An AI and sensor data-based sleep health assessment management method and system
By collaboratively monitoring users' sleep states using millimeter-wave radar, EEG electrodes, and sensor arrays, and combining this with AI model analysis, a user sleep model is established. This solves the problem of sleep state monitoring errors in existing technologies and enables more accurate sleep health assessment and environmental interference identification.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- ZHEJIANG YANKE INFORMATION TECH CO LTD
- Filing Date
- 2026-01-13
- Publication Date
- 2026-07-03
AI Technical Summary
In existing technologies, there are errors in the monitoring of users' sleep stages, which affects the accuracy of sleep health status assessment.
The system uses millimeter-wave radar, EEG electrodes, and sensor arrays to monitor the user's sleep state in a coordinated manner. It analyzes reflected signals, brainwave data, and physiological parameter data through an AI model to establish a user sleep model and conducts sleep health assessments in conjunction with preset evaluation algorithms.
It improves the accuracy of sleep state assessment, can identify environmental disturbances, and provides useful information for users with sleep disorders.
Smart Images
Figure CN122320459A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of sleep health assessment technology, specifically to a sleep health assessment and management method and system based on AI and sensor data. Background Technology
[0002] As people pay more attention to their health, their requirements for sleep quality are also increasing. Among these requirements, monitoring sleep health is not only about recording sleep duration, but also about decoding the body's nighttime repair code through multidimensional physiological data. This provides a scientific anchor for disease early warning, chronic disease management, and quality of life improvement. Sleep health monitoring mainly involves acquiring data such as the proportion of different sleep stages (N1 light sleep, N2 medium sleep, and N3 deep sleep) and REM latency. The user's sleep health status is then assessed based on this data.
[0003] In existing technologies, the monitoring of a user's sleep stages can be achieved in various ways. A common method is to determine the sleep stage by the characteristics of brain waves. When the main characteristic of brain waves is delta waves (with delta waves accounting for more than 50%), it indicates that the user has entered a deep sleep state. The brain wave characteristics of the light sleep stage are sleep spindle waves and K-complex waves, with delta waves accounting for less than 20%. Therefore, the sleep stages of a user can be divided by the characteristics of brain waves.
[0004] Because each individual's sleep state is different, the existing process of dividing sleep stages has a certain degree of error. When assessing a user's overall sleep state, this error has a relatively small impact. However, when conducting detailed analysis of a user's sleep state in stages, the existing error will affect the accuracy of the judgment results, and thus affect the assessment of the user's sleep health status. Therefore, how to more accurately assess a user's sleep health status is the fundamental problem that this invention aims to solve. Summary of the Invention
[0005] The purpose of this invention is to provide a sleep health assessment and management method and system based on AI and sensor data, to solve the following technical problems: How to more accurately assess a user's sleep health status.
[0006] The objective of this invention can be achieved through the following technical solutions: A sleep health assessment and management method based on AI and sensor data, the method comprising: By continuously transmitting frequency-modulated continuous waves and receiving human body reflection signals through millimeter-wave radar, and analyzing the human body reflection signals based on AI models, the first sleep data of the person can be obtained. Brainwave data of a person is obtained through electroencephalogram (EEG) electrodes, and second sleep data of the person is obtained based on the brainwave data. Physiological parameter data of personnel is acquired through a sensor array, and third sleep data of personnel is obtained based on the physiological parameter data; A user sleep model is established based on the differences between the first sleep data, the second sleep data, and the third sleep data. The user's sleep health status is then assessed based on the user sleep model.
[0007] Furthermore, the process of acquiring the AI model includes: Acquire sample data and perform standard data annotation and anomaly event annotation to obtain an annotated sample; The labeled samples are preprocessed, including obtaining the original ADC signal, performing range-dimensional FFT, Doppler-dimensional FFT, removing static clutter, and obtaining the time-frequency spectrum. Spatial patterns in the temporal spectrogram are extracted using CNN branches, and long-term dependencies are captured using BiLSTM branches; a loss function is established and trained, the trained model is optimized and deployed, and an AI model is obtained. The process of acquiring the second sleep data includes: The user's sleep stage is determined based on the waveform type in the brainwave data, and secondary sleep data is determined based on the changes in the sleep stage during the user's sleep process.
[0008] Furthermore, the process of acquiring third sleep data includes: Obtain the blood pressure curve bp(t), heart rate curve H(t), blood oxygen curve bo(t), and body temperature curve T(t) from the personnel's physiological parameter data, and then use the model:
[0009]
[0010] The Rem stage matching coefficient R is calculated and compared with the Rem stage critical value Rc. When R≥Rc, the user's sleep state is determined to be in the Rem stage; otherwise, the user's sleep depth stage is determined based on the blood pressure curve, heart rate curve and body temperature curve, and it is used as the third sleep data. in, and These are the average blood oxygen and average blood pressure values of the user during the waking phase, respectively. Let i be the ambient temperature, m be the number of time points selected at fixed time intervals before the current time point t, and i ∈ [1, m]. For the i-th time point, The threshold for heart rate variability. , , and These are the unit quantities representing the influence of heart rate, blood oxygen, blood pressure, and body temperature, respectively.
[0011] Furthermore, the process of determining the user's sleep depth stage based on blood pressure, heart rate, and body temperature curves includes: Through the model:
[0012] The user's sleep depth coefficient Dp(t) is calculated and compared with a preset threshold range. The corresponding user sleep depth stage is determined based on the preset threshold range in which the user's sleep depth coefficient Dp(t) is located. in, and These are the average heart rate and average body temperature of the user during the waking phase, respectively. and This is the proportional adjustment coefficient.
[0013] Furthermore, the process of establishing the user sleep model includes: Values are assigned to different user sleep depth stages and Rem stages, and curves S1(t), S2(t), and S3(t) of the first sleep state over time are obtained based on the first sleep data, the second sleep data, and the third sleep data, respectively. Obtain the area enclosed between any two curves S1(t), S2(t), and S3(t) within the preset analysis time period. , and ,Will , and Each area is compared with a preset area threshold St: when , and When all values are less than the preset area threshold St, the average of the time points when the sleep state changes in the first sleep data, the second sleep data, and the third sleep data is selected as the time point when the state changes. when , and If only one value is less than the preset area threshold St, then the average of the time points when the sleep state changes in the two sets of sleep data corresponding to that value is selected as the time point when the state changes. Otherwise, the acquisition process of the first sleep data, second sleep data, and third sleep data should be checked; The time points of state changes and different stages of sleep are used as the user's sleep model.
[0014] Furthermore, the process of assessing a user's sleep health status based on the user's sleep model includes: The user's sleep state and duration at different time points are obtained based on the user's sleep model. The user's sleep state and duration at different time points are evaluated based on the preset evaluation algorithm to obtain the evaluation value. The user's sleep health is judged based on the evaluation value.
[0015] Furthermore, the method also includes: Multiple user sleep models under different environmental conditions were obtained, and the user sleep model with the best evaluation result was selected and its corresponding environmental parameters were obtained. Through the model Calculate the environmental deviation coefficient E corresponding to the user's sleep model under non-optimal conditions; Calculate the correlation coefficient between the environmental deviation coefficient E of the user's sleep model under non-optimal conditions and the evaluation value. Compare the correlation coefficient with a preset threshold. When the correlation coefficient is greater than or equal to the preset threshold, the user's sleep is judged to be easily disturbed. The environmental parameters corresponding to the optimal evaluation result of the user's sleep model are used as recommended environmental parameters. Where n is the number of environmental parameters, k∈[1,n], For the k-th environmental parameter, The k-th environmental parameter is the one whose evaluation result in the user's sleep model is optimal. This is the dimensionless unit value of the k-th environmental parameter. This represents the influence coefficient of the k-th environmental parameter.
[0016] Furthermore, the calculation process of the correlation coefficient includes: Through the model The correlation coefficient Z was calculated. Where h is the number of user sleep models in the non-optimal state, x∈[1,h], Let x be the environmental deviation coefficient of the sleep model for the x-th user. Let h be the mean of the environmental deviation coefficients. Let x be the evaluation value of the sleep model for the xth user. The mean of h evaluation values.
[0017] A sleep health assessment and management system based on AI and sensor data, the system comprising: millimeter-wave radar, EEG electrodes, sensor array and assessment center; The millimeter-wave radar is used to continuously transmit frequency-modulated continuous waves and receive human body reflection signals. Based on an AI model, the human body reflection signals are analyzed to obtain the first sleep data of the person. The EEG electrodes are used to acquire brainwave data of a person, and to acquire the person's second sleep data based on the brainwave data. The sensor group is used to acquire physiological parameter data of personnel, and to acquire the third sleep data of personnel based on the physiological parameter data; The assessment center is used to establish a user sleep model based on the difference between the first sleep data, the second sleep data, and the third sleep data, and to assess the user's sleep health status based on the user sleep model.
[0018] The beneficial effects of this invention are: (1) The present invention can determine the user's sleep state based on physiological parameters, and at the same time, it uses three methods to monitor the user's sleep state in a coordinated manner, which can eliminate the influence of error data and improve the accuracy of the user's sleep model judgment.
[0019] (2) The present invention can make a judgment on the environmental disturbance state of the user's sleep state, and provide more useful information for users with sleep disorders. Attached Figure Description
[0020] The invention will now be further described with reference to the accompanying drawings.
[0021] Figure 1 This is a flowchart of the sleep health assessment and management method of the present invention; Figure 2 This is a logic block diagram of the sleep health assessment and management system of the present invention. Detailed Implementation
[0022] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0023] Please see Figure 1 As shown, in one embodiment, a sleep health assessment and management method based on AI and sensor data is provided. Please refer to [link to relevant documentation]. Figure 1 As shown, the method includes: First, a millimeter-wave radar continuously transmits frequency-modulated continuous waves and receives reflected signals from the human body. Based on an AI model, the reflected signals are analyzed to obtain the person's first sleep data. Next, brainwave data is acquired using EEG electrodes, and the second sleep data is obtained based on this brainwave data. Finally, physiological parameter data is acquired using a sensor array, and the third sleep data is obtained based on this physiological parameter data. A user sleep model is then established based on the differences between the first, second, and third sleep data, and the user's sleep health status is assessed based on this user sleep model.
[0024] As can be seen from the above scheme, this embodiment uses three methods to monitor the user's sleep state in a coordinated manner. Among them, by continuously transmitting frequency-modulated continuous waves and receiving human body reflection signals through millimeter-wave radar, it is possible to acquire micro-motion data of the human body in the sleep state. Based on the AI model, the micro-motion data is analyzed to determine the user's sleep stage, thereby achieving the acquisition of the first sleep data.
[0025] The AI model acquisition process includes: acquiring sample data and performing standard data annotation and abnormal event annotation to obtain annotated samples; the sample acquisition process is the most difficult part of this step, which is mainly obtained through real sleep radar data of volunteers, and at the same time, clinical positive samples are established through data augmentation to achieve sample data acquisition. Then, the annotated samples are preprocessed, including acquiring the original ADC signal, distance dimension FFT (separating objects at different distances), Doppler dimension FFT (extracting respiratory / heartbeat micro-motions) and removing static clutter to obtain the time spectrum; extracting spatial patterns in the time spectrum based on CNN branches and capturing long-term dependencies based on BiLSTM branches; establishing a loss function and training, optimizing and deploying the trained model to obtain the AI model; in the above scheme, the AI model establishment process is existing technology and will not be described in detail here.
[0026] In addition, the process of acquiring second sleep data includes: determining the user's sleep stage based on the waveform type in the brainwave data, where the waveform type includes theta waves, delta waves, alpha waves, and beta waves, etc., and determining the second sleep data based on the changes in the user's sleep stages during sleep; the process of acquiring third sleep data includes: acquiring the blood pressure curve bp(t), heart rate curve H(t), blood oxygen curve bo(t), and body temperature curve T(t) from the person's physiological parameter data, and using a model:
[0027]
[0028] The matching coefficient R of the Rem stage is calculated, where, and These are the average blood oxygen and average blood pressure values of the user during the waking phase, respectively. Let i be the ambient temperature, m be the number of time points selected at fixed time intervals before the current time point t, and i ∈ [1, m]. For the i-th time point, The threshold for heart rate variability. , , and These are unit quantities representing the influence of heart rate, blood oxygen, blood pressure, and body temperature, among which the heart rate variability threshold is included. The influence unit quantities of heart rate, blood oxygen, blood pressure, and body temperature were obtained by fitting laboratory data. , , and The parameters are set comprehensively based on their numerical range, dimensions, and degree of influence. Therefore, through... , , and It can remove dimensions and normalize data, while also assigning weights to each parameter, thereby achieving a comprehensive matching process of multiple factors for the Rem stage.
[0029] Because heart rate fluctuates dramatically during the Rem stage of sleep, the body experiences hypothermia, blood oxygen saturation decreases, and blood pressure approaches the level of wakefulness, this embodiment establishes a Rem stage matching coefficient R model based on the characteristics of changes in human physiological parameters during the Rem stage. The more dramatic the heart rate fluctuations, the closer the body temperature is to the ambient temperature, the closer the blood pressure is to that of the wakefulness state, and the more significant the decrease in blood oxygen, the higher the Rem stage matching degree, i.e., the larger the Rem stage matching coefficient R. Therefore, by comparing R with the Rem stage critical value Rc, which is obtained by fitting empirical data, when R ≥ Rc, the user's sleep state is determined to be in the Rem stage; otherwise, the user's sleep depth stage is determined based on the blood pressure curve, heart rate curve, and body temperature curve.
[0030] In one embodiment, the specific process of determining a user's sleep depth stage includes: using a model: The user's sleep depth coefficient Dp(t) is calculated, where, and These are the average heart rate and average body temperature of the user during the waking phase, respectively. and The proportional adjustment coefficient is set based on the degree of influence of different parameters in empirical data. Since human heart rate, body temperature, and blood pressure decrease with increasing sleep depth during non-REM sleep, the total user sleep depth coefficient model obtains the user sleep depth coefficient Dp(t). That is, sleep depth is determined based on the degree of decrease in human heart rate, body temperature, and blood pressure. The user sleep depth coefficient Dp(t) is compared with a preset threshold range. The preset threshold range is obtained based on test data and corresponds to different sleep depth stages. Therefore, the corresponding user sleep depth stage can be determined based on the preset threshold range in which the user sleep depth coefficient Dp(t) is located. Through the above process, the user's sleep state can be determined based on physiological parameters, thereby realizing the acquisition of third sleep data. It should be noted that the health monitoring of the above user physiological parameters will be completed in advance through an additional analysis model. Therefore, the physiological parameters appearing in this embodiment are all within a reasonable range.
[0031] In addition, the process of establishing the user sleep model includes: firstly, assigning values to different user sleep depth stages and the Rem stage. For example, N1 light sleep stage is assigned a value of 1, N2 medium sleep stage is assigned a value of 2, N3 deep sleep stage is assigned a value of 3, and the Rem stage is assigned a value of 0. It should be noted that the assignment has no practical meaning and is only used to distinguish different sleep stages. Based on the first sleep data, second sleep data, and third sleep data, obtain the time-varying curves S1(t), S2(t), and S3(t) of the first sleep state, respectively. Then, obtain the area enclosed between any two curves of S1(t), S2(t), and S3(t) within the preset analysis period. , and ,Will , and Each area is compared with a preset area threshold St, which is set based on error data from empirical data. Therefore, when... , and When all three sleep data points are less than the preset area threshold St, it indicates that the three sets of sleep data are relatively consistent. Therefore, selecting the average of the sleep state change time points in the first, second, and third sleep data as the state change time points can improve the accuracy of the assessment results. , and If only one value is less than the preset area threshold St, it indicates that one set of sleep data deviates from the other two sets. Therefore, the average of the time points when the sleep state changes in the two sets of sleep data corresponding to this value is selected as the state change time point. This can eliminate the influence of erroneous data and improve the accuracy of the user sleep model. Otherwise, it indicates that there are significant differences among the three sets of sleep data. This problem is relatively rare and mainly occurs during the data collection process. Therefore, the acquisition process of the first, second, and third sleep data is checked to avoid inaccurate judgment results caused by abnormal data collection. The acquired state change time points and sleep states at different stages are used as the user sleep model. Based on the user sleep model, the user's sleep state and duration at different time points are obtained. Based on the preset evaluation algorithm, the user's sleep state and duration at different time points are evaluated to obtain the evaluation value. The preset evaluation algorithm can adopt existing technology and mainly judges the proportion of different sleep stages, the latency of the REM stage, etc., which will not be detailed here. The user's sleep health is judged based on the evaluation value, which can more accurately achieve the judgment of the user's sleep state.
[0032] In one embodiment, the sleep health assessment and management method further includes: acquiring multiple sets of user sleep models under different environmental conditions, selecting the user sleep model with the best assessment result and acquiring its corresponding environmental parameters, and then using the model... The environmental deviation coefficient E corresponding to the user's sleep model under non-optimal conditions is calculated; where n is the number of environmental parameters, and the environmental parameters collected in this embodiment mainly include temperature and humidity, k∈[1,n]. For the k-th environmental parameter, The k-th environmental parameter is the one whose evaluation result in the user's sleep model is optimal. This is the dimensionless unit value of the k-th environmental parameter. Let be the influence coefficient of the k-th environmental parameter, where the dimensionless unit value is set according to the unit of the k-th environmental parameter. Based on the influence of corresponding environmental parameters on sleep state from empirical data, the correlation coefficient between the environmental deviation coefficient E and the evaluation value of the user's sleep model under non-optimal conditions is calculated. This process includes: using the model... The correlation coefficient Z is calculated; where h is the number of user sleep models under non-optimal conditions, and x∈[1,h]. Let x be the environmental deviation coefficient of the sleep model for the x-th user. Let h be the mean of the environmental deviation coefficients. Let x be the evaluation value of the sleep model for the xth user. The correlation coefficient Z is the mean of h evaluation values. Therefore, the closer the correlation coefficient Z is to 1, the stronger the correlation. The correlation coefficient is compared with a preset threshold, which is less than 1 and greater than 0.7. The specific value is selected based on empirical data. When the correlation coefficient is greater than or equal to the preset threshold, it is considered a high correlation, thus the user's sleep is judged to be easily disturbed. The environmental parameters corresponding to the optimal evaluation result of the user's sleep model are used as recommended environmental parameters. Through the above process, the environmental disturbance status of the user's sleep state can be judged, providing more useful information for users with sleep disorders. It should be noted that the above multiple sets of different environmental states are all within the normal range of sleep environment parameters.
[0033] In one embodiment, a sleep health assessment and management system based on AI and sensor data is provided; please refer to the appendix. Figure 2As shown, the system includes: millimeter-wave radar, EEG electrodes, a sensor array, and an assessment center. The millimeter-wave radar continuously transmits frequency-modulated continuous waves and receives reflected signals from the human body. Based on an AI model, it analyzes these reflected signals to obtain the user's first sleep data. The EEG electrodes acquire the user's brainwave data and, based on this, acquire the user's second sleep data. The sensor array acquires the user's physiological parameter data and, based on this, acquires the user's third sleep data. The assessment center establishes a user sleep model based on the differences between the first, second, and third sleep data and assesses the user's sleep health status based on this model.
[0034] The foregoing has provided a detailed description of one embodiment of the present invention, but this description is merely a preferred embodiment and should not be construed as limiting the scope of the invention. All equivalent variations and modifications made within the scope of the claims of this invention should still fall within the patent coverage of this invention.
Claims
1. An AI and sensor data-based sleep health assessment management method, characterized by, The method includes: By continuously transmitting frequency-modulated continuous waves and receiving human body reflection signals through millimeter-wave radar, and analyzing the human body reflection signals based on AI models, the first sleep data of the person can be obtained. Brainwave data of a person is obtained through electroencephalogram (EEG) electrodes, and second sleep data of the person is obtained based on the brainwave data. Physiological parameter data of personnel is acquired through a sensor array, and third sleep data of personnel is obtained based on the physiological parameter data; A user sleep model is established based on the differences between the first sleep data, the second sleep data, and the third sleep data. The user's sleep health status is then assessed based on the user sleep model.
2. The sleep health assessment and management method based on AI and sensor data according to claim 1, characterized in that, The process of acquiring the AI model includes: Acquire sample data and perform standard data annotation and anomaly event annotation to obtain an annotated sample; The labeled samples are preprocessed, including obtaining the original ADC signal, performing range-dimensional FFT, Doppler-dimensional FFT, removing static clutter, and obtaining the time-frequency spectrum. Spatial patterns in the temporal spectrogram are extracted using CNN branches, and long-term dependencies are captured using BiLSTM branches; a loss function is established and trained, the trained model is optimized and deployed, and an AI model is obtained. The process of acquiring the second sleep data includes: The user's sleep stage is determined based on the waveform type in the brainwave data, and secondary sleep data is determined based on the changes in the sleep stage during the user's sleep process.
3. The sleep health assessment and management method based on AI and sensor data according to claim 2, characterized in that, The process of acquiring third-sleep data includes: Obtain the blood pressure curve bp(t), heart rate curve H(t), blood oxygen curve bo(t), and body temperature curve T(t) from the personnel's physiological parameter data, and then use the model: ； ； The Rem stage matching coefficient R is calculated and compared with the Rem stage critical value Rc. When R≥Rc, the user's sleep state is determined to be in the Rem stage; otherwise, the user's sleep depth stage is determined based on the blood pressure curve, heart rate curve and body temperature curve, and it is used as the third sleep data. in, and These are the average blood oxygen and blood pressure values of the user during the waking phase, respectively. Let i be the ambient temperature, m be the number of time points selected at fixed time intervals before the current time point t, and i ∈ [1, m]. For the i-th time point, The threshold for heart rate variability. , , and These are the unit quantities representing the influence of heart rate, blood oxygen, blood pressure, and body temperature, respectively.
4. The sleep health assessment and management method based on AI and sensor data according to claim 3, characterized in that, The process of determining a user's sleep depth stage based on blood pressure, heart rate, and body temperature curves includes: Through the model: ； The user's sleep depth coefficient Dp(t) is calculated and compared with a preset threshold range. The corresponding user sleep depth stage is determined based on the preset threshold range in which the user's sleep depth coefficient Dp(t) is located. in, and These are the average heart rate and average body temperature of the user during the waking phase, respectively. and This is the proportional adjustment coefficient.
5. The sleep health assessment and management method based on AI and sensor data according to claim 4, characterized in that, The process of establishing the user sleep model includes: Values are assigned to different user sleep depth stages and Rem stages, and curves S1(t), S2(t), and S3(t) of the first sleep state over time are obtained based on the first sleep data, the second sleep data, and the third sleep data, respectively. Obtain the area enclosed between any two curves among S1(t), S2(t), and S3(t) within the preset analysis time period. , and ,Will , and Each area is compared with a preset area threshold St: when , and When all values are less than the preset area threshold St, the average of the time points when the sleep state changes in the first sleep data, the second sleep data, and the third sleep data is selected as the state change time point. when , and If only one value is less than the preset area threshold St, then the average of the time points when the sleep state changes in the two sets of sleep data corresponding to that value is selected as the time point when the state changes. Otherwise, the acquisition process of the first sleep data, second sleep data, and third sleep data should be checked; The time points of state changes and different stages of sleep are used as the user's sleep model.
6. The sleep health assessment and management method based on AI and sensor data according to claim 5, characterized in that, The process of assessing a user's sleep health status based on a user sleep model includes: The user's sleep state and duration at different time points are obtained based on the user's sleep model. The user's sleep state and duration at different time points are evaluated based on the preset evaluation algorithm to obtain the evaluation value. The user's sleep health is judged based on the evaluation value.
7. The sleep health assessment and management method based on AI and sensor data according to claim 5, characterized in that, The method further includes: Multiple user sleep models under different environmental conditions were obtained, and the user sleep model with the best evaluation result was selected and its corresponding environmental parameters were obtained. Through the model Calculate the environmental deviation coefficient E corresponding to the user's sleep model under non-optimal conditions; Calculate the correlation coefficient between the environmental deviation coefficient E of the user's sleep model under non-optimal conditions and the evaluation value. Compare the correlation coefficient with a preset threshold. When the correlation coefficient is greater than or equal to the preset threshold, the user's sleep is judged to be easily disturbed. The environmental parameters corresponding to the optimal evaluation result of the user's sleep model are used as recommended environmental parameters. Where n is the number of environmental parameters, k∈[1,n], For the k-th environmental parameter, The k-th environmental parameter is the one whose evaluation result in the user's sleep model is optimal. This is the dimensionless unit value of the k-th environmental parameter. This represents the influence coefficient of the k-th environmental parameter.
8. The sleep health assessment and management method based on AI and sensor data according to claim 7, characterized in that, The calculation process of the correlation coefficient includes: Through the model The correlation coefficient Z was calculated. Where h is the number of user sleep models in the non-optimal state, x∈[1,h], Let x be the environmental deviation coefficient of the sleep model for the x-th user. Let h be the mean of the environmental deviation coefficients. Let x be the evaluation value of the sleep model for the xth user. The mean of h evaluation values.
9. A sleep health assessment and management system based on AI and sensor data, characterized in that, The system employs a sleep health assessment and management method based on AI and sensor data as described in any one of claims 1-8, comprising: millimeter-wave radar, EEG electrodes, sensor array, and assessment center; The millimeter-wave radar is used to continuously transmit frequency-modulated continuous waves and receive human body reflection signals. Based on an AI model, the human body reflection signals are analyzed to obtain the first sleep data of the person. The EEG electrodes are used to acquire brainwave data of a person, and to acquire the person's second sleep data based on the brainwave data. The sensor group is used to acquire physiological parameter data of personnel, and to acquire the third sleep data of personnel based on the physiological parameter data; The assessment center is used to establish a user sleep model based on the difference between the first sleep data, the second sleep data, and the third sleep data, and to assess the user's sleep health status based on the user sleep model.