Physical state judgment method and device based on undisturbed physical sign continuous monitoring

Abstract

The invention relates to a physical state judgment method and device based on undisturbed physical sign continuous monitoring, and the method comprises the steps: continuously collecting the physiological physical sign data of a user through an undisturbed monitoring device which is integrated in a wearable device; preprocessing the physiological sign data, and extracting a physiological feature set related to the physique state; and inputting the physiological feature set into a pre-trained physique judgment model, and outputting a judgment result of the physique state of the user so as to achieve the purpose of converting the long-term physiological data of the user into accurate physique state evaluation and personalized health management suggestions.

Description

Technical Field

[0001] This invention relates to the field of medical and health monitoring technology, and in particular to a method and device for judging physical condition based on non-intrusive continuous monitoring of vital signs. Background Technology

[0002] In the current field of health and wellness, personalized health management is receiving increasing attention, especially with the growing popularity of health preservation concepts based on traditional Chinese medicine's constitution theory. However, existing health monitoring technologies have significant limitations: mainstream wearable devices mostly focus on short-term monitoring of single physiological parameters, lacking the ability to conduct long-term, systematic assessments of constitution, which characterizes the overall state of the human body; traditional constitution identification methods rely on medical history taking and scales, which are highly subjective and lack dynamic tracking capabilities. These technological limitations cause health management plans to often lag behind changes in the user's actual physiological state, making it difficult to achieve truly personalized and preventative health interventions. Summary of the Invention

[0003] The main objective of this invention is to provide a method and device for judging physical condition based on non-intrusive continuous monitoring of vital signs. Through non-intrusive continuous monitoring and intelligent analysis, the invention aims to transform users' long-term physiological data into accurate physical condition assessments and personalized health management recommendations.

[0004] To achieve the above objectives, the present invention provides a method for judging physical condition based on non-intrusive continuous monitoring of vital signs, comprising the following steps: The user's physiological data is continuously collected through a non-intrusive monitoring device, which is integrated into the wearable device. The physiological sign data are preprocessed, and a set of physiological features related to physical condition is extracted; The physiological feature set is input into a pre-trained physical condition assessment model, which outputs the assessment result of the user's physical condition.

[0005] Furthermore, the step of continuously collecting users' physiological data using a non-intrusive monitoring device includes: Physiological parameters are synchronously acquired by a multimodal sensor array integrated into a wearable device, the sensor array including optical sensors, bioimpedance sensors and inertial measurement units; The sampling frequency is adaptively adjusted based on the user's activity status to collect the user's original physiological characteristics data. The basic sampling rate is used when the user is at rest, and the enhanced sampling rate is increased when the user is in motion. The original physiological sign data is corrected in real time by a motion compensation algorithm to obtain the physiological sign data.

[0006] Furthermore, the step of continuously collecting users' physiological data through a non-intrusive monitoring device also includes: By fusing impedance detection circuit and accelerometer signal, the contact status between wearable device and user's body surface can be determined in real time. When the contact resistance continuously exceeds the preset threshold range and the accelerometer signal characteristics disappear, the device is automatically determined to be out of contact, and data acquisition and storage are suspended. When the device is detected to be connected to a charging power source, it automatically switches to continuous monitoring mode, removes the battery power consumption limit, activates high sampling rate and full parameter acquisition functions, and uses the regulated output of the power management chip to provide continuous operating current to the sensor array. A data quality assessment matrix is ​​established based on the real-time calculated data signal-to-noise ratio and impulse response characteristics. When the quality assessment matrix value is lower than the set threshold, the sensor calibration sequence is automatically triggered. The sensor sensitivity is adaptively corrected by emitting a test signal of known amplitude and detecting the response curve.

[0007] Further, the steps of preprocessing the physiological sign data and extracting a set of physiological features related to physical condition include: The physiological sign data are segmented into sliding windows, and signal quality assessment is performed on the data within each window to select qualified data segments; Denoising is performed on qualified data segments to retain valid physiological signals; Extract physiological feature sets from effective physiological signals, including time-domain statistical features, frequency-domain energy features, and nonlinear dynamic features.

[0008] Further, the step of inputting the physiological feature set into a pre-trained physical fitness assessment model and outputting the assessment result of the user's physical fitness status includes: The extracted physiological feature set is input into the physical condition assessment model based on the attention mechanism; The feature weighting module in the physical fitness assessment model evaluates the contribution of each feature in the input physiological feature set to the physical fitness classification at different time steps, and generates corresponding weight coefficients to weight the feature representation based on the contribution. Based on weighted feature representation, a softmax classification layer outputs probability distributions corresponding to different body types. The dominant constitution type is determined based on the probability distribution, and a corresponding credibility index is generated.

[0009] Furthermore, the steps for constructing the physical fitness assessment model include: Collect long-term physiological monitoring data and corresponding body type labels from multiple users, and collect background covariates corresponding to each user. The background covariates include at least one of age, gender, and basal metabolic rate information. Construct a deep neural network model, including a feature encoding layer, a temporal modeling layer, and a classification output layer; During the model training phase, the background covariates are fused with the set of physiological features extracted from long-term physiological monitoring data to train the model and achieve personalized calibration of the physical fitness assessment results. A focus loss function is used for model training to address the class imbalance problem in the training data for different physical states. The model performance is evaluated using a validation set, and the optimal model parameters are saved for online inference.

[0010] Furthermore, after the step of outputting the judgment result of the user's physical condition, it also includes: Establish a personal baseline model based on the user's historical physical condition data; The current assessment results are compared and analyzed with the individual's baseline to identify changes in physical condition; When abnormal changes are detected, an early warning signal is generated and a corresponding health intervention plan is recommended.

[0011] The present invention also provides a physical condition assessment device based on non-intrusive continuous monitoring of vital signs, comprising: The data acquisition module is used to continuously collect users' physiological data through a non-intrusive monitoring device. The feature extraction module is used to preprocess the physiological sign data and extract a set of physiological features related to physical condition. The data analysis module is used to input the physiological feature set into the pre-trained physical fitness assessment model and output the assessment result of the user's physical fitness status.

[0012] The present invention also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the above-described method for judging physical condition based on non-intrusive continuous monitoring of vital signs.

[0013] The present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the above-described method for judging physical condition based on non-intrusive continuous monitoring of vital signs.

[0014] The present invention provides a method and device for judging physical condition based on non-intrusive continuous monitoring of vital signs, which has the following beneficial effects: The present invention achieves high-quality continuous collection of physiological data without the user's awareness through multimodal sensor fusion and adaptive sampling mechanisms, effectively solving the pain points of traditional monitoring methods that rely on active user participation and are susceptible to data interference. Deeply integrating long-term physiological characteristics with the user's multidimensional background information enables the physical condition judgment model to accurately distinguish between stable physical characteristics and temporary fluctuations, improving the accuracy and personalization of physical condition identification. By establishing a personal dynamic baseline model, it can keenly identify abnormal trends in physical condition, realizing a shift from passive monitoring to proactive early warning, providing a window of opportunity for early health intervention. Through a complete technical closed loop from data collection and intelligent analysis to intervention recommendations, the present invention organically combines traditional Chinese medicine physical condition theory with modern artificial intelligence technology, providing users with a truly scientific, systematic, and personalized health management solution. Attached Figure Description

[0015] Figure 1 This is a flowchart illustrating a method for judging physical condition based on non-intrusive continuous monitoring of vital signs in one embodiment of the present invention. Figure 2 This is a structural block diagram of a physical condition judgment device based on non-intrusive continuous monitoring of vital signs in one embodiment of the present invention; Figure 3 This is a schematic block diagram of the structure of a computer device according to an embodiment of the present invention.

[0016] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0017] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0018] Reference Figure 1 This is a flowchart illustrating a method for judging physical condition based on non-intrusive continuous monitoring of vital signs proposed in this invention, including the following steps: S1, continuously collects the user's physiological data through a non-intrusive monitoring device, wherein the non-intrusive monitoring device is integrated into the wearable device; S2, preprocess the physiological signs data and extract the set of physiological features related to physical condition; S3, input the physiological feature set into the pre-trained physical condition judgment model, and output the judgment result of the user's physical condition status.

[0019] In one embodiment, for step S1, The steps for continuously collecting users' physiological data using a non-intrusive monitoring device include: Physiological parameters are synchronously acquired by a multimodal sensor array integrated into a wearable device, the sensor array including optical sensors, bioimpedance sensors and inertial measurement units; The sampling frequency is adaptively adjusted based on the user's activity status to collect the user's original physiological characteristics data. The basic sampling rate is used when the user is at rest, and the enhanced sampling rate is increased when the user is in motion. The original physiological sign data is corrected in real time by a motion compensation algorithm to obtain the physiological sign data.

[0020] In practical implementation, a multimodal sensor array integrated into the wearable device is used for synchronous data acquisition. This array includes optical sensors (primarily for acquiring photoplethysmography (PPG) signals, which can obtain parameters such as blood oxygen saturation and heart rate variability), bioimpedance sensors (to obtain information on body fluid distribution and skin conductance by measuring changes in tissue impedance), and an inertial measurement unit (IMU) (to acquire triaxial acceleration and angular velocity data to characterize the user's activity state). A multi-parameter synchronous acquisition strategy ensures the integrity and synchronicity of the multi-dimensional physiological information required for subsequent physical condition assessment. During actual acquisition, the sampling strategy is intelligently adjusted based on the user's activity state identified in real time by the IMU: when the user is stationary, a basic sampling rate (e.g., 32Hz) is used to balance power consumption and data requirements; when the user is detected to be in motion, an enhanced sampling rate (e.g., 128Hz) is automatically switched to capture richer physiological details and cope with motion interference. This adaptive sampling mechanism resolves the inherent contradiction between power consumption and data quality in continuous monitoring. Furthermore, to address the unavoidable signal quality issues during motion, this embodiment introduces a motion compensation algorithm at the signal processing front end. The compensation algorithm uses motion acceleration data acquired by the IMU as a noise reference and performs real-time correction on the original PPG and impedance signals through adaptive filtering techniques (such as the Normalized Least Mean Square NLMS algorithm) to separate physiological signals from motion artifacts, thereby outputting reliable physiological sign data.

[0021] In one embodiment, the step of continuously collecting the user's physiological data using a non-intrusive monitoring device further includes: By fusing impedance detection circuit and accelerometer signal, the contact status between wearable device and user's body surface can be determined in real time. When the contact resistance continuously exceeds the preset threshold range and the accelerometer signal characteristics disappear, the device is automatically determined to be out of contact, and data acquisition and storage are suspended. When the device is detected to be connected to a charging power source, it automatically switches to continuous monitoring mode, removes the battery power consumption limit, activates high sampling rate and full parameter acquisition functions, and uses the regulated output of the power management chip to provide continuous operating current to the sensor array. A data quality assessment matrix is ​​established based on the real-time calculated data signal-to-noise ratio and impulse response characteristics. When the quality assessment matrix value is lower than the set threshold, the sensor calibration sequence is automatically triggered. The sensor sensitivity is adaptively corrected by emitting a test signal of known amplitude and detecting the response curve.

[0022] In practical implementation, to achieve a truly non-intrusive monitoring experience and ensure the continuity and reliability of data acquisition, step S1 establishes a device status management and quality assurance mechanism: The contact impedance at the electrode-skin interface is monitored in real time via an impedance detection circuit, and combined with accelerometer signal characteristics for fusion judgment. When the contact impedance continuously exceeds the preset normal range (e.g., greater than 2MΩ) and the accelerometer signal variance drops to a near-zero rest threshold, it can be comprehensively determined that the device is in a detached state. At this time, data acquisition and storage functions are automatically paused to avoid invalid data occupying storage resources and reduce system power consumption. To further overcome the limitations of battery life on long-term continuous monitoring, an intelligent power supply switching strategy is also set up: When the device is detected to be connected to a charging power source, it immediately and automatically switches to continuous monitoring mode. At this time, all sampling restrictions based on battery power consumption are lifted, and high sampling rate (up to 256Hz) and full parameter acquisition functions are simultaneously activated. The regulated output provided by the power management chip provides a clean and stable operating current to the sensor array, thereby improving data density and quality during the charging period. To address the potential performance drift of sensors over long-term use, an automatic calibration mechanism based on real-time signal quality assessment was constructed. This mechanism establishes a multi-dimensional data quality assessment matrix by calculating the signal-to-noise ratio (SNR) of the data and analyzing its response characteristics to standard test pulses. When the overall score of this matrix falls below a preset threshold, a built-in sensor calibration sequence is automatically triggered: a signal generator emits an electrical or optical test signal with known amplitude and waveform; the sensor's response curve is detected and compared with a standard response model, ultimately achieving adaptive correction of the sensor's sensitivity. This embodiment ensures the accuracy and consistency of monitoring data throughout its entire lifecycle through a closed-loop calibration mechanism, providing users with a truly maintenance-free and disturbance-free monitoring experience.

[0023] In one embodiment, for step S2, The steps of preprocessing the physiological sign data and extracting a set of physiological features related to physical condition include: The physiological sign data are segmented into sliding windows, and signal quality assessment is performed on the data within each window to select qualified data segments; Denoising is performed on qualified data segments to retain valid physiological signals; Extract physiological feature sets from effective physiological signals, including time-domain statistical features, frequency-domain energy features, and nonlinear dynamic features.

[0024] In practice, a fixed-duration sliding window (e.g., a 5-minute window with a 2-minute overlap) is used to segment continuously collected physiological data. The overlapping window design ensures data continuity and avoids boundary effects during feature extraction. For the data within each window, a multi-index signal quality assessment is performed, including calculating the signal-to-noise ratio (SNR), detecting signal saturation, and baseline drift. Only data segments that simultaneously meet the SNR threshold, have no saturation, and have baseline drift within acceptable limits are marked as qualified data segments. The selected qualified data segments are then processed using a denoising algorithm based on wavelet transform. By selecting appropriate mother wavelet basis functions and threshold rules, noise components are separated from effective physiological signals. Optimization is specifically applied to common noise types such as motion artifacts and power frequency interference, thereby preserving effective signal components that reflect the true physiological state.

[0025] From the denoised effective physiological signals, the system comprehensively extracts a feature set across three dimensions: time-domain statistical features, including distribution characteristics such as mean, variance, and skewness, as well as the variability of peak-to-peak intervals; frequency-domain energy features, calculated using Fast Fourier Transform to measure the energy proportion and power spectral entropy of each frequency band (e.g., very low frequency, low frequency, and high frequency), revealing physiological rhythm information; and nonlinear dynamic features, calculating indicators such as sample entropy and Lyapunov exponents to quantify the complexity and stability of the physiological system. These multi-dimensional features together constitute a comprehensive physiological feature set characterizing the user's physical condition, providing rich and comprehensive input for subsequent intelligent judgment.

[0026] In one embodiment, for step S3, The steps of inputting the physiological feature set into a pre-trained physical fitness assessment model and outputting the assessment result of the user's physical fitness status include: The extracted physiological feature set is input into the physical condition assessment model based on the attention mechanism; The feature weighting module in the physical fitness assessment model evaluates the contribution of each feature in the input physiological feature set to the physical fitness classification at different time steps, and generates corresponding weight coefficients to weight the feature representation based on the contribution. Based on weighted feature representation, a softmax classification layer outputs probability distributions corresponding to different body types. The dominant constitution type is determined based on the probability distribution, and a corresponding credibility index is generated.

[0027] In practice, the multi-dimensional physiological feature set extracted in step S2 is input into a deep neural network model based on an attention mechanism. This model dynamically analyzes the differences in the contribution of each physiological feature to the constitution classification at different time steps through its built-in feature weighting module. Specifically, the model evaluates the importance of each feature in the current context by calculating the attention score of each feature in time sequence, and generates corresponding weight coefficients based on these scores to adaptively weight the original feature representation. This mechanism enables the model to focus on the feature change patterns most relevant to the current constitution state. For example, the sensitivity to nonlinear features of heart rate variability is significantly enhanced in a fatigued state. Based on the weighted feature representation, the model outputs a probability distribution corresponding to nine TCM constitution types through a softmax classification layer. This distribution accurately reflects the probability that the user belongs to each constitution type. Finally, the constitution type with the highest probability value is selected as the dominant constitution output. At the same time, a corresponding credibility index is generated based on the dispersion of the probability distribution (such as entropy calculation). When the probability distribution is relatively concentrated, high credibility is assigned. When the probabilities of multiple constitution types are close, it indicates that the judgment result is uncertain and needs to be combined with more observation data.

[0028] In one embodiment, the steps for constructing a physical fitness assessment model include: Collect long-term physiological monitoring data and corresponding body type labels from multiple users, and collect background covariates corresponding to each user. The background covariates include at least one of age, gender, and basal metabolic rate information. Construct a deep neural network model, including a feature encoding layer, a temporal modeling layer, and a classification output layer; During the model training phase, the background covariates are fused with the set of physiological features extracted from long-term physiological monitoring data to train the model and achieve personalized calibration of the physical fitness assessment results. A focus loss function is used for model training to address the class imbalance problem in the training data for different physical states. The model performance is evaluated using a validation set, and the optimal model parameters are saved for online inference.

[0029] In practice, multi-user physiological monitoring data for at least three months is collected to separate true physical characteristics from short-term fluctuations. Simultaneously, physical type labels, confirmed through a professional and standardized physical constitution identification process, are collected for each long-term data point. To build a more interpretable and accurate model, multidimensional background covariates of users are also collected, including: static physiological indicators, including body mass index (BMI) and waist-to-hip ratio in addition to age and gender, to characterize users' basal metabolic rate and body shape characteristics; long-term lifestyle habits, including users' self-reported long-term geographical residence history, typical dietary preferences (such as a preference for spicy, cold, or oily foods), and regular exercise frequency, which are long-term environmental and behavioral factors shaping physical constitution; and health status background, including a medically confirmed history of chronic diseases (such as hypertension and diabetes) and long-term medication use, which significantly affect the baseline levels and dynamic responses of physiological parameters. During the model training phase, a phased feature aggregation and multi-dimensional fusion strategy is employed: First, long-term data is segmented and features are extracted on a weekly basis to form a time-series feature set; then, the stability patterns of these features over a span of several months are learned through a time-series modeling layer. The aforementioned multi-dimensional background covariates are treated as a whole and deeply fused with the long-term stable feature representations aggregated from physiological data. This allows the model to learn, for example, that similar heart rate variability patterns may correspond to completely different constitution assessments in a user living in a humid environment with specific dietary preferences and a user living in a dry environment with vastly different lifestyles. In this way, the model internalizes the modulating effects of complex factors on physiological representations, thereby achieving a deep, personalized calibration that goes beyond surface correlation. Simultaneously, a focus loss function is used to address the class imbalance problem of different constitution types in the training data, ensuring that the model's judgments are based on a deep understanding of the user's long-term, stable physiological base and their individual background panorama. This ensures that the conclusions more accurately reflect the "constitution" attribute, improving the rationality of the technical solution, clinical interpretability, and reliability of the judgment results.

[0030] In one embodiment, after the step of outputting the judgment result of the user's physical condition, the method further includes: Establish a personal baseline model based on the user's historical physical condition data; The current assessment results are compared and analyzed with the individual's baseline to identify changes in physical condition; When abnormal changes are detected, an early warning signal is generated and a corresponding health intervention plan is recommended.

[0031] In practice, after outputting the user's physical condition assessment, a dynamic health tracking and intervention mechanism is further implemented. A personal baseline model is established based on the user's historical physical condition data: All physical condition assessments and their corresponding physiological characteristic sets over at least three months are continuously accumulated. A dynamic evolution baseline of the physical condition is constructed using a Hidden Markov Model. This baseline not only includes the probability distribution of the dominant physical condition type but also records the normal fluctuation range of each physiological characteristic and its seasonal variation. The current assessment result is compared and analyzed with the personal baseline: An anomaly detection algorithm based on Mahalanobis distance is used to perform a multi-dimensional comparison between the latest physical condition assessment result and its underlying physiological characteristic vector and the personal baseline model. While monitoring the transformation of physical condition types (e.g., from "balanced constitution" to "qi deficiency constitution"), "latent changes" deviating from the individual's normal pattern in the physiological characteristic space are detected. This means that even if the dominant physical condition type remains unchanged, multiple physiological characteristics may show a trend towards another physical condition type. When abnormal changes are detected, an early warning signal is generated and a corresponding health intervention plan is recommended: Based on the degree, duration, and direction of deviation of the abnormal change, the early warning is divided into multiple levels (e.g., attention, alert, warning). Meanwhile, based on knowledge graph technology, abnormal patterns are matched with the intervention program library to generate personalized health recommendations. For example, when a continuous decline in the nonlinear characteristics of heart rate variability is detected, accompanied by changes in body fluid distribution, it is judged as "Qi deficiency tendency" and a combination of intervention programs including dietary therapy (such as astragalus and yam porridge), breathing training (abdominal breathing method) and adjustment of work and rest are recommended.

[0032] Reference Figure 2 Here is a structural block diagram of a physical condition assessment device based on non-intrusive continuous monitoring of vital signs according to an embodiment of the present invention, comprising: The data acquisition module is used to continuously collect users' physiological data through a non-intrusive monitoring device. The feature extraction module is used to preprocess the physiological sign data and extract a set of physiological features related to physical condition. The data analysis module is used to input the physiological feature set into the pre-trained physical fitness assessment model and output the assessment result of the user's physical fitness status.

[0033] For the specific implementation of each module in the above device example, please refer to the above method embodiments, which will not be repeated here.

[0034] Reference Figure 3 This invention also provides a computer device, which can be a server, and its internal structure can be as follows: Figure 3As shown, the computer device includes a processor, memory, display screen, input device, network interface, and database connected via a system bus. The processor provides computing and control capabilities. The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores the operating system, computer programs, and database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The database stores the data corresponding to this embodiment. The network interface is used to communicate with external terminals via a network connection. When the computer program is executed by the processor, it implements the above-described method.

[0035] Those skilled in the art will understand that Figure 3 The structures shown are merely block diagrams of some structures related to the present invention and do not constitute a limitation on the computer devices on which the present invention is applied.

[0036] An embodiment of the present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-described method. It is understood that the computer-readable storage medium in this embodiment can be a volatile readable storage medium or a non-volatile readable storage medium.

[0037] In summary, this invention continuously collects users' physiological data using a non-intrusive monitoring device integrated into a wearable device; preprocesses the physiological data and extracts a set of physiological features related to physical condition; inputs the set of physiological features into a pre-trained physical condition assessment model and outputs a judgment result of the user's physical condition, thereby achieving the goal of transforming users' long-term physiological data into accurate physical condition assessments and personalized health management recommendations.

[0038] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the present invention and embodiments can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual-rate SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM, etc.

[0039] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, apparatus, article, or method that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, apparatus, article, or method. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, apparatus, article, or method that includes that element.

[0040] The above description is only a preferred embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent structural or procedural transformations made based on the content of the present invention specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. A method for judging physical condition based on non-intrusive continuous monitoring of vital signs, characterized in that, Includes the following steps: The user's physiological data is continuously collected through a non-intrusive monitoring device, which is integrated into the wearable device. The physiological sign data are preprocessed, and a set of physiological features related to physical condition is extracted; The physiological feature set is input into a pre-trained physical condition assessment model, which outputs the assessment result of the user's physical condition.

2. The method for judging physical condition based on non-intrusive continuous monitoring of vital signs according to claim 1, characterized in that, The step of continuously collecting the user's physiological data through a non-intrusive monitoring device includes: Physiological parameters are synchronously acquired by a multimodal sensor array integrated into a wearable device, the sensor array including optical sensors, bioimpedance sensors and inertial measurement units; The sampling frequency is adaptively adjusted based on the user's activity status to collect the user's original physiological characteristics data. The basic sampling rate is used when the user is at rest, and the enhanced sampling rate is increased when the user is in motion. The original physiological sign data is corrected in real time by a motion compensation algorithm to obtain the physiological sign data.

3. The method for judging physical condition based on non-intrusive continuous monitoring of vital signs according to claim 1, characterized in that, The step of continuously collecting the user's physiological data through a non-intrusive monitoring device further includes: By fusing impedance detection circuit and accelerometer signal, the contact status between wearable device and user's body surface can be determined in real time. When the contact resistance continuously exceeds the preset threshold range and the accelerometer signal characteristics disappear, the device is automatically determined to be out of contact, and data acquisition and storage are suspended. When the device is detected to be connected to a charging power source, it automatically switches to continuous monitoring mode, removes the battery power consumption limit, activates high sampling rate and full parameter acquisition functions, and uses the regulated output of the power management chip to provide continuous operating current to the sensor array. A data quality assessment matrix is ​​established based on the real-time calculated data signal-to-noise ratio and impulse response characteristics. When the quality assessment matrix value is lower than the set threshold, the sensor calibration sequence is automatically triggered. The sensor sensitivity is adaptively corrected by emitting a test signal of known amplitude and detecting the response curve.

4. The method for judging physical condition based on non-intrusive continuous monitoring of vital signs according to claim 1, characterized in that, The step of preprocessing the physiological sign data and extracting a set of physiological features related to physical condition includes: The physiological sign data are segmented into sliding windows, and signal quality assessment is performed on the data within each window to select qualified data segments; Denoising is performed on qualified data segments to retain valid physiological signals; Extract physiological feature sets from effective physiological signals, including time-domain statistical features, frequency-domain energy features, and nonlinear dynamic features.

5. The method for judging physical condition based on non-intrusive continuous monitoring of vital signs according to claim 1, characterized in that, The step of inputting the physiological feature set into a pre-trained physical fitness assessment model and outputting the assessment result of the user's physical fitness status includes: The extracted physiological feature set is input into the physical condition assessment model based on the attention mechanism; The feature weighting module in the physical fitness assessment model evaluates the contribution of each feature in the input physiological feature set to the physical fitness classification at different time steps, and generates corresponding weight coefficients to weight the feature representation based on the contribution. Based on weighted feature representation, a softmax classification layer outputs probability distributions corresponding to different body types. The dominant constitution type is determined based on the probability distribution, and a corresponding credibility index is generated.

6. The method for judging physical condition based on non-intrusive continuous monitoring of vital signs according to claim 1, characterized in that, The steps for constructing the physical fitness assessment model include: Collect long-term physiological monitoring data and corresponding body type labels from multiple users, and collect background covariates corresponding to each user. The background covariates include at least one of age, gender, and basal metabolic rate information. Construct a deep neural network model, including a feature encoding layer, a temporal modeling layer, and a classification output layer; During the model training phase, the background covariates are fused with the set of physiological features extracted from long-term physiological monitoring data to train the model and achieve personalized calibration of the physical fitness assessment results. A focus loss function is used for model training to address the class imbalance problem in the training data for different physical states. The model performance is evaluated using a validation set, and the optimal model parameters are saved for online inference.

7. The method for judging physical condition based on non-intrusive continuous monitoring of vital signs according to claim 1, characterized in that, After the step of outputting the judgment result of the user's physical condition, the method further includes: Establish a personal baseline model based on the user's historical physical condition data; The current assessment results are compared and analyzed with the individual's baseline to identify changes in physical condition; When abnormal changes are detected, an early warning signal is generated and a corresponding health intervention plan is recommended.

8. A device for judging physical condition based on non-intrusive continuous monitoring of vital signs, characterized in that, include: The data acquisition module is used to continuously collect users' physiological data through a non-intrusive monitoring device. The feature extraction module is used to preprocess the physiological sign data and extract a set of physiological features related to physical condition. The data analysis module is used to input the physiological feature set into the pre-trained physical fitness assessment model and output the assessment result of the user's physical fitness status.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the physical condition judgment method based on non-intrusive continuous monitoring of vital signs as described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the physical condition judgment method based on non-intrusive continuous monitoring of vital signs as described in any one of claims 1 to 7.

Images