Spleen-kidney deficiency syndrome dynamic evaluation system based on multi-mode biosensor

Abstract

The invention relates to the field of medical health monitoring, and provides a spleen and kidney deficiency syndrome dynamic evaluation system based on a multi-modal biosensor, which comprises a wearable device, and is characterized in that a multi-modal biosensing assembly and an ARM evaluation processor are arranged in the wearable device; the multi-modal biosensing assembly forms a first evaluation table based on sensing difference data of multi-modal sensing data in different acquisition stages within the wearing time of a user; wherein the first evaluation table is used for performing primary and secondary sensing signal scoring on the user according to the sensing difference value data; and the ARM processor is used for constructing a preset spleen and kidney deficiency evaluation model according to the primary and secondary induction scoring signals, and dynamically outputting visual spleen and kidney evaluation information of the user according to the spleen and kidney deficiency evaluation model.

Description

Technical Field

[0001] The present invention relates to the field of medical health monitoring, and in particular to a dynamic evaluation system for spleen and kidney deficiency syndrome based on a multimodal biosensor. Background Art

[0002] Lung cancer is one of the most morbid and fatal malignancies worldwide, with non-small cell lung cancer (NSCLC) accounting for over 80% of cases. Patients with advanced NSCLC often suffer from spleen and kidney deficiency, leading to insufficient vital energy. Clinically, these patients present with symptoms such as fatigue, soreness in the waist and knees. Traditional Chinese medicine (TCM) syndrome differentiation relies on physician experience and lacks objective quantitative criteria. While Western medical treatments such as chemotherapy (NP regimen) and immunotherapy (PD-1 inhibitors) can slow tumor progression, they are associated with high toxicity and side effects, and poor patient quality of life. Clinical observations of a combination of Jianpi Yishen Decoction (Jianpi Yishen Decoction) and NP regimen for the treatment of advanced NSCLC have shown that the combination significantly improves the overall efficacy rate (93.33% vs 80.00%) and reduces toxicity and side effects. However, efficacy assessment still relies on static imaging examinations and intermittent blood tests, which cannot reflect the dynamic correlation between symptoms and tumor micrometastasis in real time. In the medical review article "Research Progress on the Detection of Circulating Tumor Cells in the Clinical Application of Lung Cancer", it is pointed out that traditional Chinese medicine symptom scoring relies on manual interviews, which has defects such as strong subjectivity and lack of dynamic monitoring. Studies have shown that the number of CTCs in patients with spleen and kidney deficiency syndrome is significantly negatively correlated with the symptom scores of fatigue, sallow complexion, etc. (R 2 =0.82), but existing technologies lack real-time monitoring tools that integrate traditional Chinese and Western medicine data. Summary of the Invention

[0003] In response to the above problems, the present invention proposes a dynamic assessment system for spleen and kidney deficiency syndrome based on multimodal biosensors. By integrating sensor technologies such as pulse wave, infrared thermal imaging, and speech semantic analysis, it quantifies the TCM symptom score in real time and associates it with CTC dynamic change data, providing accurate and real-time efficacy evaluation support for integrated Chinese and Western medicine treatment.

[0004] In a first aspect, a dynamic evaluation system for spleen and kidney deficiency syndrome based on a multimodal biosensor comprises a wearable device having a built-in multimodal biosensor component and an ARM evaluation processor;

[0005] The multimodal biosensor assembly forms a first evaluation table based on the sensor difference data of the multimodal sensor data in different acquisition stages during the user's wearing time; wherein the first evaluation table is used to score the user's primary and secondary sensing signals according to the sensor difference data;

[0006] The ARM processor is used to construct a preset spleen and kidney deficiency assessment model based on the primary and secondary sensing score signals, and dynamically output the user's visualized spleen and kidney assessment information based on the spleen and kidney deficiency assessment model.

[0007] In the embodiments of the present application, by integrating multimodal biosensor components and wearable devices such as ARM evaluation processors, it is possible to continuously collect multiple physiological parameters such as pulse waveform, skin conductivity and body surface temperature. Then, based on the quantitative evaluation of the sensor difference data in the time series change process, combined with the spleen and kidney deficiency evaluation model, a dynamic evaluation of the user's physical condition is achieved. Through cross-validation of multi-dimensional biometric features and the fusion performance of multi-dimensional biometric features, the user's physical condition information is output, so that doctors can have more accurate diagnosis and treatment opinions when treating users.

[0008] In an embodiment of the present application, the multimodal biosensor component collecting sensor difference data includes:

[0009] A dynamic difference threshold model is constructed based on the user's vital signs baseline, and time series segmentation is performed based on the joint time and frequency domain features of multimodal data in the continuous acquisition stage;

[0010] The dynamic difference threshold model normalizes the phase difference between the photoplethysmography and the galvanic skin response through a sliding window mechanism to generate an abnormal fluctuation identifier within the stage;

[0011] The abnormal fluctuation identifier triggers the adaptive acquisition frequency adjustment module, which synchronously adjusts the sampling intervals of the electromyographic signal and the temperature sensor according to the user's current motion posture classification result, so that the difference data generation cycle matches the user's physiological state migration rate.

[0012] This application realizes the acquisition of individualized spleen and kidney deficiency judgment data through the combination of a dynamic difference threshold model and an adaptive acquisition mechanism. The dynamic difference threshold model constructs a personalized fluctuation standard based on the user's physical baseline to solve the problem of misjudgment of people with different physical constitutions by traditional fixed thresholds. For example, in view of the fact that the baseline fluctuation of the skin electrical response of people with yang deficiency is small, the abnormal fluctuation judgment range can be automatically relaxed, and physiological fluctuations can be misjudged as pathological signals. The adaptive acquisition frequency adjustment module optimizes the data acquisition strategy in real time through motion posture classification. When it is detected that the user is in motion, the electromyographic signal sampling interval is increased, and the temperature sensor sampling frequency is reduced at the same time to ensure the data validity under motion interference and reduce device power consumption.

[0013] In an embodiment of the present application, the different acquisition stages further include:

[0014] In the initial stage of wearing, the standard deviation of the RR interval of the user's ECG signal and the rate of change of blood oxygen saturation are extracted to form a baseline feature vector;

[0015] During the continuous acquisition phase, the spatiotemporal similarity between the current photoplethysmography morphology and the reference vector is calculated using a dynamic time warping algorithm.

[0016] This application extracts the user's unique RR interval standard deviation of the ECG signal and the rate of change of blood oxygen saturation as a reference vector in the initial stage of wearing, and constructs a personalized assessment baseline based on the user's physiological fingerprint. It solves the problem of misjudgment caused by individual differences in traditional technologies. The dynamic time warping algorithm is used to identify signal distortion caused by emotional fluctuations or environmental interference by matching the spatiotemporal characteristics of the pulse wave morphology, making the sensing results of the spleen deficiency-related pulse characteristics more accurate. When the skin conductance response curve shows abnormal changes, the multimodal feature weighted fusion mechanism will be used to take the product of the electromyographic signal spectrum energy and the body temperature change as the weight factor, which can dynamically verify the user's identity, prevent confusion of user identity data, and capture the complex signs of chills and muscle relaxation that are unique to spleen and kidney deficiency syndrome.

[0017] In an embodiment of the present application, the multimodal biosensing component is further used to:

[0018] Real-time analysis of the time-domain cross-correlation matrix of multimodal sensors, and generation of dynamic compression coding strategies based on the covariance change rate of sensor data during wear time;

[0019] When the delay difference between the photoplethysmography and the ECG signal exceeds the preset alarm threshold in the cloud, the layered encryption transmission of the multimodal data stream is initiated, using a dynamic key distribution mechanism based on user motion state recognition;

[0020] The biometric topology map is reconstructed based on the received encrypted data packets, and a visual report of the user's physiological state migration trajectory is generated by comparing the topological structure similarity of adjacent acquisition stages.

[0021] The dynamic compression coding strategy of this application is to determine the covariance change rate under multimodal sensor data, and adaptively eliminate redundant information. While reducing the amount of continuously monitored data, it also retains key physiological fluctuation characteristics, such as the characteristic peak of the morning pulse. The layered encryption transmission mechanism can dynamically bind the user's motion state to the key distribution, and automatically switch to a lightweight encryption algorithm when intense exercise is detected. Under the premise of ensuring data security, it can reduce processor energy consumption. The biometric topology reconstruction technology can intuitively present the pathological evolution path of spleen dysfunction and kidney failure to absorb qi by comparing the similarity of the thermodynamic distribution of meridians in adjacent stages.

[0022] In an embodiment of the present application, the first evaluation table is configured with an initial wearing stage and a continuous collection stage; wherein,

[0023] During the wearable initialization phase, a multimodal baseline database with timestamp alignment is established, and the standard deviation of the signal-to-noise ratio of different sensor types is calculated to assign weight coefficients to primary and secondary signals.

[0024] During the continuous acquisition phase, the main signal credibility score was calculated by the mutual information between the first-order derivative of the skin conductance response curve and the harmonic components of the photoplethysmography;

[0025] When the covariance value between the spectrum energy of the electromyographic signal and the rate of change of body temperature exceeds the dynamic threshold, the secondary signal compensation algorithm is triggered, and the attitude angle change rate of the inertial measurement unit is weighted and injected into the evaluation model to generate a primary and secondary signal fusion scoring matrix.

[0026] This application uses timestamp-aligned multimodal baseline data at the initial stage to prevent feature drift that occurs in existing technologies due to asynchronous sensor data acquisition. The primary and secondary signal weighting based on the standard deviation of the signal-to-noise ratio automatically downgrades the scoring of low-quality sensor data, ensuring the dominance of key physiological parameters. The secondary signal compensation algorithm uses covariance analysis between the rate of change of posture angle and physiological parameters to distinguish motion interference from true pathological signals.

[0027] In an embodiment of the present application, the primary and secondary sensing signal scores include:

[0028] The signal priority index is generated by convolution operation of the RR interval variation coefficient of the ECG signal and the blood oxygen saturation trend slope;

[0029] When the secondary signal score exceeds the first threshold of the primary signal, the multi-channel feature fusion mechanism is activated, and the dynamic time warping algorithm is used to align the respiratory frequency phase difference and skin impedance change curves;

[0030] The first evaluation table has a built-in Bayesian network optimization module, which dynamically updates the primary and secondary signal weight distribution strategy according to the user's current exercise intensity level, so that the scoring results are synchronized in time and space with the physiological state migration rate.

[0031] This application realizes dynamic priority determination instead of fixed priority through the convolution operation of the coefficient of variation during the RR period of the ECG signal and the slope of the blood oxygen saturation trend, thereby improving the monitoring sensitivity; the dynamic time correction algorithm aligns the respiratory phase difference and the skin impedance curve, which can eliminate the probability of false positives, and finally dynamically adjusts the weights of the primary and secondary signals through the grading of exercise intensity to achieve accurate collection of pathological characteristics.

[0032] In an embodiment of the present application, the first evaluation table includes: a three-dimensional evaluation vector including confidence intervals of primary and secondary signals, matched to a user's historical physiological feature vector library via a cloud server;

[0033] When it is detected that the main score of the electromyographic signal is lower than the secondary signal heat map threshold for multiple consecutive sampling cycles, the multimodal collaborative alarm mechanism is triggered and the sampling frequency ratio of the inertial measurement unit and the optical sensor is synchronously adjusted;

[0034] The alarm mechanism dynamically reconstructs the biometric topology map based on the signal attenuation index generated by the evaluation table and generates an abnormal contact status prompt instruction for the wearable device.

[0035] This application can characterize the user's condition in a three-dimensional manner, rather than a two-dimensional assessment method. The entire process can achieve quantitative tracking and realize the spatiotemporal evolution of the user's symptoms, and the results are more accurate.

[0036] In an embodiment of the present application, the steps of constructing the spleen and kidney deficiency assessment model include:

[0037] The pulse trough interval variation coefficient and foot three yin meridian temperature gradient data in the primary and secondary induction scores were extracted through a multi-channel feature fusion mechanism.

[0038] The coefficient of variation of pulse trough intervals and temperature gradient data of the three foot yin meridians are loaded into the ARM processor. Through the built-in dynamic weight allocation algorithm of the ARM processor and the weight ratio of spleen deficiency / kidney deficiency symptoms in the TCM syndrome element matching library, a triple burner operation state parameter matrix is ​​constructed; when it is detected that the bioimpedance covariance value of the Zusanli and Taixi acupoints exceeds the preset meridian imbalance threshold, the visualization output module is triggered to generate a dynamic meridian topology map, and the topology map is superimposed to display the spatiotemporal correlation heat map of the spleen and kidney deficiency index and the energy flow changes of the Ren and Du meridians.

[0039] This application is not an evaluation of a single parameter, but a fusion evaluation of multi-dimensional data. It integrates traditional Chinese medicine theory and bioimpedance data, and dynamically evolves meridian topology. It can realize thermal map mapping of the user's body state and achieve high spatiotemporal resolution of the evolution of deficiency syndrome, and the results are more accurate.

[0040] In an embodiment of the present application, the dynamic output of the user's visualized spleen and kidney assessment information includes:

[0041] The multimodal data fusion unit performs a time-domain convolution operation on the R-wave interval variation rate of the ECG signal and the second-order derivative of the skin conductance response curve to generate the spleen and kidney function coordination index;

[0042] The ARM processor is equipped with a Bayesian network optimization module, which automatically activates the tongue image feature transfer learning model when the ratio of the primary and secondary induction scores exceeds a preset multiple of the TCM constitution classification threshold;

[0043] The visual output interface integrates an augmented reality module, maps the spleen and kidney assessment index to a three-dimensional human meridian model, and synchronously outputs the difference in virtual and real vibration frequencies of the Taiyin spleen meridian of the foot and the Shaoyin kidney meridian of the foot through a tactile feedback unit.

[0044] This application dynamically associates the ratio of primary and secondary scores with the threshold for TCM constitution classification. When the ratio exceeds the preset threshold, the tongue image transfer learning model is automatically activated, supplementing the diagnosis with tongue coating texture features to reduce misdiagnosis rates. An augmented reality module, combined with tactile feedback, converts the abstract spleen and kidney deficiency index into perceptible meridian vibration differences, resulting in more accurate clinical results.

[0045] In an embodiment of the present application, the dynamically outputting the user's visualized spleen and kidney assessment information further includes:

[0046] The ARM processor compares the spectrum characteristics of the current user's tongue coating image with the pulse parameters in real time to build a dynamic evolution map of spleen and kidney deficiency.

[0047] According to the dynamic evolution map of spleen and kidney deficiency, when the spleen deficiency confidence score is lower than the kidney deficiency heat map threshold for multiple consecutive collection cycles, the user's physical fingerprint map is triggered to output a personalized conditioning plan;

[0048] Among them, the user's physical fingerprint map represents the medicine and food map suitable for the user's physical constitution. The personalized conditioning plan dynamically displays the Ren / Dai meridian energy flow reconstruction process through a multimodal interactive interface, and adjusts the recommended weight coefficient of Chinese medicine meridians according to the user's real-time motion posture data.

[0049] The multimodal interactive interface of this application uses energy flow field reconstruction technology to intuitively display the virtual and real evolution of the Ren and Dai meridians, and automatically adjusts the weights of Chinese medicine meridians according to the user's real-time movement posture, thereby improving the doctor's judgment of treatment plans and issuing better diagnosis and treatment opinions.

[0050] Other features and advantages of the present invention will be described in the following description, and in part will become apparent from the description, or will be understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures particularly pointed out in the written description and the accompanying drawings.

[0051] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0053] Figure 1 This is a system composition architecture diagram of a dynamic assessment system for spleen and kidney deficiency syndrome based on a multimodal biosensor in an embodiment of the present invention;

[0054] Figure 2 This is a diagram of the training deployment architecture of a traditional medical large model in an embodiment of the present invention;

[0055] Figure 3 A diagram showing the process of constructing a biometric template according to an embodiment of the present invention;

[0056] Figure 4 This is a process diagram of secure data transmission and cloud reconstruction in an embodiment of the present invention;

[0057] Figure 5 A process diagram of dynamic weight allocation of primary and secondary signals in an embodiment of the present invention;

[0058] Figure 6 A diagram illustrating a process of dynamically adjusting signal priority levels in an embodiment of the present invention;

[0059] Figure 7 3D evaluation and abnormal alarm implementation process diagram in an embodiment of the present invention;

[0060] Figure 8 A diagram showing the calculation process of TCM meridian energy analysis in an embodiment of the present invention;

[0061] Figure 9 This is a process control diagram of augmented reality interaction and tactile feedback in an embodiment of the present invention;

[0062] Figure 10 This is a process diagram of dynamic evolution and personalized adjustment in an embodiment of the present invention. DETAILED DESCRIPTION

[0063] The preferred embodiments of the present invention are described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0064] First, some nouns involved in the embodiments of this application are briefly introduced:

[0065] Deficiency of both spleen and kidney: A common syndrome in Traditional Chinese Medicine theory, which refers to the simultaneous dysfunction of both the spleen and kidney, manifested as abnormalities in digestion, reproduction, and water metabolism.

[0066] Wearable devices: including watches, bracelets, smart sensor T-shirts, smart vests, smart insoles, electric field therapy devices, knee pads, etc., which can collect users' routine health data such as electrocardiogram signals, pulse waves, electromyography signals and body temperature.

[0067] Multimodal biosensing components: devices used to collect physiological signals from the wearer.

[0068] ARM evaluation processor, the built-in processor of the wearable device, is used to connect to the cloud and process multimodal sensor data based on the local spleen and kidney deficiency evaluation model built into the wearable device;

[0069] User vital signs baseline, the baseline of the wearable user's physical condition and health status, is used to determine the lowest value of the user's physical condition data collected by the non-biological sensing component when the user is in a healthy state.

[0070] Figure 2 The figure shows the deployment diagram of traditional medical big model training. The training and verification of the model rely on blood oxygen data, electrocardiogram data and pulse wave data. Through the combination of these data, the health status of the human body is used to realize the parameter configuration of the medical big model. The spleen and kidney deficiency assessment model of this application is also a medical big model. In the process of evaluating the user's physical condition, the traditional medical big model determines the specific parameters of the deployed big model according to the type of physical sign analysis ordered by the user, that is, controls which aspect the deployed big model is evaluated for. For example, if the user wants to evaluate the cardiovascular health status, the deployment method is to set the medical big model to mainly input the cardiovascular-related physical sign data features in the user's physical sign data, thereby achieving a comprehensive evaluation.

[0071] Although traditional methods can collect data, they can only determine which aspect of the user's health status they want to test based on the user's intention, and then extract corresponding features from the user's vital sign data to achieve detection. Therefore, the accuracy and professionalism of the detection are not comprehensive, and the evaluation results are directly output during the evaluation process.

[0072] This application aims to solve the defects of traditional technical solutions. Figure 1 ,The wearable device serves as the core carrier, which includes two major modules: multimodal biosensor components and ARM evaluation processors.

[0073] The multimodal biosensor component forms a first evaluation table based on the sensor difference data of the multimodal sensor data in different acquisition stages during the user's wearing time;

[0074] The first evaluation table is used to score the user's primary and secondary sensing signals based on the sensing difference data;

[0075] The multimodal biosensor component is responsible for collecting the user's physiological signals, such as photoplethysmography and skin conductance. It can use three types of devices: plethysmography (PPG), microcurrent skin response (EDR), and infrared thermal imaging (IRT). Different collection stages are collected through sampling frequency or sampling time, or user-defined collection stages.

[0076] The first evaluation table belongs to the middle layer of signal processing and is used to score and classify the primary and secondary signals. For example, the ECG signal is listed as the primary signal and the temperature sensor signal as the secondary signal.

[0077] The ARM evaluation processor is the inference and decision-making center, which inputs the scoring results into a preset medical model. The medical model belongs to the Traditional Chinese Medicine syndrome model. This application uses the spleen and kidney deficiency evaluation model to ultimately generate visual evaluation information. The spleen and kidney deficiency evaluation model of this application is based on the training of a general Traditional Chinese Medicine syndrome model.

[0078] During the specific implementation process, multimodal sensor data is collected by different wearable devices, or different types of data are collected by the same wearable device, and then difference processing is performed. Difference processing is to perform difference processing on data of the same type, such as electrocardiogram signal difference and pulse wave difference, and then score classification is performed. After scoring and classification, model matching is performed, and finally visual output is achieved to realize closed-loop monitoring.

[0079] The wearable device of the present application has a built-in multimodal biosensor component for continuously collecting the user's physiological signals.

[0080] In the initial stage of wearing, baseline data is established. For example, by extracting the time domain variation characteristics of the ECG signal and the trend of blood oxygen saturation changes, a user-specific physiological characteristic template is formed.

[0081] During the continuous monitoring process, the multimodal biosensor component is based on dynamic difference analysis, for example: comparing the waveform phase difference of the photoplethysmography and the instantaneous change rate of the skin conductance response through a sliding window mechanism to generate a first evaluation table reflecting the user's real-time physiological state.

[0082] In this application, the ARM evaluation processor is combined with the TCM syndrome model. For example, the characteristics of sluggish Qi and blood circulation related to spleen deficiency are correlated with the characteristics of abnormal energy metabolism related to kidney deficiency. Then, the primary and secondary signal scores are integrated based on the dynamic weight allocation algorithm to generate a visual evaluation report.

[0083] In this process, multimodal signal fusion can eliminate the environmental interference of a single sensor, thereby improving the stability of the assessment; dynamic difference analysis can capture subtle changes in physiological status, for example: identifying the trend of spleen and kidney imbalance in the early stages of qi and blood fluctuations; the spleen and kidney deficiency assessment model is combined with biosensor data to combine the dialectical thinking of traditional Chinese medicine and output user-visualized spleen and kidney assessment information in a way that is closer to clinical monitoring.

[0084] In this application, by adopting the traditional method of multiple acquisition stages, the symptom assessment is performed by sensor difference rather than absolute value in the prior art, so that there will be no misjudgment of health status assessment caused by individual physiological baseline differences. In the scoring stage, through the hierarchical weight distribution method, sudden abnormal signals can be combined with gradual changes. In the recognition of complex symptoms, the fusion of the two can also improve the recognition accuracy. Then, the visual spleen and kidney assessment information based on time series is output to provide doctors with assessment data in the spatiotemporal dimension. This application only provides assessment data for doctors and does not directly analyze the user's disease.

[0085] This application uses multimodal sensing components to achieve dynamic quantitative scoring by taking advantage of the difference data collected at different stages. This dynamic quantitative scoring is then combined with an ARM processor and a traditional Chinese medicine model to output dynamic visualization data on spleen and kidney deficiency.

[0086] See Figure 1 , this application has dynamic difference analysis and adaptive acquisition in the process of collecting multimodal sensor data:

[0087] A dynamic difference threshold model is constructed based on the user's vital signs baseline. The dynamic difference threshold model is a difference calculation threshold model based on the user's personalized physiological state sensor data. It is mainly constructed based on a personalized biometric baseline library. In the initial calibration stage, multimodal data of the user in a resting state can be collected, and then a baseline distribution area curve on the user's biometric characteristics can be constructed. For example: in the skin electrical response, the absolute integral of the first-order derivative of the GSR signal in each sliding window is calculated as an indicator of sympathetic nerve activation. If the standard deviation of the integral value of multiple consecutive windows is less than a certain normal value, it is used as a benchmark stable period. Then, according to the mean phase difference of the benchmark stable period, the dynamic threshold anchor point is determined, so that the calculation of the difference data is more in line with the individual characteristics of the user.

[0088] Perform time series segmentation based on the time-domain and frequency-domain joint features of multimodal data in the continuous acquisition phase of the dynamic difference threshold model, and determine the sensor difference data after the time series segmentation;

[0089] Specifically, the dynamic difference threshold model configures time-series segmentation points based on joint time-domain and frequency-domain features, such as the spectral energy distribution of the pulse wave's harmonic components, and monitors signal fluctuations in real time through a sliding window mechanism. If the signal difference falls outside the reasonable range of physiological fluctuations, such as a transient sudden change in skin conductance, the adaptive adjustment module is triggered. Then, through adaptive acquisition frequency, the sensor sampling method is adjusted according to the user's motion state, such as static and walking dynamic optimization, and the myoelectric signal sampling rate is increased during strenuous exercise to capture muscle state. This comprehensive approach achieves dynamic optimization of data acquisition, balancing signal accuracy and device power consumption to adapt to different activity scenarios.

[0090] Then, the dynamic difference threshold model normalizes the phase difference between the photoplethysmogram and the skin conductance response through a sliding window mechanism to generate an abnormal fluctuation identifier within the stage; the dynamic difference threshold model of the present application adopts a multi-dimensional signal joint analysis method, which extracts the harmonic components of the photoplethysmogram and the transient response characteristics of the skin conductance through time-frequency transformation to construct a cross-modal correlation indicator.

[0091] When abnormal physiological fluctuations are detected, such as a sudden increase in the second-order derivative of the skin conductance curve or a distortion in the pulse wave morphology, the system triggers the adaptive acquisition strategy.

[0092] During the specific implementation of this application, the sensor sampling frequency will be dynamically adjusted according to the user's motion posture classification results. For example, during strenuous exercise, the sampling density of electromyographic signals will be prioritized, and then the energy efficiency of the temperature sensor will be optimized in a resting state. The real-time response capability to abnormal fluctuations can capture sudden health risks in a timely manner; the sampling strategy that is adaptive to the motion state takes into account both data accuracy and device battery life. For example, it can continuously provide stable monitoring during marathon training; the collaborative analysis of multimodal signals can reduce misjudgments caused by motion artifacts and improve the reliability of outdoor use scenarios.

[0093] When implementing this application, the core is to combine phase difference normalization with motion posture classification to improve the accuracy of dynamic evaluation, that is: based on the physiological state migration rate related to TCM syndrome types, the sampling frequency is adjusted to achieve targeted dynamic matching of difference data and user physiological state.

[0094] The dynamic difference threshold model and multimodal sensing components form a closed-loop, high-precision data acquisition system. Sliding window normalization eliminates individual difference noise, ensuring that the primary and secondary sensing scores are weighted and matched to the ARM evaluation processor.

[0095] For example, if it is detected that the user is sitting quietly at work, the myoelectric sampling interval is reduced, and based on the relaxation data of the spleen meridian-related muscle groups, specific symptoms such as drowsiness after eating are identified.

[0096] This application proposes a method of constructing a certificate using a biometric template. Figure 3 :

[0097] In the initial stage of wearing, that is, the stage of determining the baseline feature vector, this application extracts the standard deviation of the RR interval of the user's ECG signal and the rate of change of blood oxygen saturation to form the baseline feature vector;

[0098] In practical implementation, a user's identity reference vector is determined by extracting stable features of the ECG signal and blood oxygen saturation, such as the coefficient of variation of the RR interval. Dynamic time warping is used to align multimodal signals and identify timing differences, such as the phase offset between the respiratory signal and the pulse wave.

[0099] Then, the authentication weight factor is determined, and the spectral energy of the electromyographic signal and the trend of body temperature change are integrated to make the identity recognition unique. Through the spectral energy of the electromyographic signal and the trend of body temperature change, the fluctuation characteristics of the physiological signal are combined with the identity authentication, so that traditional biometrics will not fail in dynamic scenarios.

[0100] During the continuous acquisition phase, user identity verification is combined with a multimodal biometric fusion mechanism to align the dynamic time warping algorithm with the timing characteristics of the ECG signal and photoplethysmography to determine the user's personalized biometric template.

[0101] During continuous monitoring, when abnormal changes are detected in the skin conductance signal (changes in the spectrum energy of the electromyographic signal), such as energy mutations or phase shifts in a specific frequency band, multimodal weight fusion is initiated, which nonlinearly combines the spectrum distribution characteristics of the electromyographic signal with the body temperature change trend to form a user-unique authentication factor. The authentication factor is obtained by a dynamic fusion mechanism of biometrics and can resist counterfeit attacks, such as by simulating a single signal, because it is impossible to crack a multi-dimensional authentication system.

[0102] The timing alignment technology of this application is used to eliminate recognition errors caused by individual physiological rhythm differences; the adaptive adjustment of multimodal weights enables high-precision authentication to be maintained even when the user sweats or has body temperature fluctuations.

[0103] This application's identity authentication mechanism uses a five-point central difference method based on the second-order inverse of the skin conductance response curve. This method synchronously collects myoelectric signals when the instantaneous rate of rise of the detection conductance exceeds a threshold, performs wavelet decomposition, and extracts the energy contribution of specific frequency bands. This energy contribution is then used as a weighting factor, and the user's sensor signals are adaptively collected by adjusting the parameters of the sliding time window and the user's activity.

[0104] The benchmark feature vector of this application is combined with the primary and secondary sensing signal scores for individualized calibration. For example, the user-specific SDNN range is used as the normalization standard for the heart rate variability score, and the pulse conditions of people with different physical conditions can be compared.

[0105] The dynamic time warping algorithm of this application is used to improve the data accuracy of pulse wave difference, reduce the amplitude difference calculation error related to spleen deficiency by eliminating time domain distortion interference, and the identity authentication weight factor and hierarchical scoring mechanism can constitute a double check. When the myoelectric energy is abnormally reduced, indicating that the device is off, the scoring weight of the corresponding sensor channel is automatically reduced to prevent erroneous data, and scoring is performed through the evaluation model.

[0106] This application proposes a method for secure data transmission and cloud reconstruction, see Figure 4 :

[0107] First, the multimodal biosensor assembly of the present application is also used to perform data compression on multimodal sensor data. In a specific implementation, the time-domain cross-correlation matrix of the multimodal biosensor assembly is analyzed in real time to calculate the covariance change rate of the multimodal sensor data during wear time. The covariance change rate can be used to determine the dynamic compression coding strategy, the coordinated change pattern of the time-domain cross-correlation matrix and the multimodal signal, and the monitoring of data anomalies. The time-domain cross-correlation matrix represents the temporal correlation of sensor data of different types and sources.

[0108] For example: if there is a correlation between the conduction delay of the electrocardiogram and the pulse wave, the presence of a sudden physiological event is determined based on the covariance change rate, and the physiological event is synchronously associated with the abnormal monitoring data. Therefore, during dynamic compression coding, a dynamic compression coding strategy of wavelet packet decomposition is adopted to retain the low-frequency approximate coefficients and truncate the high-frequency approximate coefficients to achieve zero distortion retention.

[0109] During real-time analysis, namely the calculation of the covariance and covariance rate of change of multimodal sensor data, the delay difference between the photoplethysmogram and the ECG signal is determined by comparing it with a preset alarm threshold. If the test difference exceeds the alarm threshold, the multimodal sensor data is encrypted and transmitted in layers. This layered encryption mechanism introduces dual-threshold triggering logic. If the delay difference between the photoplethysmogram and the ECG signal indicates a risk such as arrhythmia, the layered encryption mechanism is activated, transmitting the user's core physiological data. Based on the user's movement or physical status during the wearable phase, dynamic key distribution is implemented to generate encrypted data. The dynamic compression encoding strategy is a dynamic compression encryption strategy that performs encryption after key distribution by the dynamic key distribution mechanism.

[0110] In this application, layered encrypted transmission dynamically selects encryption strength based on the level of signal anomaly, for example, abnormalities in key vital signs, to ensure real-time data.

[0111] Specifically, the present application will also reconstruct the biometric topology map based on the received encrypted data, and generate a visualization report of the user's physiological state migration trajectory by comparing the topological structure similarity of adjacent acquisition stages. In this process, the biometric topology map is reconstructed in the cloud by comparing the data of adjacent stages, such as the acupoint impedance distribution pattern, to achieve the physiological state migration trajectory, so that the present application can track the health status across time dimensions while ensuring privacy security. The present application realizes the migration trajectory report of state vectors in multiple dimensions by calculating the graph structure similarity of adjacent acquisition stages. In the final visualization, the protection of the thermodynamic streamline diagram superimposed on the pulse waveform waterfall diagram is used to convert the process of the transmission of TCM syndromes into an interactive dynamic map.

[0112] The data security transmission of this application combines layered encryption and dynamic topology analysis. The coordinated change pattern of multimodal signals is monitored in real time through the time domain cross-correlation matrix. When data anomalies are detected, the physiological correlation between the pulse wave conduction time and the electrocardiogram signal is broken, triggering the encryption transmission mechanism. Then, a dynamic key is generated based on the user's motion state. For example, a lightweight encryption algorithm is used in low-intensity activities such as yoga, while a multiple verification mechanism is enabled in high-dynamic scenarios such as basketball.

[0113] The cloud reconstructs and analyzes biometric topology maps, for example, by comparing impedance distribution patterns at acupoints along the three Yin meridians of the foot over different time periods, enabling visual tracking of physiological status. Dynamic encryption strategies balance data security with transmission efficiency, prioritizing the real-time delivery of critical vital signs in emergency medical situations. Visual reconstruction of biotopology maps helps doctors quickly locate areas of meridian energy imbalance. A collaborative verification mechanism for multimodal signals effectively identifies risks of unauthorized device disassembly or tampering.

[0114] The dynamic compression coding of this application and the ARM evaluation processor form a closed loop of data processing, which improves the efficiency of primary and secondary induction score calculation by eliminating redundant data;

[0115] The layered encryption mechanism of this application can ensure the security of multimodal sensor data during transmission, especially preventing the leakage of sensitive physiological data during motion;

[0116] The biometric topology reconstruction technology of this application upgrades the visual evaluation information into a spatiotemporal dynamic map, which can trace and quantify the user's spleen and kidney status.

[0117] This application proposes a method for dynamically allocating weights of primary and secondary signals. Figure 5 :

[0118] In the wear initialization phase, a multimodal baseline database with timestamp alignment is constructed;

[0119] Then, the weight coefficients of the primary and secondary signals are assigned based on the standard deviation of the signal-to-noise ratio of different sensor types. In this application, the baseline database assigns an initial weight to each sensor. For example, the photoelectric capacitance sensor has a higher weight than the temperature sensor. The weight coefficients of the primary and secondary signals are then assigned based on the signal-to-noise ratio and signal stability.

[0120] The main signal credibility score is determined by the mutual information between the skin conductance response curve and the pulse wave harmonic components, and is used to adjust the signal priority.

[0121] The sub-signal compensation algorithm is used to address sudden increases in sub-signal quality, such as significant changes in the spectral energy of the electromyographic signal. Inertial measurement data is weighted and injected into the evaluation model to address evaluation bias caused by sensor performance fluctuations and enhance robustness. During the initialization phase, a baseline weighting model is established based on the sensor's signal-to-noise ratio characteristics, for example, assigning higher confidence to photoelectric capacitance sensors with strong anti-interference capabilities.

[0122] During the continuous acquisition phase, the present application calculates the credibility score of the main signal by using the mutual information between the first-order derivative of the skin conductance response curve and the harmonic components of the photoplethysmogram. That is, during the continuous acquisition phase of the present application, the skin conductance response curve and the harmonic components of the photoplethysmogram will have overlapping correlated information. Based on the correlated information of the two, the present application determines the credibility of the main signal and outputs a credibility score result.

[0123] During continuous monitoring, when the secondary signal quality exceeds a preset threshold, such as a sudden increase in EMG spectrum energy accompanied by abnormal body temperature fluctuations, the compensation algorithm is triggered. This involves spatially and temporally aligning the posture change data from the inertial measurement unit with the primary physiological signal to generate a fusion evaluation matrix, which provides a confidence score for the primary signal.

[0124] The dynamic weight allocation mechanism enhances robustness against fluctuations in sensor performance, for example, automatically switching to a backup signal source when the electrode pads are in poor contact; the multi-dimensional compensation algorithm ensures continuous monitoring of key physiological parameters and avoids evaluation bias caused by data interruptions; and the spatiotemporal alignment technology eliminates the impact of motion artifacts on core indicators and improves monitoring quality in mobile scenarios.

[0125] This application proposes a method for scoring primary and secondary sensing signals with dynamic adjustment of signal priority. Figure 6 :

[0126] First, based on the standard deviation of the RR interval of the user's ECG signal and the rate of change of blood oxygen saturation, a signal priority index is generated through a convolution operation; the standard deviation of the RR interval of the user's ECG signal can determine the coefficient of variation of the RR interval of the ECG signal, and the rate of change of blood oxygen saturation is used to determine the trend slope of blood oxygen saturation. Through the convolution operation of the two, a signal priority index is generated. The signal priority index is used to quantify the confidence of different signals. For example, the coefficient of variation of the ECG signal can determine the autonomic nervous state. In the process of generating the signal priority index, the present application adopts an optimized reference signal setting method guided by physiological rhythms. For example, by analyzing the variability of the ECG signal and the circadian rhythm characteristics of blood oxygen saturation, a personalized signal evaluation benchmark is established.

[0127] Signal priority index. During signal detection, if the secondary signal score exceeds the first threshold of the main signal, the multi-channel feature fusion mechanism is activated, and the dynamic time warping algorithm is used to align the respiratory frequency phase difference and the skin impedance change curve; the first threshold is the preset difference between the main signal and the secondary signal. The scoring result of the secondary signal needs to be lower than the main signal. When the scoring result of the secondary signal score exceeds the threshold of the main signal, the index value exceeds the threshold to trigger multi-channel fusion. In this application, multi-channel feature fusion is applicable when the importance of the secondary signal is increased. For example: the process of nonlinearly stacking the time-varying characteristics between the respiratory frequency and skin impedance uses dynamic time warping to align the signal phase.

[0128] Finally, the first evaluation table has a built-in Bayesian network optimization module, which dynamically updates the primary and secondary signal weight distribution strategy according to the user's current exercise intensity level, so that the scoring results are synchronized in time and space with the physiological state migration rate.

[0129] In this application, the Bayesian network optimization can update the weight distribution strategy in real time according to the intensity of user activities, such as resting, exercising, etc., and perform dynamic adaptation of the evaluation model to the user's real-time status to prevent misjudgment caused by fixed weights.

[0130] In this application, during the convolution operation, a separated convolution kernel method can be used. The first layer extracts frequency domain features, the second layer determines the time domain change pattern through blood oxygen saturation, and the second layer of convolution kernel fusion generates a priority index.

[0131] When the sub-signal quality changes significantly, such as when the correlation between respiratory rate and skin impedance increases, the system initiates multi-channel feature fusion.

[0132] The specific implementation includes using dynamic time warping technology to align the phase differences of different physiological signals, and updating the weight coefficients of each signal in real time through the Bayesian network.

[0133] The circadian rhythm-guided assessment strategy enables the system to better align with the body's natural metabolic patterns, for example accurately identifying the characteristics of spleen deficiency during the post-meal digestive period;

[0134] Dynamic phase alignment technology eliminates the timing misalignment problem during multi-modal signal acquisition;

[0135] The continuous optimization capability of the Bayesian network enables the weight distribution to evolve automatically with the user's health status, enhancing the adaptability of long-term monitoring.

[0136] In this application, the primary and secondary signal scoring mechanism is optimized through priority indexing, and the sensitivity of spleen deficiency-related cardiovascular indicators can be enhanced through convolutional feature extraction; the alignment mechanism improves the spatiotemporal consistency of multimodal difference data, and improves the accuracy in the evaluation of complex syndromes. The dynamic weight of the Bayesian network and the ARM processor form an intelligent evaluation closed loop, and the response speed is also higher when switching between exercise and resting states.

[0137] This application proposes a three-dimensional assessment and abnormal alarm method, see Figure 7 :

[0138] This application uses a three-dimensional evaluation vector containing the primary signal confidence interval, the secondary signal heat map threshold, and the fusion score for multi-dimensional status description. The collaborative alarm mechanism triggers the sensor sampling rate adjustment when signal attenuation persists, for example, when the electromyography signal strength continuously falls below the threshold.

[0139] In actual implementation, the historical physiological feature vector library is a deep learning database, and the main signal heat map threshold is generated through user historical data training. This application uses a Gaussian mixture model to jointly calibrate the spleen meridian night temperature drop rate and the kidney meridian morning impedance phase offset, and dynamically adjust the confidence interval boundary. When the main electromyography score is lower than the secondary signal heat map threshold for multiple consecutive sampling cycles (in actual implementation, the sampling cycle will be set, for example, 5) , it will be determined that a complex pathological state exists, and then the corresponding response mechanism will be triggered. The multimodal collaborative high-precision mechanism will generate different levels of response mechanisms based on the number of pathological signals, thereby achieving synchronous adjustment of the sampling frequency ratio of the inertial measurement unit and the optical sensor; synchronously activate redundant sensor channels to achieve a ratio design of synchronous enhancement or synchronous weakening of the sampling frequency of the inertial measurement unit and the optical sensor.

[0140] Finally, a multimodal alarm mechanism dynamically reconstructs the biometric topology map based on the signal attenuation index generated by the evaluation table, generating instructions to indicate abnormal contact status of the wearable device. Biometric topology reconstruction analyzes changes in contact impedance to identify device wear anomalies, such as sensor displacement or detachment. This application is suitable for long-term health monitoring scenarios and addresses device wear reliability issues.

[0141] The 3D evaluation vector construction module uses a multi-level anomaly detection mechanism. For example, it builds a 3D confidence interval model for primary and secondary signals by matching historical cloud data. When persistent signal attenuation is detected, such as when the EMG signal strength continuously falls below the secondary signal threshold, the system triggers a coordinated alarm.

[0142] In specific implementations, this application reconstructs a biometric topological map to analyze device wear status, for example, identifying sensor displacement or risk of dislodgment through changes in contact impedance. A three-dimensional confidence interval model enhances real-time representation of complex physiological states, such as Qi and blood deficiency associated with spleen deficiency and abnormal energy metabolism associated with kidney deficiency. Furthermore, a collaborative alarm mechanism enables dual monitoring of device status and physiological indicators. Topological map reconstruction technology rapidly locates sensor contact issues and reduces data distortion caused by improper wear.

[0143] In this application, the sliding window can be normalized into a three-dimensional evaluation vector to provide noise-suppressed baseline data, thereby reducing the error in the calculation of the main signal confidence interval. Furthermore, by combining motion-adaptive acquisition and sampling frequency adjustment mechanisms, when the device is in abnormal contact, the PPG sampling strategy optimized by posture classification results can be used to improve signal integrity under motion interference. It is also possible to integrate spatiotemporal features into hierarchical scoring to achieve three-dimensional topological analysis, and based on meridian node anomaly detection, identify specific features of spleen and kidney disease.

[0144] This application proposes the process of Chinese medicine meridian energy analysis, see Figure 8 ;

[0145] In the process of constructing a spleen and kidney deficiency assessment model, this application uses a multi-channel feature fusion mechanism to extract the pulse trough interval variation coefficient and the temperature gradient data of the three foot yin meridians in the primary and secondary induction scores. This application uses adaptive threshold segmentation to determine the trough feature points and the pulse trough interval variation coefficient. Then, combined with the gradient data of the corresponding body surface temperature sensors of the three foot yin meridians (i.e., the Taiyin spleen meridian, Shaoyin kidney meridian, and Jueyin liver meridian), the covariance matrix of the three meridian temperature gradients is calculated to determine the temperature gradient data of the three foot yin meridians.

[0146] This application is based on the computing function of the ARM processor, configures a dynamic weight allocation algorithm, and combines the symptom elements of spleen deficiency / kidney deficiency. For example, the spleen deficiency weight focuses on digestive function indicators to construct a triple burner energy matrix, that is, by extracting the temperature gradient and impedance change characteristics of specific acupoints, a triple burner energy operation status matrix is ​​constructed.

[0147] This application will also trigger the visualization output module to generate a dynamic meridian topology map when it is detected that the bioimpedance covariance value of the Zusanli and Taixi acupoints exceeds the preset meridian imbalance threshold; in the actual implementation process, the dynamic meridian topology map is a spatiotemporal correlation thermal map that superimposes the spleen and kidney deficiency index and the energy flow changes of the Ren and Du meridians. The dynamic topology map is generated after the meridian imbalance threshold is triggered, and the thermal map visualizes the energy flow differences of the spleen and kidney meridians. For example, by extracting the temperature gradient and impedance change characteristics of specific acupoints, a triple burner energy operation state matrix is ​​constructed. When signs of meridian imbalance are detected, such as the breakdown of the bioelectric signal synergy of the Zusanli and Taixi acupoints, the system generates a dynamic meridian topology map.

[0148] In practice, this application converts spleen and kidney syndrome elements into visual parameters of a thermal map, for example, using color gradients to reflect the fullness of the spleen meridian's qi and blood. The integration of Traditional Chinese Medicine (TCM) meridian theory and bioelectrical signals opens up new dimensions for health assessment, for example, early detection of chronic fatigue syndrome through impedance changes in the foot-Shaoyin Kidney Meridian. The dynamic rendering of this application's thermal map visualizes abstract TCM concepts, improving doctor-patient communication. The multi-dimensional analysis capabilities of the triple-burner energy matrix support the development of precise conditioning plans.

[0149] This application's motion state sensing combined with IMU posture classification can optimize the temperature gradient sampling strategy for the three Yin meridians of the foot, reducing temperature measurement errors under motion interference. Hierarchical evaluation and the triple energizer operating parameter matrix can form a multi-dimensional fusion evaluation system, improving the accuracy of monitoring data for spleen and kidney deficiency symptoms.

[0150] This application proposes a method of augmented reality interaction and tactile feedback, see Figure 9 :

[0151] Multimodal data fusion generates spleen and kidney function coordination indicators, such as the time domain correlation of electrocardiogram and skin conductance, to drive the rendering of three-dimensional meridian models.

[0152] In the actual implementation process, the interval variation rate between the R waves of the ECG signal and the second-order derivative of the skin conductance response curve are subjected to time-domain convolution operation to generate the spleen-kidney function coordination index. Specifically, the interval variation rate of the R waves of the ECG signal and the second-order derivative of the skin conductance response curve are subjected to time-domain convolution operation, so that the correlation between the ECG signal and the skin signal can output more correlation indicators in the identification of heart-spleen deficiency and heart-kidney disharmony.

[0153] The ARM processor is equipped with a Bayesian network optimization module. When the ratio of the primary and secondary induction scores exceeds the preset multiple of the TCM constitution classification threshold, the tongue image feature transfer learning model is automatically activated. Specifically, the Bayesian network optimization dynamically associates the primary and secondary score ratio with the TCM constitution classification threshold. If the ratio exceeds the preset threshold, such as the maximum value of the spleen deficiency constitution, the tongue image transfer learning model is automatically activated to supplement the diagnostic data through the tongue coating texture feature to reduce the misjudgment rate. In the visual output display, the ARM processor of this application maps the spleen and kidney function synergy index to the dynamic meridian topology map, combines tactile feedback, converts the kidney deficiency and excess index into a perceptible meridian vibration difference, and generates a visual meridian state map.

[0154] The augmented reality interaction module of this application realizes the spatial presentation of physiological data. For example, it generates spleen and kidney function coordination index through multimodal data fusion and then maps it to a three-dimensional human meridian model.

[0155] When the present application detects significant symptom characteristics, such as the coordinated abnormality of tongue characteristics and pulse parameters, the tactile feedback mechanism is activated.

[0156] In specific implementation, the difference in vibration frequency is used to reflect the deficiency or excess state of the spleen meridian and the kidney meridian. For example, high-frequency vibration indicates the risk of insufficient kidney essence.

[0157] The spatial mapping of dynamic meridian topology enhances the user's spatial cognition of health status, for example: intuitively showing the location of qi and blood blockage;

[0158] The tactile feedback mechanism establishes a new health warning channel suitable for scenes with limited vision;

[0159] The collaborative analysis of tongue and pulse patterns breaks through the limitations of traditional single-dimensional syndrome differentiation and improves the accuracy of TCM constitution identification.

[0160] The Bayesian network of this application forms a closed-loop optimization, which automatically reduces the tongue image migration weight when the secondary signal is abnormal to prevent doctors from misdiagnosing. The multimodal fusion augmented reality module can convert the results of the visual output into multi-sensory interactive information, so that the symptoms of spleen and kidney deficiency can evolve in time and space and be dynamically perceived.

[0161] This application proposes a method of dynamic evolution and personalized adjustment, see Figure 10 :

[0162] This application is based on the real-time comparison of the current user's tongue coating image spectrum characteristics and pulse parameters through the ARM processor to construct a dynamic evolution map of spleen and kidney deficiency. In actual implementation, the dynamic evolution map of spleen and kidney deficiency can be combined with the cloud-based Chinese medicine diagnosis case library. For example: the trend of spleen deficiency indicators and the prediction of the development direction of syndromes. This application generates personalized conditioning plans by combining the user's physical characteristics through the generation of suggestions based on the same origin of medicine and food, such as yang deficiency and phlegm-dampness, combined with real-time exercise status, exercise intensity, usable medicines and foods that users can eat.

[0163] The multimodal interactive interface of this application can visualize the changes in the energy flow of the Ren / Dai meridians and optimize the recommended weights of Chinese medicine meridians based on activity data.

[0164] In terms of the spleen deficiency confidence score, this application triggers a suggestion of food and medicine from the same source if the score is below the kidney deficiency heat map threshold for multiple consecutive collection cycles. The application then outputs a personalized conditioning plan based on the recommended plan and the user's constitution fingerprint. Based on the personalized conditioning plan, dynamic evolution prediction is performed. For example, by analyzing long-term monitoring data to establish a development trajectory model for spleen and kidney deficiency, when a symptom evolution trend is detected, such as when the spleen deficiency indicator is continuously weaker than the kidney deficiency parameter, the system triggers the food and medicine from the same source recommendation mechanism.

[0165] The specific implementation includes matching real-time motion data with the meridian characteristics of traditional Chinese medicine, such as dynamically adjusting the recommended weight of qi-invigorating drugs based on the current activity intensity.

[0166] In this application, long-term trend prediction is based on long-term testing to determine the user's optimal maintenance time and the best maintenance plan under the optimal maintenance time, such as strengthening spleen and stomach maintenance before the change of seasons;

[0167] The dynamic difference threshold is used to provide noise-suppressed pressure gradient data for pulse analysis, reducing the error in floating pulse judgment; motion posture perception is achieved through IMU data to optimize the adjustment of Chinese medicine meridian weights and improve the accuracy of recommendations; hierarchical evaluation and syndrome evolution maps can achieve multi-dimensional verification. If there is a conflict between primary and secondary scores, the tongue and pulse fusion analysis results are used first to reduce the misjudgment rate.

[0168] Obviously, those skilled in the art may make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if such changes and modifications fall within the scope of the claims and their equivalents, the present invention is intended to include such changes and modifications.

Claims

1. A dynamic assessment system for spleen and kidney deficiency syndrome based on multimodal biosensors, including a wearable device, characterized in that: The wearable device has a built-in multimodal biosensor component and an ARM evaluation processor; The multimodal biosensor assembly forms a first evaluation table based on the sensor difference data of the multimodal sensor data in different acquisition stages during the user's wearing time; wherein the first evaluation table is used to score the user's primary and secondary sensing signals according to the sensor difference data; The ARM evaluation processor is used to construct a preset spleen and kidney deficiency evaluation model based on the primary and secondary sensing score signals, and dynamically output the user's visualized spleen and kidney evaluation information based on the spleen and kidney deficiency evaluation model.

2. A dynamic assessment system for spleen and kidney deficiency syndrome based on a multimodal biosensor as claimed in claim 1, characterized in that: The multimodal biosensor component collects sensor difference data including: A dynamic difference threshold model is constructed based on the user's vital signs baseline. The time domain and frequency domain joint features of multimodal data in multiple stages are continuously collected according to the dynamic difference threshold model to perform time series segmentation and determine the sensor difference data after time series segmentation. The dynamic difference threshold model normalizes the phase difference between the photoplethysmography and the galvanic skin response through a sliding window mechanism to generate an abnormal fluctuation identifier within the stage; The abnormal fluctuation identifier triggers adaptive acquisition frequency adjustment, and synchronously adjusts the sampling interval of the electromyographic signal and the temperature sensor according to the user's current motion posture classification result, so that the generation cycle of the difference data matches the user's physiological state migration rate.

3. A dynamic assessment system for spleen and kidney deficiency syndrome based on a multimodal biosensor as claimed in claim 1, characterized in that: The different collection stages include an initial wearing stage and a continuous collection stage; The initial wearing stage is used to extract the standard deviation of the RR interval of the user's ECG signal and the rate of change of blood oxygen saturation to form a baseline feature vector; The continuous acquisition phase is used to calculate the spatiotemporal similarity between the current photoplethysmography morphology and the baseline feature vector through a dynamic time warping algorithm to determine the user's personalized biometric template.

4. A dynamic assessment system for spleen and kidney deficiency syndrome based on a multimodal biosensor as claimed in claim 1, characterized in that: The multimodal biosensing assembly is further configured to: Real-time analysis of the time-domain cross-correlation matrix of multimodal biosensor components to determine the rate of change of the covariance of multimodal sensor data during wear time; When the time delay difference between the photoelectric volumetric pulse wave and the electrocardiogram signal detected in real-time analysis exceeds the preset alarm threshold in the cloud, the layered encryption transmission of the multimodal data stream is initiated, and a dynamic key distribution mechanism based on the user's motion state recognition is used to generate encrypted data. The biometric topology map is reconstructed based on the received encrypted data packets, and the similarity of the topological structures in adjacent acquisition stages is compared to generate a visual report of the user's physiological state migration trajectory.

5. A dynamic assessment system for spleen and kidney deficiency syndrome based on a multimodal biosensor as claimed in claim 2, characterized in that: The first evaluation table is configured with an initial wearing stage and a continuous collection stage; wherein, During the wearable initialization phase, a multimodal baseline database with timestamp alignment is established, and the standard deviation of the signal-to-noise ratio of different sensor types is calculated to assign weight coefficients to primary and secondary signals. According to the weight coefficients of the primary and secondary signals, the primary signal credibility score was calculated by the mutual information between the first-order derivative of the skin conductance response curve and the harmonic components of the photoplethysmography during the continuous acquisition stage.

6. A dynamic assessment system for spleen and kidney deficiency syndrome based on a multimodal biosensor as claimed in claim 3, characterized in that: The primary and secondary sensing signal scores include: Generate a signal priority index through convolution operation based on the standard deviation of the RR interval of the user's ECG signal and the rate of change of blood oxygen saturation; According to the signal priority index, when the secondary signal score exceeds the first threshold of the primary signal, the multi-channel feature fusion mechanism is activated, and the dynamic time warping algorithm is used to align the respiratory frequency phase difference and skin impedance change curve.

7. A dynamic assessment system for spleen and kidney deficiency syndrome based on a multimodal biosensor as claimed in claim 5, characterized in that: The first evaluation table includes a three-dimensional evaluation vector of the primary and secondary signal confidence intervals, which is matched to the user's historical physiological feature vector library through the cloud server; wherein the historical physiological feature vector library includes the primary signal heat map and the secondary signal heat map threshold of the primary and secondary signal confidence intervals; When it is detected that the main score of the electromyographic signal is lower than the secondary signal heat map threshold for multiple consecutive sampling cycles, the multimodal collaborative alarm mechanism is triggered and the sampling frequency ratio of the inertial measurement unit and the optical sensor is synchronously adjusted; Among them, the multimodal alarm mechanism dynamically reconstructs the biometric topology map according to the signal attenuation index generated by the evaluation table, and generates an abnormal contact status prompt instruction for the wearable device.

8. A dynamic evaluation system for spleen and kidney deficiency syndrome based on a multimodal biosensor according to claim 6, characterized in that: The steps of constructing the spleen and kidney deficiency assessment model include: The pulse trough interval variation coefficient and foot three yin meridian temperature gradient data in the primary and secondary induction scores were extracted through a multi-channel feature fusion mechanism. The pulse wave trough interval variation coefficient and the foot three yin meridian temperature gradient data were loaded into the ARM processor. Through the built-in dynamic weight allocation algorithm of the ARM processor and the spleen deficiency / kidney deficiency symptom weight ratio in the TCM syndrome element matching library, the triple energizer operation state parameter matrix and the spleen and kidney deficiency evaluation model were constructed. Among them, according to the triple burner operation state parameter matrix, when it is detected that the bioimpedance covariance value of the Zusanli and Taixi acupoints exceeds the preset meridian imbalance threshold, a dynamic meridian topology map is generated.

9. A dynamic evaluation system for spleen and kidney deficiency syndrome based on a multimodal biosensor according to claim 8, characterized in that: The dynamically outputting the user's visualized spleen and kidney assessment information includes: The R-wave interval variation rate of the ECG signal is convolved with the second-order derivative of the skin conductance response curve in the time domain to generate the spleen and kidney function coordination index; Based on the spleen and kidney function synergy index and the Bayesian network optimization module configured with the ARM processor, the tongue image feature transfer learning model is automatically activated when the ratio of the primary and secondary induction scores exceeds the preset multiple of the TCM constitution classification threshold; Based on the ARM processor, the spleen and kidney function coordination index is mapped to the dynamic meridian topology map, and the difference in the virtual and real vibration frequencies of the Taiyin spleen meridian of the foot and the Shaoyin kidney meridian of the foot is synchronously output through tactile feedback.

10. The dynamic evaluation system for spleen and kidney deficiency syndrome based on multimodal biosensor according to claim 1, characterized in that: The method of dynamically outputting the user's visualized spleen and kidney assessment information further includes: The ARM processor compares the spectrum characteristics of the current user's tongue coating image with the pulse parameters in real time to build a dynamic evolution map of spleen and kidney deficiency. According to the dynamic evolution map of spleen and kidney deficiency, when the spleen deficiency confidence score is lower than the kidney deficiency heat map threshold for multiple consecutive collection cycles, the user's physical fingerprint map is triggered to output a personalized conditioning plan; Among them, the user's physical fingerprint map represents the medicine and food map suitable for the user's physical constitution. The personalized conditioning plan dynamically displays the Ren / Dai meridian energy flow reconstruction process through a multimodal interactive interface, and adjusts the recommended weight coefficient of Chinese medicine meridians according to the user's real-time motion posture data.

Images