Traditional chinese medicine sleep health monitoring system based on millimeter wave radar
By using millimeter-wave radar to collect sleep data non-contactly and combining it with traditional Chinese medicine theory, the invasiveness and data fusion problems of existing technologies have been solved. This enables seamless and continuous sleep monitoring that integrates with traditional Chinese medicine diagnosis, forming a closed-loop management system and improving the systematicness and accuracy of sleep health management.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- SHANGHAI UNIV OF T C M
- Filing Date
- 2026-03-31
- Publication Date
- 2026-06-09
AI Technical Summary
Existing sleep monitoring technologies are highly invasive, making it impossible to achieve long-term, continuous, and comfortable home monitoring. Furthermore, modern physiological data lacks effective integration with traditional Chinese medicine theories, and TCM assessments lack quantitative basis. The monitoring and intervention processes are disconnected, making it difficult to achieve personalized health management.
Using millimeter-wave radar for non-contact data acquisition, combined with the dynamic syndrome differentiation theory of traditional Chinese medicine, physiological parameters are transformed into quantitative indicators of traditional Chinese medicine through a quantitative mapping unit, a spatiotemporal correlation database is established, traditional Chinese medicine syndrome differentiation labels are generated, and analysis and decision-making are carried out through machine learning models to form a closed-loop health management system.
It achieves seamless and continuous sleep monitoring combined with objective and quantitative TCM syndrome differentiation, providing systematic sleep health management, supporting early warning and continuous conditioning, and improving the systematicness and accuracy of management.
Smart Images
Figure CN122163185A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the fields of intelligent health monitoring and digital TCM technology, specifically to a TCM sleep health monitoring system based on millimeter-wave radar. Background Technology
[0002] Sleep health is an important indicator of overall human health. Among existing sleep monitoring technologies, polysomnography (PSG) is cumbersome to use, disrupts sleep, and is not suitable for home use; its data also lacks correlation with the overall state of the body according to Traditional Chinese Medicine (TCM). While consumer-grade wearable devices are convenient for home use, they still require wearing and have limited accuracy, failing to provide in-depth health insights. Furthermore, TCM assessments primarily rely on subjective, momentary consultations, lacking continuous, objective nighttime data as a basis for diagnosis, making quantitative assessment and closed-loop management difficult. Therefore, the current field mainly faces the following four problems: First, existing monitoring technologies have invasive limitations in data collection methods. Whether it is a professional PSG device or a consumer-grade wearable device, they all need to be in direct contact with the body, which is either cumbersome to wear or restrictive for a long time. Their physical invasiveness not only interferes with the user's natural sleep state and affects the authenticity and ecological validity of the data, but also limits the feasibility of long-term, continuous and comfortable monitoring in the home environment, making it difficult to obtain long-term dynamic data that reflects the true health status.
[0003] Secondly, the physiological parameters collected by existing technologies mainly serve modern medical analysis. The core parameters output, such as sleep staging based on EEG and heart rate monitoring based on optics, are derived from the modern physiological framework and cannot be directly converted into the core diagnostic elements required by traditional Chinese medicine assessment, such as the circulation of qi and blood, the waxing and waning of yin and yang, and the functions of the internal organs. This results in a lack of effective integration between modern sensor data and traditional Chinese medicine theoretical models, making it difficult for massive amounts of physiological data to directly serve the personalized constitution identification and syndrome analysis of traditional Chinese medicine.
[0004] Third, in the traditional Chinese medicine health assessment process, doctors mainly rely on the patient's subjective complaints during daytime visits and the instantaneous pulse and tongue appearance to make judgments. They completely lack continuous and objective records of the key physiological and pathological process of nighttime sleep. This data blind spot makes it difficult to make quantitative basis for diagnosis, and it is impossible to accurately locate the dynamic pattern and severity of abnormalities, which restricts the pertinence and precision of intervention measures.
[0005] Fourth, the monitoring, evaluation, intervention, and verification processes in the existing technology combination are fragmented and form an open-loop process. There is a lack of a systematic platform that can automatically connect "non-intrusive continuous monitoring - dynamic objective analysis - personalized intervention generation - quantitative feedback of intervention effect". As a result, health management cannot achieve dynamic optimization and personalized iteration based on objective data, and its effects are difficult to scientifically verify and continuously track. Summary of the Invention
[0006] To address the problems existing in current technologies, this invention provides a Traditional Chinese Medicine (TCM) sleep health monitoring system based on millimeter-wave radar. The system includes: a data acquisition module for acquiring raw vital sign signals via millimeter-wave radar; a data processing module for extracting a set of sleep physiological parameters; a sleep stage identification module for identifying sleep stages corresponding to wakefulness, light sleep, deep sleep, and REM sleep; a TCM sleep assessment module for converting sleep parameters into TCM quantitative indicators through a quantification mapping unit, generating TCM syndrome differentiation labels through a spatiotemporal correlation unit, and then performing fusion analysis through an analysis and decision-making unit to output TCM health monitoring results; and a monitoring result output module for generating a visualized health report. This invention achieves a deep integration of seamless, continuous sleep monitoring with objective, quantitative TCM syndrome differentiation, providing a systematic solution for sleep health management.
[0007] This invention adopts the following technical solution: a traditional Chinese medicine sleep health monitoring system based on millimeter-wave radar, comprising: The data acquisition module is used to deploy millimeter-wave radar sensors in a set manner and to collect raw vital signs signals of the target using millimeter-wave radar sensors. The data processing module is used to receive the raw vital signs signals and perform preprocessing operations to output a set of sleep physiological parameters; the set of sleep physiological parameters includes at least: respiratory rate, heart rate, and body movement rate; The sleep stage identification module is used to receive the set of sleep physiological parameters, identify the sleep stages based on a preset integrated classification model, and output the sleep stage results; the sleep stage results include: sleep stage corresponding to the wakefulness stage, light sleep stage, deep sleep stage, and REM sleep stage respectively; The TCM sleep assessment module includes a quantitative mapping unit, a spatiotemporal correlation unit, and an analysis and decision-making unit; The quantitative mapping unit is used to convert the sleep staging results and the set of sleep physiological parameters into TCM quantitative indicators according to preset TCM theory mapping rules. The spatiotemporal association unit is used to establish a three-level mapping database of time, meridians, and internal organs based on the theory of meridian flow in traditional Chinese medicine. Based on the sleep stage results and their corresponding detection times, the corresponding organ differentiation rules are queried in the three-level mapping database to generate corresponding TCM differentiation tags. The analysis and decision-making unit takes the TCM quantitative indicators and the TCM syndrome differentiation labels as inputs and uses a pre-trained machine learning model to generate the TCM health monitoring results of the target. The monitoring results output module is used to integrate the TCM health monitoring results and generate a visualized health report.
[0008] Furthermore, the data acquisition module deploys a millimeter-wave radar sensor in a predetermined manner, specifically: The millimeter-wave radar sensor is deployed under the mattress, and the installation position is perpendicular to the mattress surface and points towards the user's sleeping position; The main lobe of the millimeter-wave signal emitted by the millimeter-wave radar sensor covers the core area of the torso when the target is lying supine on the mattress surface.
[0009] Furthermore, the data processing module receives the original vital sign signal and performs preprocessing operations, including: a filtering unit, a phase demodulation unit, a blind source separation unit, and a feature extraction unit; The filtering unit is used to filter the original vital signs signal using a digital filter; The phase demodulation unit is used to demodulate the phase components of the filtered original vital signs signal and extract the micro-displacement signal related to the chest cavity displacement. The blind source separation unit is used to separate the respiratory signal component and the heartbeat signal component in the micro-motion displacement signal using a blind source separation algorithm. The feature extraction unit is used to perform time-frequency domain analysis on the respiratory signal component and the heartbeat signal component to extract the sleep physiological parameters that constitute the sleep physiological parameter set.
[0010] Furthermore, in the sleep stage identification module, the process of identifying sleep stages based on a preset integrated classification model of the sleep physiological parameter set is as follows: Based on the set of sleep physiological parameters, a multidimensional feature vector is extracted for input into a preset integrated classification model; the multidimensional feature vector includes respiratory features, heart rate features, body movement features, and cross-modulation features. The preset integrated classification model is composed of at least two of the following: long short-term memory neural network, support vector machine, and random forest classifier; The multidimensional feature vectors are input into the ensemble classification model for parallel reasoning and fusion decision-making, and the sleep stage results corresponding to each sleep stage information are output.
[0011] Furthermore, the quantification mapping unit is used to convert the sleep staging results and the sleep physiological parameter set into TCM quantitative indicators according to preset TCM theoretical mapping rules, including: Features related to deep sleep, rapid eye movement (REM) sleep, and wakefulness are extracted from the sleep stage results and the set of sleep physiological parameters. According to the preset TCM theory mapping rules, the deep sleep-related features are mapped to the first TCM quantitative index representing the storage of Yin energy. The rapid eye movement phase-related features are mapped to a second TCM quantitative index that characterizes the degree of yang entering yin and the smoothness of yin-yang interaction. The aforementioned awakening-related characteristics are mapped to a third TCM quantitative indicator characterizing the level of abnormal Wei Qi circulation or restlessness of the mind.
[0012] Furthermore, features related to deep sleep, REM sleep, and wakefulness are extracted from the sleep stage results and the set of sleep physiological parameters, including: The relevant characteristics of deep sleep include the total duration of deep sleep and the average duration of a single episode. The rapid eye movement (REM) related characteristics include the regularity of periodic occurrence, the average period length, and the proportion of REM activity in total sleep duration. The relevant characteristics of the awakening period include the number of awakenings, the average duration, and the frequency of awakenings in the second half of the night.
[0013] Furthermore, in the aforementioned spatiotemporal correlation unit, based on the theory of meridian flow in Traditional Chinese Medicine, a three-level mapping database of time-meridian-organs is established, specifically as follows: Based on the theory of meridian flow in traditional Chinese medicine, the 24-hour cycle is divided into twelve consecutive preset time intervals. For each preset time interval, the corresponding dominant human meridians and core organs are configured according to traditional Chinese medicine theory; For each preset time interval, a differentiation rule for the viscera and bowels is established. The differentiation rule for the viscera and bowels is as follows: Multiple abnormal conditions are set based on the sleep stage results in each preset time interval, and the functional status of the dominant human meridians and core organs in each preset time interval is obtained under each abnormal condition. When the sleep stage results meet the set abnormal conditions within a preset time interval, the functional state of the human meridians and core organs dominated by that preset time interval is used as the corresponding TCM syndrome label. A three-level mapping database of time-meridian-organ is constructed based on the organ differentiation rules of all preset time intervals.
[0014] Furthermore, in the spatiotemporal correlation unit, the corresponding organ differentiation rules are queried in the three-level mapping database to generate corresponding TCM differentiation tags, including: Obtain the sleep stage results and their corresponding detection times, and map the detection times to the corresponding specific time intervals according to the theory of meridian flow in traditional Chinese medicine. The system queries the organ differentiation rules corresponding to the specific time interval in the time-meridian-organ three-level mapping database and matches the sleep stage results with the organ differentiation rules corresponding to the characteristic time interval. Output the corresponding TCM syndrome differentiation labels based on the matching results.
[0015] The beneficial effects of this invention are as follows: By integrating millimeter-wave radar non-contact sensing technology with the dynamic syndrome differentiation theory of traditional Chinese medicine, this invention achieves an innovative breakthrough in the paradigm of sleep health monitoring. It acquires continuous, multi-dimensional raw vital sign signal data in a non-contact manner and innovatively constructs a mapping process from parameters such as respiration, heart rate, and body movement to TCM phenomena, thereby transforming static sleep quality scores into dynamic identification of the spatiotemporal evolution of sleep-syndrome patterns. This not only provides a new, objective, and quantitative method for TCM sleep diagnosis but also forms a precise health management closed loop from monitoring to syndrome differentiation to intervention and finally effect feedback. This effectively supports early warning and continuous conditioning guided by disease prevention, and significantly improves the systematicness, accuracy, and practicality of sleep health management. Attached Figure Description
[0016] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0017] Figure 1 This is a schematic diagram of the structure of a traditional Chinese medicine sleep health monitoring system based on millimeter-wave radar according to an embodiment of the present invention; Figure 2 This is a schematic diagram illustrating the deployment method of millimeter-wave radar in a data acquisition module according to an embodiment of the present invention; Figure 3 This is a partially enlarged schematic diagram of a millimeter-wave radar deployment method according to an embodiment of the present invention; Figure 4 This is a schematic diagram illustrating the technical principle of a traditional Chinese medicine sleep health monitoring system according to an embodiment of the present invention.
[0018] In the diagram, 1. External power supply; 2. Target; 3. Mattress; 4. Visualization terminal; 5. Millimeter-wave radar sensor. Detailed Implementation
[0019] 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.
[0020] A schematic diagram of the TCM sleep health monitoring system based on millimeter-wave radar according to an embodiment of the present invention is shown below. Figure 1 As shown, it includes: The data acquisition module is used to deploy millimeter-wave radar sensors in a set manner and to collect raw vital signs signals of the target using millimeter-wave radar sensors. The data processing module is used to receive the raw vital signs signals and perform preprocessing operations to output a set of sleep physiological parameters; the set of sleep physiological parameters includes at least: respiratory rate, heart rate, and body movement rate; The sleep stage identification module is used to receive the set of sleep physiological parameters, identify the sleep stages based on a preset integrated classification model, and output the sleep stage results; the sleep stage results include: sleep stage corresponding to the wakefulness stage, light sleep stage, deep sleep stage, and REM sleep stage respectively; The TCM sleep assessment module includes a quantitative mapping unit, a spatiotemporal correlation unit, and an analysis and decision-making unit; The quantitative mapping unit is used to convert the sleep staging results and the set of sleep physiological parameters into TCM quantitative indicators according to preset TCM theory mapping rules. The spatiotemporal association unit is used to establish a three-level mapping database of time, meridians, and internal organs based on the theory of meridian flow in traditional Chinese medicine. Based on the sleep stage results and their corresponding detection times, the corresponding organ differentiation rules are queried in the three-level mapping database to generate corresponding TCM differentiation tags. The analysis and decision-making unit takes the TCM quantitative indicators and the TCM syndrome differentiation labels as inputs and uses a pre-trained machine learning model to generate the TCM health monitoring results of the target. The monitoring results output module is used to integrate the TCM health monitoring results and generate a visualized health report.
[0021] In a specific embodiment of the present invention, a schematic diagram of the health monitoring principle of the system of the present invention is shown below. Figure 4 As shown, the details are as follows: The data acquisition module deploys millimeter-wave radar sensors in a predefined manner to collect raw vital signs signals of the target. In embodiments of the present invention, such as Figure 2 and Figure 3 As shown, the millimeter-wave radar sensor 5 is deployed under the mattress 3 and connected to an external power supply 1 via a USB power cable. The sensor is installed perpendicular to the mattress surface, pointing towards the target's sleeping position. Specifically, it is placed horizontally under the mattress. During installation, it is necessary to ensure that the main beam direction of the millimeter-wave radar sensor antenna faces upwards towards the mattress, corresponding to the direction perpendicular to the mattress surface and pointing towards the user's sleeping position. The main lobe range of the frequency-modulated continuous millimeter-wave signal emitted can be adjusted to completely cover the area where the target's torso is located when lying supine. In other words, the main lobe range of the millimeter-wave signal emitted by the millimeter-wave radar sensor covers the core area of the target's torso when lying supine on the mattress surface. Figure 2 The measurement area is shown in the figure; at the same time, the signal sent by the millimeter-wave radar sensor is configured as a frequency modulated continuous wave (FMCW) millimeter-wave signal, and the echo reflected by the periodic undulation of the human chest cavity and the micro-movement of the body surface is received simultaneously; by mixing and analog-to-digital conversion of the echo signal, the original vital sign signal stream containing modulation information of various vital signs such as breathing, heartbeat and body movement is obtained.
[0022] In one specific embodiment of the present invention, the parameters of the millimeter-wave radar sensor are set to 60-64Hz, with a sampling rate greater than or equal to 100Hz, to meet the requirements for extracting heartbeat signals. The millimeter-wave radar sensor can be placed approximately 10 cm below the mattress to ensure that the 3dB main lobe width of its emitted beam can completely cover the core area of the target's torso. This core area can be a rectangular region approximately 80 cm long and 50 cm wide, centered on the chest and abdomen of the target; or it can be... Figure 2 The elliptical region shown.
[0023] The data processing module receives raw vital signs signals, performs preprocessing operations, and outputs a set of sleep physiological parameters. In this embodiment of the invention, the raw vital sign signals acquired by millimeter-wave radar can be represented as: In the formula, These are primitive vital signs signals. For the starting frequency, For bandwidth, The chirp period is typically 50 milliseconds. For phase modulation term, This represents the signal amplitude.
[0024] The data processing module mainly employs the following units to preprocess the raw vital signs signals: a filtering unit, a phase demodulation unit, a blind source separation unit, and a feature extraction unit. Specifically, the filtering unit uses a digital filter to filter the raw vital signs signals; the phase demodulation unit demodulates the phase components of the filtered raw vital signs signals to extract the micro-displacement signals related to chest cavity displacement; the blind source separation unit uses a blind source separation algorithm to separate the respiratory and heartbeat signal components from the micro-displacement signals; and the feature extraction unit performs time-frequency domain analysis on the respiratory and heartbeat signal components to extract the sleep physiological parameters constituting the sleep physiological parameter set.
[0025] In a specific embodiment of the present invention, the digital filter used in the filtering unit is a recursive least squares (RLS) adaptive filter; in the phase demodulation unit, the demodulation of the phase component of the filtered original vital sign signal mainly includes three steps: phase extraction, phase unwinding, and micro-displacement signal extraction. The expression for phase extraction is: In the formula, The extracted raw phase signal, These are the orthogonal components of the original vital signs signals. This refers to the in-phase component of the original vital signs signal.
[0026] The expression for phase unwinding is: In the formula, The phase signal after unwinding, The extracted raw phase signal, For cumulative summation function, For phase difference, This indicates dividing the phase difference by Round down.
[0027] The expression for extracting the micro-displacement signal is: In the formula, This is a micro-displacement signal. The phase signal after unwinding, is a coefficient.
[0028] The process by which the blind source separation unit separates the respiratory signal component and the heartbeat signal component from the micro-motion displacement signal is as follows: This invention, in its embodiment for blind source separation algorithm, first employs a fourth-order Busterworth bandpass filter to filter the micro-displacement signal. The passband frequency of this filter... Interval setting Within Hz, the stopband attenuation is greater than or equal to 40dB. Simultaneously, the coefficients used in the separation process are set, including: the lower passband angular limit frequency. Upper passband angular frequency ; center angular frequency ;bandwidth After the coefficients are set, for the respiratory signal components, this embodiment of the invention analyzes and extracts them based on the instantaneous respiratory variation coefficient and the respiratory harmonic index, wherein the instantaneous respiratory variation coefficient is expressed as: In the formula, The standard deviation between respirations. The mean is used; the respiratory harmonic index is expressed as: In the formula, This represents the power spectral density value corresponding to the respiratory rate. The power spectral density is given. For the heartbeat signal component, an adaptive noise canceller is first used to eliminate respiratory harmonics in the micro-displacement signal. Then, wavelet transform is used for denoising and enhancement. In this embodiment of the invention, the wavelet transform uses the db4 wavelet basis for 5-level decomposition. The instantaneous heart rate is calculated on the denoised and enhanced signal to obtain the heartbeat signal component.
[0029] In another specific embodiment of the present invention, the blind source separation unit also performs Doppler analysis on the original phase, calculates the micro-Doppler velocity and sets a threshold, and determines the time of body movement of the target when the signal exceeds the threshold, thereby statistically analyzing data such as body movement frequency, intensity and duration.
[0030] The data processed in the above way are integrated and extracted by the feature extraction unit to construct a high temporal resolution set of sleep physiological parameters. This set contains data such as respiratory waveforms and frequencies, heart rate, and body movement frequency for each time period, providing direct input for the subsequent sleep stage identification module.
[0031] The sleep stage identification module receives the set of sleep physiological parameters and performs sleep stage identification on the set of sleep physiological parameters based on a preset integrated classification model, and outputs the sleep stage result corresponding to each sleep stage information; In this embodiment of the invention, the preset ensemble classification model consists of multiple heterogeneous base classifiers. In this embodiment, at least two of the following are selected: a long short-term memory neural network focused on temporal pattern recognition, a support vector machine adept at handling high-dimensional nonlinear features, and a random forest classifier with strong interpretability. Each base classifier performs parallel inference with the aforementioned feature vectors as independent inputs, and its output is fused through a weighted voting or meta-learner to finally generate an ensemble decision, thereby improving the stage recognition, especially the classification accuracy and robustness for non-REM sleep sub-stages and REM sleep stages.
[0032] When performing sleep stage identification, this embodiment of the invention analyzes the set of sleep physiological parameters using a fixed duration as a basic analysis unit. For each basic analysis unit, its basic features are calculated to make a preliminary state judgment, primarily determining whether the target is currently in the wakeful or sleep stage. For basic analysis units determined to be in the sleep stage, a classifier trained on sleep data is further used, and an extended feature set containing higher-order temporal dependencies is input to further distinguish between light sleep, deep sleep, and REM sleep. Finally, the preliminary stage sequence output for the entire night is subjected to temporal smoothing post-processing based on a Hidden Markov Model. Using prior probability constraints of sleep state transitions, isolated misclassification points caused by transient signal interference are corrected, thereby outputting a continuous and stable sleep stage result sequence that conforms to physiological laws.
[0033] In a specific embodiment of the present invention, during the preliminary state judgment process of calculating the basic features of each basic analysis unit, the basic features include respiratory features, heart rate features, body movement features, and cross-modulation features. Among them, respiratory features include average respiratory rate, respiratory depth, and respiratory rhythm entropy; heart rate features include average heart rate, instantaneous heart rate standard deviation, high-frequency power, low-frequency power, and the ratio of low-frequency to high-frequency power in heart rate variability analysis; body movement features include the cumulative intensity and average duration of body movement events in each basic analysis unit; and cross-modulation features are calculated by measuring the amplitude modulation coherence of respiratory signals and heartbeat signals in a specific frequency band. The above basic features constitute a multidimensional feature vector input to the preset integrated classification model.
[0034] The Traditional Chinese Medicine sleep assessment module includes a quantitative mapping unit, a spatiotemporal correlation unit, and an analysis and decision-making unit, among which: The quantitative mapping unit converts the sleep staging results and the set of sleep physiological parameters into TCM quantitative indicators according to the preset TCM theory mapping rules. In this embodiment of the invention, the quantitative mapping unit first extracts features related to deep sleep, rapid eye movement (REM) sleep, and wakefulness from the sleep stage results and sleep physiological parameter set; then, according to the preset TCM theory mapping rules, it maps the features related to deep sleep to a first TCM quantitative index representing the storage of Yin Qi; maps the features related to REM sleep to a second TCM quantitative index representing the degree of Yang entering Yin and the smooth transition between Yin and Yang; and maps the features related to wakefulness to a third TCM quantitative index representing the level of abnormal Wei Qi circulation or restlessness of the spirit.
[0035] In a specific embodiment of the present invention, the characteristic quantities related to the deep sleep period mainly include the total duration of the deep sleep period and the average single duration; the characteristic quantities related to the rapid eye movement (REM) period mainly include the appearance regularity of the REM period, the average cycle length, and its proportion in the total night sleep duration; the characteristic quantities related to the wakefulness period mainly include the total number of night awakenings, the average duration, and the frequency of awakenings in the second half of the night.
[0036] The time-space correlation unit establishes a three-level mapping database of time- meridian -zang-fu according to the theory of traditional Chinese medicine's midnight-noon ebb-flow. According to the sleep stage results and their corresponding detection times, it queries the corresponding zang-fu syndrome differentiation rules in the three-level mapping database and generates corresponding traditional Chinese medicine syndrome differentiation labels. In the embodiment of the present invention, in the process of establishing the three-level mapping database of time- meridian -zang-fu, first, based on the theory of traditional Chinese medicine's midnight-noon ebb-flow, the 24-hour cycle is divided into twelve consecutive preset time intervals; then for each preset time interval, the corresponding dominant human meridians and core zang-fu organs are associated and configured according to traditional Chinese medicine theory; the zang-fu syndrome differentiation rules are set for each preset time interval, specifically: multiple abnormal conditions are set according to the sleep stage results in each preset time interval, and the functional states of the dominant human meridians and core zang-fu organs corresponding to this preset time interval under each abnormal condition are obtained; when the sleep stage results in the preset time interval meet the set abnormal conditions, the functional states of the dominant human meridians and core zang-fu organs in this preset time interval are used as the corresponding traditional Chinese medicine syndrome differentiation labels; the three-level mapping database of time- meridian -zang-fu is constructed according to the zang-fu syndrome differentiation rules of all preset time intervals.
[0037] In a specific embodiment of the present invention, based on the theory of traditional Chinese medicine's midnight-noon ebb-flow, the 24-hour cycle is divided into twelve consecutive preset time intervals. For example, the zi period is between 23:00 and 01:00, the chou period is between 01:00 and 03:00, the yin period is between 03:00 and 05:00, and so on, to obtain twelve preset time intervals; when associating and configuring the corresponding dominant human meridians and core zang-fu organs for each preset time interval according to traditional Chinese medicine theory, the relevant content in the existing traditional Chinese medicine theory can be referred to. Specifically, the corresponding dominant meridians and core zang-fu organs for each preset time interval are all in the prior art. For example, the dominant meridian corresponding to the zi period is the gallbladder meridian, and the core zang-fu organ is the gallbladder; the chou period corresponds to the liver meridian and the liver.
[0038] The embodiment of the present invention further takes the zi period, the chou period, and the yin period as examples to illustrate the specific process of setting the zang-fu syndrome differentiation rules for each preset time interval: For the Zi hour (11 PM - 1 AM), the sleep stage results corresponding to this time are obtained. These results are primarily quantitative indicators output by the quantitative mapping unit. These include a first TCM quantitative indicator representing the storage of Yin energy, a second TCM quantitative indicator representing the degree of Yang entering Yin and the smooth transition between Yin and Yang, and a third TCM quantitative indicator representing the level of Wei Qi (defensive Qi) dysfunction or restlessness of the spirit. Therefore, the abnormal conditions for the sleep stage results at this time are set as follows: when the sleep latency is greater than 45 minutes and the target's turning frequency is greater than 5 times per hour, the target's meridian and core organ functions are considered to be in a state of Gallbladder Qi deficiency or Shao Yang pivot dysfunction. The TCM diagnostic labels corresponding to the staged sleep results are gallbladder qi deficiency or Shaoyang pivot dysfunction, thus establishing the organ differentiation rules for the Zi hour (11 PM - 1 AM). For the Chou hour (1 AM - 3 AM), the abnormal conditions for the sleep staged sleep results are set as follows: when there are continuous body movement signals in the quantitative indicators corresponding to the sleep staged sleep results or the target is in the awakening period, the functional state of the target's meridians and core organs is considered to be liver failing to store blood or liver fire disturbing the spirit. For the Yin hour (3 AM - 5 AM), the abnormal conditions for the sleep staged sleep results are set as follows: when the quantitative indicators corresponding to the sleep staged sleep results detect that the target has transitioned from deep sleep to light sleep and then back to the awakening period, the functional state of the target's meridians and core organs is considered to be lung qi deficiency or insufficient qi and blood production. Based on the organ differentiation rule establishment process illustrated above, and in accordance with the relevant content of the TCM meridian-noon flow theory, organ differentiation rules for each hour can be established sequentially, thereby obtaining the three-level mapping database of hour-meridian-organ required by the embodiments of the present invention. This database is stored in the system of the present invention as pre-established content for easy access at any time.
[0039] When monitoring a target, the system collects raw vital signs data and performs a series of transformations to obtain corresponding sleep staging results. Each sleep staging result has a corresponding precise timestamp. Based on the data in the sleep staging results within this timestamp, the system queries the three-level mapping database of time-meridian-organ to determine whether it meets the corresponding organ differentiation rules. This generates corresponding TCM differentiation labels as the output of the spatiotemporal correlation module. It should be noted that for data from a single night, multiple TCM differentiation labels may be generated because different rules are met at different times or at the same time. In other words, the TCM differentiation labels output by this module may be multiple.
[0040] The analysis and decision-making unit takes TCM quantitative indicators and TCM syndrome differentiation labels as inputs, and uses a pre-trained machine learning model to generate TCM health monitoring results for the target. In this embodiment of the invention, the pre-trained machine learning model used by the analysis and decision-making unit can be a support vector machine or a multilayer perceptron. This machine learning model is trained on a large dataset of sleep data and TCM syndrome pairings labeled by clinical experts, and is capable of learning complex, nonlinear mapping relationships. When using TCM quantitative indicators and TCM syndrome labels as input, the two need to be encoded and fused to form a joint feature vector. The machine learning model is then used to calculate and output a probabilistic score for various common TCM syndromes. Based on the probabilistic scores for various common TCM syndromes, a structured TCM syndrome probabilistic analysis result is compiled, thereby obtaining the target TCM health monitoring result.
[0041] The monitoring results output module integrates the TCM health monitoring results to generate a visualized health report.
[0042] In this embodiment of the invention, the monitoring result output module outputs health reports through two visualization platforms. The first is a cloud platform deployed on a cloud server and accessed via a web browser. This platform is typically geared towards health management institutions and supports multiple users viewing the vital sign waveforms, real-time sleep staging, and abnormal event warnings of the current monitored target in real time. It provides a complete and detailed version of the structured health report and supports historical report review, comparative analysis, and PDF export. The second is... Figure 2 The built-in mini-program platform of the Zhongguancun 4 visualization terminal is designed for individual users. It provides simple real-time monitoring and health report results, and automatically marks and alerts users who show severe symptoms or abnormal sleep structure for several consecutive days.
[0043] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.
Claims
1. A TCM sleep health monitoring system based on millimeter-wave radar, characterized in that, include: The data acquisition module is used to deploy millimeter-wave radar sensors in a set manner and to collect raw vital signs signals of the target using millimeter-wave radar sensors. The data processing module is used to receive the raw vital signs signals and perform preprocessing operations to output a set of sleep physiological parameters; The set of sleep physiological parameters includes at least: respiratory rate, heart rate, and body movement rate; The sleep stage identification module is used to receive the set of sleep physiological parameters, identify the sleep stages based on a preset integrated classification model, and output the sleep stage results; the sleep stage results include: sleep stage corresponding to the wakefulness stage, light sleep stage, deep sleep stage, and REM sleep stage respectively; The TCM sleep assessment module includes a quantitative mapping unit, a spatiotemporal correlation unit, and an analysis and decision-making unit; The quantitative mapping unit is used to convert the sleep staging results and the set of sleep physiological parameters into TCM quantitative indicators according to preset TCM theory mapping rules. The spatiotemporal association unit is used to establish a three-level mapping database of time, meridians, and internal organs based on the theory of meridian flow in traditional Chinese medicine. Based on the sleep stage results and their corresponding detection times, the corresponding organ differentiation rules are queried in the three-level mapping database to generate corresponding TCM differentiation tags. The analysis and decision-making unit takes the TCM quantitative indicators and the TCM syndrome differentiation labels as inputs and uses a pre-trained machine learning model to generate the TCM health monitoring results of the target. The monitoring results output module is used to integrate the TCM health monitoring results and generate a visualized health report.
2. The millimeter wave radar-based traditional Chinese medicine sleep health monitoring system according to claim 1, characterized in that: The data acquisition module deploys millimeter-wave radar sensors in a predetermined manner, specifically as follows: The millimeter-wave radar sensor is deployed under the mattress, and the installation position is perpendicular to the mattress surface and points towards the user's sleeping position; The main lobe of the millimeter-wave signal emitted by the millimeter-wave radar sensor covers the core area of the torso when the target is lying supine on the mattress surface.
3. The TCM sleep health monitoring system based on millimeter-wave radar according to claim 1, characterized in that: The data processing module receives the raw vital signs signal and performs preprocessing operations, including: a filtering unit, a phase demodulation unit, a blind source separation unit, and a feature extraction unit; The filtering unit is used to filter the original vital signs signal using a digital filter; The phase demodulation unit is used to demodulate the phase components of the filtered original vital signs signal and extract the micro-displacement signal related to the chest cavity displacement. The blind source separation unit is used to separate the respiratory signal component and the heartbeat signal component in the micro-motion displacement signal using a blind source separation algorithm. The feature extraction unit is used to perform time-frequency domain analysis on the respiratory signal component and the heartbeat signal component to extract the sleep physiological parameters that constitute the sleep physiological parameter set.
4. The TCM sleep health monitoring system based on millimeter-wave radar according to claim 1, characterized in that: In the sleep stage identification module, the process of identifying sleep stages based on a preset integrated classification model of the sleep physiological parameter set is as follows: Based on the set of sleep physiological parameters, a multidimensional feature vector is extracted for input into a preset integrated classification model; the multidimensional feature vector includes respiratory features, heart rate features, body movement features, and cross-modulation features. The preset integrated classification model is composed of at least two of the following: long short-term memory neural network, support vector machine, and random forest classifier; The multidimensional feature vectors are input into the ensemble classification model for parallel reasoning and fusion decision-making, and the sleep stage results corresponding to each sleep stage information are output.
5. The TCM sleep health monitoring system based on millimeter-wave radar according to claim 1, characterized in that: The quantitative mapping unit is used to convert the sleep staging results and the set of sleep physiological parameters into TCM quantitative indicators according to preset TCM theoretical mapping rules, including: Features related to deep sleep, rapid eye movement (REM) sleep, and wakefulness are extracted from the sleep stage results and the set of sleep physiological parameters. According to the preset TCM theory mapping rules, the deep sleep-related features are mapped to the first TCM quantitative index representing the storage of Yin energy. The rapid eye movement phase-related features are mapped to a second TCM quantitative index that characterizes the degree of yang entering yin and the smoothness of yin-yang interaction. The aforementioned awakening-related characteristics are mapped to a third TCM quantitative indicator characterizing the level of abnormal Wei Qi circulation or restlessness of the mind.
6. The TCM sleep health monitoring system based on millimeter-wave radar according to claim 5, characterized in that: Features related to deep sleep, REM sleep, and wakefulness were extracted from the sleep stage results and the set of sleep physiological parameters, including: The relevant characteristics of deep sleep include the total duration of deep sleep and the average duration of a single episode. The REM sleep-related characteristics include the regularity of periodic occurrence, the average period length, and the proportion of REM sleep in total sleep duration. The relevant characteristics of the awakening period include the number of awakenings, the average duration, and the frequency of awakenings in the second half of the night.
7. The TCM sleep health monitoring system based on millimeter-wave radar according to claim 1, characterized in that: In the aforementioned spatiotemporal correlation unit, based on the theory of meridian flow in Traditional Chinese Medicine, a three-level mapping database of time-meridian-organ is established, specifically as follows: Based on the theory of meridian flow in traditional Chinese medicine, the 24-hour cycle is divided into twelve consecutive preset time intervals. For each preset time interval, the corresponding dominant human meridians and core organs are configured according to traditional Chinese medicine theory; For each preset time interval, a differentiation rule for the viscera and bowels is established. The differentiation rule for the viscera and bowels is as follows: Multiple abnormal conditions are set based on the sleep stage results in each preset time interval, and the functional status of the dominant human meridians and core organs in each preset time interval is obtained under each abnormal condition. When the sleep stage results meet the set abnormal conditions within a preset time interval, the functional state of the human meridians and core organs dominated by that preset time interval is used as the corresponding TCM syndrome label. A three-level mapping database of time-meridian-organ is constructed based on the organ differentiation rules of all preset time intervals.
8. The TCM sleep health monitoring system based on millimeter-wave radar according to claim 1, characterized in that: In the spatiotemporal correlation unit, the corresponding organ differentiation rules are queried in the three-level mapping database to generate corresponding TCM differentiation tags, including: Obtain the sleep stage results and their corresponding detection times, and map the detection times to the corresponding specific time intervals according to the theory of meridian flow in traditional Chinese medicine. The system queries the organ differentiation rules corresponding to the specific time interval in the time-meridian-organ three-level mapping database, matches the sleep stage results with the organ differentiation rules corresponding to the specific time interval, and outputs the corresponding TCM differentiation labels based on the matching results.