state transition probability matrix, an observation probability distribution, and an initial state probability vector. Use the historical sequence features and their corresponding sleep states to train the hidden Markov model to obtain a trained sleep analysis model.
[0028] S3. The pre-processed real-time signal is input into the trained sleep analysis model for processing to obtain the user's sleep results, which include the user's sleep state and the user's sleep transition pattern, and generate a health report based on the user's sleep results; Specifically, the user's sleep results include the user's sleep state and the user's sleep transition pattern. The user's sleep quality score is obtained based on the user's sleep state, and a health report is generated based on the user's sleep state, the user's sleep quality score and the user's sleep transition pattern.
[0029] S4. Upload the sleep results and health report to cloud storage or local database for storage.
[0030] Furthermore, we regularly connect with mental health experts and nutritionists to provide online consultation services based on health reports, and simultaneously improve dietary recipes to provide complete health services.
[0031] This embodiment provides a sleep monitoring method that obtains real-time and historical signals collected by sensors, preprocesses the real-time and historical signals, and uses the preprocessed historical signals to train a pre-built sleep analysis model to obtain a trained sleep analysis model. The trained sleep analysis model is then used to process the preprocessed real-time signals to obtain the user's sleep results, and a corresponding health report is generated based on the user's sleep results. The sleep analysis model, constructed using a machine learning algorithm and a hidden Markov model, can accurately analyze the user's sleep state during sleep and derive a sleep quality score based on the distribution of sleep states, facilitating observation of the user's sleep state and analysis of their health indicators.
[0032] Example 2: This embodiment provides a sleep monitoring system, characterized by comprising a sleep monitoring belt, a server and a terminal; The sleep monitoring belt includes a sensor unit and a communication module; the sensor unit is used to detect real-time signals; The communication module is communicatively connected to the sensor unit and is used to upload real-time signals to the terminal; The terminal includes a student terminal and a teacher terminal, and the sleep monitoring belt, the student terminal and the teacher terminal are all connected to the server for communication; The server is configured to execute the sleep monitoring method described in Example 1; The student terminal is used to check personal sleep status and generate a personal sleep report based on the personal sleep status; The teacher terminal is used to view the individual sleep status of all student terminals and generate a class sleep report based on the individual sleep status of all student terminals.
[0033] In one possible design, the sensor unit includes but is not limited to a pressure sensor, a health monitoring sensor, a blood oxygen sensor, a heart rate sensor, a respiratory rate sensor, a sound sensor, and a human presence sensor. The pressure sensor is used to detect pressure; the health monitoring sensor is used to detect blood glucose concentration; the blood oxygen sensor is used to detect real-time blood oxygen saturation; the heart rate sensor is used to detect heart rate changes; the respiratory rate sensor is used to record the number of breaths per minute; the sound sensor is used to detect environmental sounds; and the human presence sensor is used to detect whether the target is human activity.
[0034] Specifically, the sleep monitoring belt has multiple sensors, including a human presence sensor, pressure sensor, and sound sensor integrated inside the pillow; a respiratory rate sensor and a heart rate sensor are attached to the surface of the pillow, and other sensors are connected to the outside of the pillow, such as a blood oxygen sensor and a health monitoring sensor.
[0035] In this embodiment, the sleep monitoring belt and the pillow cooperate with each other to monitor pressure, blood oxygen saturation, heart rate, respiratory rate and human activity signals without having to be worn on the wrist, thus avoiding the phenomenon of forgetting to wear or losing the belt.
[0036] In a possible design, the user's sleep result also includes the user's sleep score. The server is further configured to determine whether the user's sleep score exceeds a preset score threshold. If so, a warning message is generated and uploaded to the terminal.
[0037] In one possible design, the server is also used to determine whether the user has fallen asleep at the preset bedtime based on real-time signals, and if not, trigger an intelligent sleep reminder; it is also used to determine whether the user is in a sleeping state at the preset wake-up time based on real-time signals, and if in a sleeping state, trigger a wake-up reminder.
[0038] This embodiment discloses a sleep monitoring system, including a sleep monitoring belt, a server, and a terminal. The sleep monitoring belt includes a sensor unit and a communication module, wherein the sensor unit detects the user's real-time signal, and the communication module uploads the real-time signal to the terminal. Through the cooperation of multiple sensors, the physiological state and physical data during sleep are collected to facilitate subsequent model processing. The terminal includes a student terminal and a teacher terminal. The sleep monitoring belt, the student terminal, and the teacher terminal are all communicatively connected to the server, wherein the server processes and stores historical signals, real-time signals, the user's sleep results, and health reports. The visualization display module visualizes the user's sleep results and health reports in the form of charts, and uses intelligent reminders to urge the user to develop good work and rest conditions. The user's sleep results are displayed on a visual dashboard, making it convenient for teachers to view abnormal students. The teacher terminal can view the user's personal sleep report and the class sleep report, while students can only view their personal sleep reports. It is less likely to forget to wear or lose the belt. Since the belt is wireless, there is no need to worry about electrical safety issues. It is suitable for students and has broad application prospects.
[0039] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention shall be included in the scope of protection of the present invention.
Claims
1. A sleep monitoring method, characterized in that: include: Acquire real-time signals collected by sensors and historical signals stored in cloud storage or a local database, the real-time signals including real-time pressure, real-time blood oxygen saturation, real-time heart rate, real-time respiratory rate, and real-time human activity signals; the historical signals including historical pressure, historical blood oxygen saturation, historical heart rate, historical respiratory rate, and historical human activity signals, and preprocess the real-time and historical signals; Training a pre-built sleep analysis model based on the pre-processed historical signal to obtain a trained sleep analysis model; Inputting the pre-processed real-time signal into the trained sleep analysis model for processing to obtain the user's sleep results, wherein the user's sleep results include the user's sleep state and the user's sleep transition pattern, and generating a health report based on the user's sleep results; Upload sleep results and health reports to cloud storage or local database for storage.
2. The sleep monitoring method according to claim 1, characterized in that: Preprocessing of real-time and historical signals, including: Perform data cleaning operations on real-time signals and historical signals to obtain cleaned real-time signals and historical signals; Performing denoising on the cleaned real-time signal and historical signal to obtain denoised real-time signal and historical signal; The denoised real-time signal and the historical signal are normalized to obtain the preprocessed real-time signal and the historical signal.
3. The sleep monitoring method according to claim 1, wherein: The sleep analysis model is constructed based on a machine learning algorithm and a hidden Markov model; the pre-constructed sleep analysis model is trained based on the pre-processed historical signal to obtain a trained sleep analysis model, including: Label the preprocessed historical signals to obtain labeled historical signals; Use machine learning algorithms to extract features from labeled historical signals to obtain historical sequence features; Taking the historical sequence features as observation values, and obtaining corresponding sleep states based on the historical sequence features, the sleep states include awake state, light sleep, deep sleep, and rapid eye movement (REM) sleep; Initializing the parameters of the hidden Markov model, including a state transition probability matrix, an observation probability distribution, and an initial state probability vector, and training the hidden Markov model using historical sequence features and their corresponding sleep states to obtain a trained sleep analysis model.
4. The sleep monitoring method according to claim 3, characterized in that: Machine learning algorithms are used to extract features from labeled historical signals to obtain historical sequence features, including: A convolutional neural network is used to extract features from labeled historical signals to obtain historical sleep feature sequences. The RNN model is used to process the historical sleep feature sequence to obtain the historical sequence features.
5. The sleep monitoring method according to claim 4, characterized in that: A convolutional neural network is used to extract features from labeled historical signals to obtain historical sleep feature sequences, including: The convolution layer is used to extract features from the labeled historical signals to obtain local historical features; The pooling layer reduces the dimension of local historical features to obtain key sleep features; The key features are combined through the fully connected layer to obtain the historical sleep feature sequence.
6. A sleep monitoring system, characterized in that: Includes sleep monitoring belt, server and terminal; The sleep monitoring belt includes a sensor unit and a communication module; the sensor unit is used to detect real-time signals; The communication module is communicatively connected to the sensor unit and is used to upload real-time signals to the terminal; The terminal includes a student terminal and a teacher terminal, and the sleep monitoring belt, the student terminal and the teacher terminal are all connected to the server for communication; The server is configured to execute the sleep monitoring method according to any one of claims 1 to 5; The student terminal is used to check personal sleep status and generate a personal sleep report based on the personal sleep status; The teacher terminal is used to view the individual sleep status of all student terminals and generate a class sleep report based on the individual sleep status of all student terminals.
7. The sleep monitoring system according to claim 6, characterized in that: The sensor unit includes a pressure sensor, a health monitoring sensor, a blood oxygen sensor, a heart rate sensor, a respiratory rate sensor, and a human presence sensor. The pressure sensor is used to detect pressure; the health monitoring sensor is used to detect blood sugar concentration; the blood oxygen sensor is used to detect blood oxygen saturation; the heart rate sensor is used to detect heart rate changes; and the respiratory rate sensor is used to record the number of breaths per minute. The human presence sensor is used to detect whether a target is a human activity.
8. The sleep monitoring system according to claim 6, characterized in that: The user's sleep result also includes the user's sleep score. The server is further configured to determine whether the user's sleep score exceeds a preset score threshold. If so, a warning message is generated and uploaded to the terminal.
9. The sleep monitoring system according to claim 6, characterized in that: The server is also used to determine whether the user has fallen asleep at the preset bedtime based on real-time signals, and if not, trigger an intelligent sleep reminder; it is also used to determine whether the user is in a sleeping state at the preset wake-up time based on real-time signals, and if so, trigger a wake-up reminder.