A Health Monitoring, Analysis, and Early Warning Method and System Based on Multimodal Data Fusion

Abstract

This invention relates to the field of medical and health monitoring technology, and discloses a multimodal data fusion health monitoring, analysis, and early warning method and system. The method includes: synchronously collecting motion and heart rate data streams through a wearable device; generating a coupling vector between motion intensity and heart rate change rate within a short time window; accumulating this vector over a long period to form a coupling vector bundle; and assessing the stability of physiological regulation by evaluating changes in its morphological dispersion. This invention transforms micro-motion into a natural probe, achieving a leap from observing vital sign values ​​to assessing autonomic nervous system regulation capabilities through the dynamic evolution of the motion-heart rate coupling relationship. It can detect early functional decline before routine indicators become abnormal, and simultaneously ensures assessment reliability in complex environments through frequency domain feature filtering.

Description

Technical Field

[0001] This invention relates to a health monitoring, analysis, and early warning method and system based on multimodal data fusion, belonging to the field of medical and health monitoring technology. Background Technology

[0002] Existing technologies typically trigger alarms by continuously collecting physiological parameters such as heart rate and blood oxygen and comparing them with preset standard ranges. However, these methods have a fundamental limitation: they are essentially passive observations of physiological regulation results, rather than assessments of the body's intrinsic ability to maintain homeostasis. When elderly people living alone are in the incubation period of a disease, their autonomic nervous system has already begun to bear the burden, and their regulatory ability has begun to decline insidiously, but routine vital signs still show normal values, making it impossible for the system to detect early risks.

[0003] Specifically, existing technologies have three main shortcomings: 1. They focus too much on the absolute value of vital signs and ignore the dynamic stability of the regulatory process, resulting in a lag in the key warning window; 2. Non-physiological vibrations in complex scenarios, such as the bumps of vehicles, can contaminate motion signals and cause misjudgments; 3. Simply quantifying dispersion cannot distinguish between hyperactive and inhibitory instability modes, thus weakening the value of intervention guidance.

[0004] Industry attempts to improve accuracy by adding sensors or enhancing algorithms have exacerbated the conflict between power consumption and cost, failing to overcome the inherent limitations of the observation method. With the surge in demand for home health management, how to achieve early assessment of adjustment capabilities through low-power wearable devices has become the technical problem to be solved by this invention. Summary of the Invention

[0005] This invention provides a health monitoring, analysis and early warning method and system based on multimodal data fusion, the main purpose of which is to solve the problem that existing technologies cannot capture the early decline of the body's regulatory capacity through low-power solutions.

[0006] To achieve the above objectives, the present invention provides a health monitoring, analysis, and early warning method based on multimodal data fusion, comprising the following steps:

[0007] Step a: Synchronously collect a first physiological data stream and a second physiological data stream through a wearable device. The first physiological data stream contains user motion information, and the second physiological data stream contains user heart rate information. The collection is performed in a low-frequency non-uniform manner.

[0008] Step b: Within a specific short time window, perform frequency domain analysis on the first physiological data stream to determine whether the frequency distribution of its motion energy conforms to physiological low-frequency characteristics.

[0009] Step c: Only when the frequency distribution of exercise energy conforms to the physiological low-frequency characteristics, determine an exercise intensity scalar based on the first physiological data stream, and determine a heart rate scalar based on the second physiological data stream;

[0010] Step d: Pair the exercise intensity scalar and the heart rate change rate scalar to form a coupling vector characterizing the physiological coupling response within the short time window;

[0011] Step e: Over a specific long period of time, a series of coupling vectors are accumulated to form a coupling vector bundle, which reflects the intrinsic pattern of the user's physiological activities and heart rate response.

[0012] Step f: Evaluate the morphological dispersion of the coupled vector bundle, and determine the stability of the user's physiological coupling regulation based on the changing trend of the morphological dispersion over a continuous time period.

[0013] Step g: When the stability of physiological coupling regulation shows a declining trend, health risk warning information is generated.

[0014] Preferably, in step a, the first physiological data stream is acquired by a triaxial accelerometer sensor, and the second physiological data stream is acquired by a photoplethysmography (PPG) sensor.

[0015] Preferably, in step c, the exercise intensity scalar is the variance of the acceleration signal within a specific short time window, and the heart rate change rate scalar is the slope of the heart rate sequence within a specific short time window.

[0016] Preferably, in step b, the frequency domain analysis specifically involves performing a short-time Fourier transform on the first physiological data stream and determining whether the energy proportion of the high-frequency part in the short-time Fourier transform result is lower than a specific judgment threshold; in step f, the method for evaluating the morphological dispersion of the coupled vector bundle specifically involves calculating the convex hull area of ​​the coupled vector bundle in a two-dimensional coordinate system.

[0017] Preferably, the method further includes a personal baseline self-calibration step: continuously recording the convex hull area and calculating its statistical average over a very long period of 24 to 72 hours, using the statistical average as a personal benchmark for judging the stability of physiological coupling regulation.

[0018] Preferably, in step g, the generation of the warning information is triggered when the convex hull area of ​​the current time window is higher than the alarm threshold determined by the individual benchmark for at least one hour.

[0019] Preferably, in the two-dimensional coordinate system of the coupled vector bundle, the X-axis represents exercise intensity and the Y-axis represents the rate of change of heart rate. It is divided into the following functional quadrants: high response region, where the rate of change of heart rate is higher than 50 percent of its average value at rest; sluggish or disjointed region, where the absolute value of the rate of change of heart rate is less than 5 percent of its average value at rest.

[0020] Preferably, in step g, when it is determined that the stability of physiological coupling regulation is showing a declining trend, the method further includes: counting the number of coupling vectors falling into the high-response region. and the number of coupling vectors falling into sluggish or disconnected regions By comparison and The relative magnitudes are used to determine the pattern characteristics of physiological coupling regulatory instability, and the criteria for judging the pattern characteristics are: if If so, it is determined to be hyperactive instability; if If it is, it is determined to be an inhibited instability, in which The preset ratio threshold is greater than 1.

[0021] Preferably, the data acquisition is performed in a low-frequency, non-uniform manner, specifically by randomly waking up the device for five to ten seconds every minute to acquire data.

[0022] A health monitoring, analysis, and early warning system based on multimodal data fusion, comprising:

[0023] The data acquisition module is configured to simultaneously acquire a first physiological data stream of user motion information and a second physiological data stream of user heart rate information, and the acquisition is performed in a low-frequency non-uniform manner.

[0024] The frequency analysis module is configured to perform frequency domain analysis on the first physiological data stream within a specific short time window to determine whether the frequency distribution of its motion energy conforms to physiological low-frequency characteristics.

[0025] The coupling vector generation module is configured to determine an exercise intensity scalar based on the first physiological data stream and a heart rate change rate scalar based on the second physiological data stream only when the frequency distribution of exercise energy conforms to the physiological low-frequency characteristics. The exercise intensity scalar and the heart rate change rate scalar are then paired to form a coupling vector characterizing the physiological coupling response within the short time window.

[0026] The Coupled Vector Bundle Construction Module is configured to accumulate a series of coupled vectors over a specific long period of time to form a Coupled Vector Bundle, which reflects the intrinsic pattern of user physiological activity and heart rate response.

[0027] The stability assessment module is configured to evaluate the morphological dispersion of the coupled vector bundle and determine the stability of the user's physiological coupling regulation based on the trend of the morphological dispersion over a continuous time period.

[0028] The early warning generation module is configured to generate health risk early warning information when the stability of physiological coupling regulation shows a declining trend.

[0029] Compared with the prior art, the beneficial effects of the present invention are:

[0030] 1. By transforming micro-movements into natural probes, the system no longer relies on static values ​​of vital signs, but instead captures the coupling relationship between the scalar of exercise intensity and the scalar of heart rate change. When these coupled vectors form a vector bundle over a long period, the evolution of its morphological dispersion directly reflects the regulatory stability of the autonomic nervous system. This shift from observation results to assessment capabilities enables health monitoring to penetrate superficial data and capture early declines in bodily functions before routine physiological indicators become abnormal.

[0031] 2. Traditional alarms rely on threshold breaches, while this invention is based on dynamic assessment of the changing trend of convex hull area. It anchors the early warning logic to the instability process of an individual's own regulatory mode. When the vector bundle changes from a compact shape to a diffuse distribution, the system identifies a systematic decline in regulatory capacity. This mechanism, which uses loss of stability as the core trigger, avoids sensitivity to absolute value fluctuations, allowing the early warning window to be moved from the symptom outbreak period to the initial stage of dysfunction, creating a golden intervention opportunity for chronic disease management. The newly added frequency domain fingerprint identification mechanism forms a synergistic closed loop with this invention. By performing physiological low-frequency feature judgment on the acceleration signal, the system filters out non-physiological vibration interference before generating the coupling vector. This source filtering based on frequency domain energy distribution does not increase hardware costs and ensures the purity of the exercise-heart rate coupling relationship, so that the stability assessment still maintains the credibility of the conclusions in complex scenarios such as buses and near electrical appliances.

[0032] 3. When the convex hull area expands abnormally, the system further activates the quadrant density statistical mechanism. By analyzing the difference in distribution density of statistical vector bundles in the high-response and sluggish regions, the single unstable signal is decomposed into qualitative features of hyperactivity or inhibition. This mechanism reuses existing data structures and achieves initial screening of pathological tendencies simply by partition counting, providing actionable guidance for family health management. The core mechanism only requires a basic combination of accelerometer and PPG sensor. It replaces complex models with low-level calculations such as variance, slope, and convex hull area. Frequency domain analysis uses 8-point short-time FFT to filter interference, and partition statistics only require counter accumulation, achieving the ultimate utilization of universal hardware and realizing high-quality early warning with low energy consumption. Attached Figure Description

[0033] Figure 1 This is a timing diagram of the multimodal data fusion health monitoring, analysis, and early warning method of the present invention;

[0034] Figure 2This is a graph showing the relationship between exercise intensity and heart rate change rate according to the present invention.

[0035] Figure 3 This is a flowchart of the multimodal data fusion health monitoring and early warning system of the present invention.

[0036] The objectives, features, and advantages of this invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0037] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0038] This application provides a health monitoring, analysis, and early warning method based on multimodal data fusion, comprising the following steps:

[0039] Step a: Synchronously collect a first physiological data stream and a second physiological data stream through a wearable device. The first physiological data stream contains user motion information, and the second physiological data stream contains user heart rate information. The collection is performed in a low-frequency non-uniform manner.

[0040] Step b: Within a specific short time window, perform frequency domain analysis on the first physiological data stream to determine whether the frequency distribution of its motion energy conforms to physiological low-frequency characteristics.

[0041] Step c: Only when the frequency distribution of exercise energy conforms to the physiological low-frequency characteristics, determine an exercise intensity scalar based on the first physiological data stream, and determine a heart rate scalar based on the second physiological data stream;

[0042] Step d: Pair the exercise intensity scalar and the heart rate change rate scalar to form a coupling vector characterizing the physiological coupling response within the short time window;

[0043] Step e: Over a specific long period of time, a series of coupling vectors are accumulated to form a coupling vector bundle, which reflects the intrinsic pattern of the user's physiological activities and heart rate response.

[0044] Step f: Evaluate the morphological dispersion of the coupled vector bundle, and determine the stability of the user's physiological coupling regulation based on the changing trend of the morphological dispersion over a continuous time period.

[0045] Step g: When the stability of physiological coupling regulation shows a declining trend, health risk warning information is generated.

[0046] Preferably, in step a, the first physiological data stream is acquired by a triaxial accelerometer sensor, and the second physiological data stream is acquired by a photoplethysmography (PPG) sensor.

[0047] Preferably, in step c, the exercise intensity scalar is the variance of the acceleration signal within a specific short time window, and the heart rate change rate scalar is the slope of the heart rate sequence within a specific short time window.

[0048] Preferably, in step b, the frequency domain analysis specifically involves performing a short-time Fourier transform on the first physiological data stream and determining whether the energy proportion of the high-frequency part in the short-time Fourier transform result is lower than a specific judgment threshold; in step f, the method for evaluating the morphological dispersion of the coupled vector bundle specifically involves calculating the convex hull area of ​​the coupled vector bundle in a two-dimensional coordinate system.

[0049] Preferably, the method further includes a personal baseline self-calibration step: continuously recording the convex hull area and calculating its statistical average over a very long period of 24 to 72 hours, using the statistical average as a personal benchmark for judging the stability of physiological coupling regulation.

[0050] Preferably, in step g, the generation of the warning information is triggered when the convex hull area of ​​the current time window is higher than the alarm threshold determined by the individual benchmark for at least one hour.

[0051] Preferably, in the two-dimensional coordinate system of the coupled vector bundle, the X-axis represents exercise intensity, and the Y-axis represents the rate of change of heart rate. It is divided into the following functional quadrants: a high-response region, where the rate of change of heart rate is higher than 50% of its average value at rest; and a sluggish or disjointed region, where the absolute value of the rate of change of heart rate is less than 5% of its average value at rest. In this invention, the proposed convex hull area, as an indicator of the dispersion of the coupled vector bundle, is mainly based on the following physiological regulatory principle: in a healthy state, there is a relatively consistent and regular synergistic relationship between an individual's exercise behavior and heart rate response, exhibiting a relatively compact distribution in the coupled vector coordinate system; while when... When the autonomic nervous system's regulatory capacity declines, the body's heart rate response to the same type of exercise stimulus becomes unstable, delayed, or abnormal in amplitude, leading to an increase in the spatial divergence of the coupling vector. The convex hull area, as a spatial measure of the minimum envelope region, can directly reflect the overall distribution trend and degree of convergence and divergence of the coupling response. By continuously recording and comparing the trajectory of changes in this area, the evolution of individual regulatory stability can be dynamically perceived without relying on absolute numerical fluctuations. This approach has good physiological rationality and engineering operability, and also helps to complete early risk assessment without the need for complex model reasoning. All of these are extended implementation methods known to those skilled in the art.

[0052] Preferably, in step g, when it is determined that the stability of physiological coupling regulation is showing a declining trend, the method further includes: counting the number of coupling vectors falling into the high-response region. and the number of coupling vectors falling into sluggish or disconnected regions By comparison and The relative magnitudes are used to determine the pattern characteristics of physiological coupling regulatory instability, and the criteria for judging the pattern characteristics are: if If so, it is determined to be hyperactive instability; if If it is, it is determined to be an inhibited instability, in which The preset ratio threshold is greater than 1.

[0053] Preferably, the data acquisition is performed in a low-frequency, non-uniform manner, specifically by randomly waking up the device for five to ten seconds every minute to acquire data.

[0054] A multimodal data fusion health monitoring, analysis, and early warning system includes: a data acquisition module configured to simultaneously acquire a first physiological data stream of user exercise information and a second physiological data stream of user heart rate information, with acquisition performed in a low-frequency, non-uniform manner; a frequency analysis module configured to perform frequency domain analysis on the first physiological data stream within a specific short time window to determine whether the frequency distribution of its exercise energy conforms to physiological low-frequency characteristics; and a coupling vector generation module configured to determine an exercise intensity scalar based on the first physiological data stream and a heart rate variability scalar based on the second physiological data stream only when the frequency distribution of exercise energy conforms to physiological low-frequency characteristics, and to pair the exercise intensity scalar and the heart rate variability scalar to form a coupling vector characterizing the physiological coupling response within the short time window. The system includes: a coupled vector bundle construction module, configured to accumulate a series of coupled vectors over a specific long period to form a coupled vector bundle, which reflects the inherent pattern of user physiological activity and heart rate response; a stability assessment module, configured to assess the morphological dispersion of the coupled vector bundle and determine the stability of the user's physiological coupling regulation based on the trend of morphological dispersion over a continuous time period; an early warning generation module, configured to generate health risk warning information when the stability of physiological coupling regulation shows a declining trend; and a heart rate variability rate, which refers to the trend of heart rate change calculated from continuous pulse intervals within a specific time window. Its calculation is based on heart rate sampling results once per second, with a time granularity accurate to the second, serving as a reference for the trend and facilitating the capture of rapid regulatory responses. Furthermore, the paired analysis of exercise intensity and heart rate variability is based on a unified time segmentation to ensure the comparability and consistency of physiological coupling characteristics over time. A time-by-time update evaluation mechanism is employed in long-term monitoring, and statistical methods are used to extract the change patterns of the coupling vector bundle. For example, the discrete trends formed by long-term individual data can be dynamically compared with the current state to avoid interference from static indicators. In specific implementation, the system can set a reference range based on actual test samples. For instance, in certain scenarios, when the spatial dispersion of the coupling vector bundle exceeds a predetermined percentage range of the individual's historical mean over multiple consecutive time periods, it is considered a signal of decreased stability. The system thus triggers an early warning and identifies abnormal patterns in the coupling response, providing clear judgment criteria and actionable adjustment suggestions for subsequent health management. Moreover, the evaluation results generated by the system not only reflect the current state but also focus on the dynamic evolution trajectory of the regulatory process. Especially when resting physiological data remain normal, potential abnormal regulatory trends can be detected early through subtle fluctuations in the consistency of the response between exercise and heart rate. These are all extended implementation methods known to those skilled in the art.

[0055] Example 1: This example proposes a multimodal data fusion health monitoring, analysis, and early warning method and system. It aims to assess early changes in the body's regulatory capabilities using low-power wearable devices without relying on traditional absolute values ​​of vital signs, and generate health risk warning information accordingly. First, the wearable device simultaneously collects the user's motion data stream and heart rate data stream. Specifically, the first physiological data stream is collected by a triaxial accelerometer sensor to reflect the user's micro-motion information; the second physiological data stream is acquired by a photoplethysmography (PPG) sensor to represent the user's dynamic heart rate. This data acquisition process uses low-frequency non-linear pulse wave (NLP) technology. The uniform wake-up strategy involves the device randomly activating the data acquisition module every minute and running it for 5 to 10 seconds. This strategy effectively reduces energy consumption while meeting the time resolution and data coverage requirements for health monitoring. The acquired motion data is sent to the frequency domain analysis process within a specific short time window. This invention uses short-time Fourier transform to process the first physiological data stream and determines whether the frequency distribution of its motion energy has physiological low-frequency characteristics. Physiological low-frequency characteristics refer to the proportion of high-frequency energy in the spectrum being lower than a set judgment threshold. Only when this condition is met will subsequent processing steps continue.

[0056] After the frequency domain filtering described above, this invention calculates two scalar indicators representing physiological responses based on acceleration and heart rate data: one is a scalar of exercise intensity, specifically the variance of the acceleration signal within the short time window; the other is a scalar of heart rate change rate, which is the slope of the heart rate sequence within the same time window. These two scalars, after pairing, constitute a coupling vector characterizing the physiological coupling response state within the short time window. Over a set long period, the system continuously accumulates the aforementioned coupling vectors, forming a coupling vector bundle. This coupling vector bundle, as a collection of motion and heart rate coupling characteristics across multiple time segments, reflects the inherent regularity between the user's physiological activities and autonomic nervous system responses, forming the basis for subsequent stability assessments. To assess the user's current physiological regulation stability, this invention calculates the coupling vector bundle... The convex hull area formed by the combined vector bundle in a two-dimensional coordinate system is used to characterize its morphological dispersion. In this coordinate system, the X-axis represents exercise intensity and the Y-axis represents the rate of change of heart rate. As physiological coupling stability decreases, the spatial distribution of the coupling vectors tends to diverge, and the corresponding convex hull area also shows an expanding trend. The system identifies the systematic decline of the user's physiological regulatory ability by tracking the evolution trajectory of the convex hull area over a continuous time period. When the convex hull area is detected to be continuously expanding over a certain time period, and the trend clearly points to a decline in regulatory stability, the present invention will trigger a health risk warning mechanism. The generation of warning information does not depend on the anomaly of a single measurement value, but is based on the loss of dynamic stability reflected by the accumulation of data over multiple time periods, thereby realizing the transformation of the warning mechanism from result-driven to capability assessment-driven.

[0057] To further improve system adaptability and reduce the interference of inter-individual differences on judgment results, this invention introduces an individual baseline self-calibration mechanism. The system continuously records the user's convex hull area and calculates the statistical average of this indicator over an ultra-long period of 24 to 72 hours, using it as an individual reference benchmark for the user's physiological coupling regulation stability. In actual monitoring, only when the convex hull area of ​​the current time period is higher than the alarm threshold determined by this benchmark for at least one hour is it determined that there is a risk of physiological regulation decline, and a warning message is generated accordingly. To enhance the interpretability and intervention guidance significance of the warning system, this invention further divides the functional image in the two-dimensional coordinate space of the coupling vector. The system defines a high-response region as one where the heart rate variability is 50% higher than the average at rest; and a sluggish or disjointed region as one where the absolute value of the heart rate variability is 5% lower than the average at rest. While the coupled vector bundle distribution shows significant morphological diffusion, the system also counts the number of vectors in each quadrant to analyze the instability pattern. Specifically, if the number of coupled vectors falling into the high-response region is significantly greater than that in the sluggish region, it is determined to be hyperactive instability; otherwise, it is determined to be inhibitory instability. The determination condition for this pattern is: if the former is greater than the latter multiplied by a preset proportional threshold greater than 1, it is confirmed as hyperactive; otherwise, it is confirmed as inhibitory.

[0058] Example 2: This example proposes a multimodal data fusion health monitoring, analysis, and early warning method and system. It aims to utilize low-power wearable devices to assess early decline in the body's regulatory capacity and generate health risk warnings without relying on traditional absolute values ​​of vital signs. The specific implementation steps are as follows: First, two physiological data streams are simultaneously collected using a wearable device: the first physiological data stream is collected by a triaxial accelerometer sensor to reflect the user's micro-movements; the second physiological data stream is acquired by a photoplethysmography (PPG) sensor to reflect changes in the user's heart rate. Data acquisition employs a low-frequency non-uniform wake-up strategy, specifically randomly waking the device for 5 to 10 seconds per minute for data acquisition. This strategy effectively reduces energy consumption while ensuring sufficient temporal resolution and data coverage. After data acquisition, frequency domain analysis is first performed on the motion data stream. Specifically, the first physiological data stream is processed using a Short-Time Fourier Transform (STFT) to determine whether the frequency distribution of exercise energy conforms to physiological low-frequency characteristics. The criterion for physiological low-frequency characteristics is that the proportion of high-frequency energy in the spectrum is lower than a set threshold. Only when this condition is met will the system continue to execute subsequent processing steps. Then, two scalars are calculated based on acceleration data and heart rate data respectively: the exercise intensity scalar quantifies the exercise intensity by calculating the variance of the acceleration signal within a short time window; the heart rate change rate scalar quantifies the heart rate change rate by calculating the slope of the heart rate sequence within a short time window. These two scalars represent the physiological response characteristics of exercise and heart rate, respectively. They are then paired to generate a coupling vector, where the X-axis represents exercise intensity and the Y-axis represents the heart rate change rate, characterizing the physiological coupling response state within a short time window.

[0059] Over a long time period, the system accumulates multiple coupling vectors, forming a coupling vector bundle. This bundle reflects the user's exercise-heart rate coupling pattern over a longer period, revealing the intrinsic relationship between the user's physiological activities and the autonomic nervous system response. To assess the user's physiological regulation stability, this embodiment measures the dispersion of the coupling vectors by calculating the area of ​​the convex hull formed by the coupling vector bundle in a two-dimensional coordinate system. As physiological regulation stability decreases, the distribution of coupling vectors tends to diverge, and the convex hull area increases. By tracking the evolution of the convex hull area over continuous time periods, the system can identify the decline in the user's physiological regulation ability. When the system detects that the convex hull area continuously increases over a certain period and the trend points to a decline in physiological regulation stability, the system triggers a health risk warning mechanism. This mechanism differs from traditional threshold-based alarm systems; instead, it generates warning information based on the dynamic stability changes of multi-time period data. To further improve the system's adaptability and reduce the impact of individual differences, this embodiment introduces an individual baseline self-calibration mechanism. The system continuously records the user's convex hull. The system calculates the area of ​​the convex hull and its statistical average over a long period of 24 to 72 hours. This statistical average serves as an individual benchmark for the stability of the user's physiological coupling regulation. In actual monitoring, the system only determines the risk of declining physiological regulation capacity and generates an early warning when the area of ​​the convex hull in the current time period continuously exceeds the alarm threshold determined by this benchmark. In addition, the system divides the coupling vector bundle into multiple functional quadrants in a two-dimensional coordinate system: the high-response region indicates that the heart rate change rate is higher than 50% of the average value under resting conditions; the sluggish or disjointed region indicates that the absolute value of the heart rate change rate is lower than 5% of the average value under resting conditions. When the distribution of the coupling vector bundle changes significantly, the system counts the number of coupling vectors in each quadrant and analyzes the instability mode by comparing the ratio of the high-response region to the sluggish region: if the number of coupling vectors in the high-response region is significantly greater than that in the sluggish region, it is judged as hyperactive instability; otherwise, it is judged as inhibitory instability. This analysis mechanism provides more interpretive information for health risk assessment and provides effective intervention guidance for health managers.

[0060] Example 3: This example aims to verify the practical efficacy of a multimodal data fusion-based health monitoring, analysis, and early warning method and system in assessing the stability of autonomic nervous system regulation and detecting early functional decline through a continuous monitoring experiment of physiological responses during daily activities. The core of this experiment lies in achieving objective quantification and trend-based early warning of changes in individual physiological regulatory stability by analyzing the dynamic evolution of the exercise-heart rate coupling relationship, without relying on traditional absolute values ​​of vital signs. In the context of increasingly widespread home-based health management, how can low-power, non-invasive wearable devices be used to monitor and analyze physiological responses in a way that is imperceptible or barely perceptible to the user? This study aims to achieve continuous and in-depth assessment of physical health. Existing technologies typically focus on threshold monitoring of absolute values ​​of physiological parameters. Such methods often lag behind subtle changes in the body's internal regulatory mechanisms, resulting in a lack of early warning windows. To overcome this limitation, this experiment seeks to validate a novel health monitoring, analysis, and early warning method. This method transforms daily micro-movements into physiological probes and analyzes their coupling patterns with heart rate dynamics to achieve real-time assessment of the dynamic stability of the autonomic nervous system's regulatory capacity. This could potentially capture early signals of functional decline before significant abnormalities appear in routine physiological indicators, providing users with forward-looking health risk warnings.

[0061] This experiment used a wearable device integrating a triaxial accelerometer sensor and a photoplethysmography (PPG) sensor for data acquisition. This device features low power consumption and communicates with the local data processing unit via Bluetooth Low Energy. To ensure the authenticity and general applicability of the data, the experimental environment was set up in a relatively free daily home setting for the subjects, including but not limited to light housework, walking, and sitting. During the experiment, environmental factors that could introduce strenuous exercise or significant non-physiological vibrations were excluded to ensure the effective representation of the physiological low-frequency characteristics of the motion signal. All data acquisition and processing were completed on the local processing unit, whose configuration simulates a real-world edge computing environment. The subjects in this experiment were digital wireless... Healthy volunteers with known cardiovascular or neurological diseases wore wearable devices for continuous monitoring and data recording. Data acquisition employed a low-frequency non-uniform wake-up strategy, where the device randomly activated sensors every minute and continuously collected data for several seconds. This strategy aimed to balance data coverage and system power consumption. Preliminary verification showed that it effectively captured the dynamic coupling relationship between motion and heart rate, while also extending the device's battery life on a single charge, meeting practical home monitoring needs. Before entering the subsequent analysis process, the raw acceleration and PPG signals were preprocessed. The acceleration signal was processed by bandpass filtering to remove DC components and high-frequency noise; the heart rate signal was converted into a heart rate sequence through peak detection and interval analysis.

[0062] Within each specific short-time window, a short-time Fourier transform is performed on the preprocessed acceleration signal. The key to this stage is determining whether the frequency distribution of motion energy conforms to physiological low-frequency characteristics. This method uses the proportion of high-frequency energy in the short-time Fourier transform results for screening: if the proportion of energy above a specific frequency in the spectrum is below a preset threshold, the motion signal within the current window is considered to primarily originate from natural, low-frequency physiological micro-movements of the human body, rather than external interference. Only then can the data from this window proceed to subsequent coupling analysis. The determination of this threshold is based on a technical consensus formed through the analysis of a large amount of daily activity data in this field, aiming to retain effective physiological information to the greatest extent while avoiding non-physiological interference, thus ensuring evaluation under complex environments. To assess the reliability of the estimation, for short time windows filtered by the frequency domain, the scalars of exercise intensity and heart rate change are calculated separately. The scalar of exercise intensity is defined as the variance of the acceleration signal within the short time window. Variance can effectively quantify the activity level of micro-movements. Its calculation method is simple and easy to implement on low-power hardware. The scalar of heart rate change is defined as the slope of the heart rate sequence within the same time window. The heart rate slope can reflect the instantaneous upward or downward trend of heart rate and its rate, which is a direct manifestation of the rapid regulatory response of the autonomic nervous system. The determination of these two scalars is based on a deep understanding of the physiological signal characteristics in this field and the effectiveness considerations in engineering practice. The above-mentioned scalars of exercise intensity and heart rate change are paired to form a characterization of physiological coupling within the short time window. The system uses a coupling vector as the response vector, where the X-axis represents exercise intensity and the Y-axis represents the rate of change in heart rate. Over a continuous long period, the system accumulates a series of such coupling vectors, forming a coupling vector bundle. To quantify the morphological dispersion of the coupling vector bundle, this method calculates the area of ​​the convex hull formed by the coupling vector bundle in a two-dimensional coordinate system. A larger convex hull area indicates a more dispersed distribution of coupling vectors, reflecting a decrease in the stability of the user's physiological coupling regulation. As a geometric algorithm, the convex hull algorithm is computationally efficient and can accurately capture the outer boundary of the point set, thus effectively quantifying its dispersion. To accommodate individual differences, this method introduces a personal baseline self-calibration mechanism. The system continuously records the user's convex hull area and updates it over 24 to 72 hours. The statistical average of the indicator is calculated over a very long period of time, and this statistical average is used as a personal benchmark for judging the stability of physiological coupling regulation. In actual monitoring, only when the convex hull area of ​​the current time period is higher than the alarm threshold determined by the personal benchmark for at least one hour is it determined that there is a risk of decline in physiological regulation capacity, and a warning message is generated accordingly. This dynamic threshold setting fully considers daily physiological fluctuations and avoids the limitations of rigid fixed thresholds, which prevent the warning logic from being misplaced in the process of instability of the individual's own regulation mode. In order to provide more instructive warning information, this method divides the following functional quadrants in the two-dimensional coordinate system of the coupled vector bundle: high response region, whose central rate of change is higher than 50% of its average value in the resting state;In sluggish or disconnected regions, the absolute value of the rate of change of the central velocity is less than five percent of its average value in the resting state. When it is determined that the physiological coupling regulation stability is showing a declining trend, the number of coupling vectors falling into the high response region is further included. and the number of coupling vectors falling into sluggish or disconnected regions By comparison and The relative magnitudes are used to determine the pattern characteristics of physiological coupling regulatory instability, and the criteria for judging the pattern characteristics are: if If so, it is determined to be hyperactive instability; if If it is, it is determined to be an inhibited instability, in which With a preset proportion threshold greater than 1, this mechanism reuses existing data structures and achieves initial screening of pathological tendencies simply by counting by partitions, providing an actionable guidance dimension for family health management.

[0063] During daily activities, the convex hull area of ​​the coupled vector bundles of subjects exhibited regular dynamic changes. During periods of emotional stability and good physiological condition, the convex hull area was typically small with minimal fluctuations, and the coupled vector bundle was compact, indicating a stable and efficient coupling relationship between exercise and heart rate. When subjects experienced mild physiological load or fatigue, a reversible increase in the convex hull area was observed, but it usually recovered to baseline levels within a short time, reflecting the normal regulatory and adaptive capacity of the physiological system. Continuous monitoring revealed that when some subjects exhibited signs of sub-health, such as mild discomfort but not meeting the diagnostic criteria for disease, even though their absolute heart rate and acceleration values ​​remained within the normal range, the convex hull area data recorded by the system over continuous time periods showed that their convex hull area remained consistently above their individual baseline alarm threshold. This verifies that this method can capture early declines in the body's regulatory stability by increasing the convex hull area before traditional indicators become abnormal. During the experiment, when subjects were subjected to vibrations that could introduce non-physiological stresses... In environments such as when operating certain household appliances, frequency domain analysis revealed that the vibration energy is mainly concentrated in the high-frequency band. Thanks to the screening mechanism of physiological low-frequency characteristics, when the proportion of high-frequency energy in the acceleration signal exceeds a preset threshold, the data in this short time window is effectively eliminated and does not participate in the subsequent coupling vector generation. This proves that the frequency domain screening mechanism can effectively filter non-physiological interference, ensuring the purity and reliability of stability assessment results in complex environments. When the physiological coupling regulation stability of the subjects is judged to be in a declining trend, we further analyzed the quadrant distribution of their coupling vector bundle. The results showed that during the period of continuous warning of their convex hull area, there was a significant difference between the number of coupling vectors falling into the high-response region and the number of coupling vectors falling into the sluggish or disjointed region, which meets the criteria for hyperactive or inhibitory instability. For example, if the number of coupling vectors in the high-response region is significantly more than that in the sluggish region, it is judged as hyperactive instability. The identification of this specific pattern provides a more precise guidance for subsequent health management interventions that may be needed.

[0064] Example 4: This example combines Figures 1 to 3 This paper describes a health monitoring, analysis, and early warning method and system based on multimodal data fusion. For example... Figure 1As shown in the diagram, the data acquisition layer includes two types of sensor devices: a triaxial accelerometer and a PPG sensor. These sensors simultaneously acquire motion data streams and heart rate data streams. The data acquisition process uses a low-frequency, non-uniform method at a rate of 5 to 10 seconds per minute. The subsequent signal processing layer includes a frequency domain analysis module and a coupling generator. The frequency domain analysis module performs frequency domain analysis on the motion data stream, filtering out signals that conform to physiological low-frequency characteristics. Only when the signals meet the conditions does the coupling generator generate scalars of exercise intensity and heart rate change rate based on the motion data stream and heart rate data stream, respectively. These scalars are paired to form coupling vectors of physiological coupling response. Finally, in the intelligent analysis layer, these coupling vectors are fed into a vector bundle builder and a stability evaluator. The vector bundle builder accumulates multiple coupling vectors over a long period to form a coupled vector bundle, while the stability evaluator analyzes the morphological dispersion of the coupled vector bundle and judges the stability of physiological coupling regulation. When the stability shows a declining trend, the system generates a health risk warning.

[0065] like Figure 2 As shown in the figure, the horizontal axis represents exercise intensity (acceleration variance), and the vertical axis represents the rate of change of heart rate (bpm / s). Different points represent measurement results at different time periods. The figure uses three different symbols to distinguish the monitoring time periods: circles represent afternoon measurement results, squares represent morning measurement results, and triangles represent evening measurement results. It can be seen that as exercise intensity increases, the rate of change of heart rate gradually increases, and the measurement data at different time periods show a similar relationship in the overall trend.

[0066] like Figure 3 As shown, the data acquisition module primarily includes spoofed device acquisition and synchronous acquisition of exercise and heart rate data. Spoofed device acquisition supports low-frequency mode synchronization and dual-modal synchronization. In the post-acquisition data management module, medical staff can manage and adjust data through historical data queries, baseline parameter settings, and early warning threshold adjustments. The subsequent intelligent analysis module includes physiological feature extraction and coupling relationship analysis. Physiological feature extraction can be achieved through real-time health assessment, while coupling relationship analysis is used to analyze the user's exercise and heart rate changes, generating personalized health risk warnings in conjunction with real-time health assessments. Simultaneously, under the early warning management module, through risk warning generation and error pattern recognition, the system can generate corresponding early warning information based on predictive analysis results and push suggestions to help users address health risks promptly. Furthermore, the system has extended functions, including extended stability assessment and individual benchmark calibration, further improving the accuracy and adaptability of health monitoring.

[0067] Example 5: This example provides a health monitoring, analysis and early warning method and system based on multimodal data fusion. Targeting the physiological state monitoring needs of users in non-wearable or low-sensitivity environments, the system is equipped with an integrated sensing module that integrates a three-axis accelerometer and a photoplethysmography (PPG) pulse wave detection unit, and periodically exchanges data with the main control unit via Bluetooth Low Energy communication protocol. During normal operation, the system pseudo-randomly wakes up the acquisition module once per minute. Each sampling period is set to 6 to 10 seconds, preferably 8 seconds. The sampled data undergoes short-time Fourier transform processing using a symmetrical Hanning window function. The spectral resolution is set to 0.5 Hz, and the window overlap rate is set to 50% to control computational resource consumption while ensuring frequency separation accuracy. The acquired acceleration modulus data first enters the frequency domain filtering stage. The system uses short-time Fourier transform to obtain the spectral power density distribution, and then calculates the energy proportion above 2 Hz. When this proportion is below 30%, the motion signal within that sampling period is determined to mainly originate from the user's natural physiological activities. The 2 Hz threshold is based on the fact that the frequency of daily human activities is mainly concentrated between 0.5 Hz and 1 Hz, while signals above 2 Hz are usually related to involuntary disturbances, such as environmental vibrations or mechanical vibrations, and therefore can be used as noise filters. In addition to the above, data filtered through the frequency domain is used to construct coupling vectors. The system calculates the variance of the acceleration modulus in the current sampling window as an indicator of exercise intensity. Simultaneously, the pulse wave peak interval sequence is extracted, and its linear slope is calculated as an indicator of heart rate variability. The above two indicators form a two-dimensional coupling vector, where exercise intensity characterizes local exercise activity, and heart rate variability reflects the immediate response trend of the autonomic nervous system to exercise stimulation. The system saves the coupling vectors constructed by continuous sampling in a sliding window for long-term trend analysis. This sliding window is set to 15 minutes, contains no less than 20 coupling vector points, and is refreshed with new samples. In this window, all coupling vectors are mapped to a two-dimensional coordinate system with exercise intensity as the horizontal axis and heart rate variability as the vertical axis. The convex hull calculation module is called to extract the envelope region and calculate its area. The obtained area is used to measure the dispersion of the physiological response pattern during this period.

[0068] To account for individual differences, the system employs an automated resting-state baseline calibration mechanism. Specifically, during the first 72 hours of continuous operation after device deployment, the system statistically analyzes the hourly convex hull area and uses the median value as the user's individual dispersion benchmark. In subsequent operations, if the convex hull area exceeds 150% of this benchmark value within three consecutive sliding windows, the system identifies a declining trend in regulatory capacity and triggers an alert. For example, in a typical application scenario, if a user is engaged in low-intensity activity (such as slow walking or light housework) for three consecutive hours, and their heart rate remains stable near resting levels during this period, the relevant coupling vectors will be concentrated in the low-response quadrant. The system further determines that the number of vectors falling into this quadrant is 13, indicating a high-response quadrant. The quadrant vectors are 3. If the current risk judgment threshold is 3, then since 13 is greater than 3 multiplied by 3, it satisfies the suppression instability characteristics. Based on this, the system merges and judges that the regulation ability declines and the response mode is abnormal, thereby generating a high-level risk warning. It should be noted that the frequency domain screening rules adopted can effectively remove interference signals such as TV sound, elevator operation and floor vibration in common living environments, improving the effectiveness of physiological data. At the same time, the calculation process involved in this embodiment only relies on conventional algorithms such as basic Fourier transform, variance calculation and convex hull construction, which has good software and hardware deployment adaptability. It can also run stably under the condition of limited embedded main control chip resources, and can achieve long-term continuous operation in button battery power supply mode.

[0069] Example 6: A scenario for health monitoring of a 70-year-old user living alone. The system's multimodal data fusion and early warning mechanism is specifically deployed and executed. The user's wearable wristband device integrates a triaxial accelerometer and a photoplethysmography (PPG) sensor, operating in a low-frequency, non-uniform manner. Every minute, the device's main control unit randomly selects a time point to wake up the sensor and simultaneously collect eight seconds of raw acceleration and pulse wave data. It then enters a sleep state. The acceleration data collected during this process first undergoes a frequency domain fingerprinting process. The system performs a short-time Fourier transform on the eight-second acceleration modulus signal, calculating its values ​​above a specific frequency. The ratio of the total energy in the (2 Hz) frequency band to the total signal energy, this frequency threshold. The setting is based on a general rule derived from statistical analysis of a large amount of physiological activity data: the energy of human activities, such as walking, getting up, and housework, is mainly concentrated in the low-frequency range below 2 Hz, while vibrations above this frequency are mostly caused by environmental noise or equipment vibration. Only when the proportion of high-frequency energy is below a preset judgment threshold will the system be considered effective. Only when the percentage is 30% is the data segment confirmed as a valid physiological signal and allowed to proceed to the subsequent analysis stage, thus ensuring the purity of the data at the source of the algorithm.

[0070] For the data window that has passed frequency domain discrimination, the system then calculates two core scalars. One is the motion intensity scalar, which calculates the variance of the synthesized triaxial acceleration signal within the window. The results show that this indicator can sensitively quantify the user's activity level in minute movements. The second indicator is the heart rate scalar. The system first extracts the heart rate value second by second from the synchronously acquired pulse wave data, and then calculates the slope of the linear fit of the heart rate sequence within an eight-second window. The slope directly reflects the dynamic change trend of heart rate over a short period of time. These two scalars are then paired to form a two-dimensional coupled vector. This vector precisely characterizes the heart rate regulation response of the user's autonomic nervous system under specific micro-motion perturbations. The system continuously accumulates these coupled vectors within a 15-minute sliding time window, forming a coupled vector bundle containing dozens of data points. To quantify the morphological dispersion of this vector bundle, the system calls the geometry calculation module to construct and calculate the minimum convex hull area of ​​these data points in a two-dimensional coordinate system. This area The size of the area directly corresponds to the consistency and stability of the user's exercise and heart rate coupling response pattern within the fifteen minutes: the smaller the area, the more stable and regular the response pattern; the larger the area, the more divergent and disordered the response pattern.

[0071] To establish personalized health assessment benchmarks, the system performs a personal baseline self-calibration step within 72 hours of initial deployment. During this period, the system records all calculated 15-minute convex hull areas. The values ​​were calculated, and these values ​​were statistically sorted. The 75th percentile was taken as the personalized stable benchmark for the user. This baseline represents a relatively relaxed but healthy upper limit to a user's ability to regulate their daily physiological fluctuations. Based on this individual baseline, the system sets dynamic alarm thresholds. The relationship is as follows: After completing baseline calibration and entering the long-term monitoring phase, the system continuously calculates the convex hull area of ​​the current 15-minute window. The warning is not triggered based on a single instance of area exceeding the limit, but rather on a well-defined trend of stability degradation. Based on this, the system calculates the moving average of the convex hull area over the past hour, i.e., four consecutive 15-minute windows. Only when the moving average Continuously exceeding the alarm threshold Only after a preset duration of one hour will the system determine that the user's physiological coupling regulation stability has experienced a clinically significant and persistent decline, and will then officially generate an early warning.

[0072] Once a declining trend in physiological coupling regulation is confirmed, the system will immediately initiate a qualitative analysis procedure for the instability mode. The system will project all coupling vectors accumulated in the two hours prior to the warning onto a clearly defined two-dimensional coordinate system, divided according to physiological significance: heart rate variability. The region exceeding 50% of the average resting heart rate variability is defined as the high-response region; and Regions with absolute values ​​below 5 percent of the average resting heart rate variability are defined as sluggish or disjointed regions. The system counts the number of vectors falling into high-response regions. The number of vectors falling into sluggish or disjointed regions The instability mode is identified by comparing the relative magnitudes of the two factors. The judgment rule is: if If so, it is judged as hyperactive instability; conversely, if If the value is 1, it is determined to be an inhibited instability. The proportional threshold is... Set to 3.0, this value is based on statistical data from a large clinical population. It is believed that when the number of one response pattern reaches more than three times that of another, this imbalance constitutes a significant pathological tendency that needs attention. Ultimately, a detailed health report containing warnings of physiological coupling regulation stability decline, qualitative conclusions of unstable patterns, and relevant data trend graphs will be automatically generated and pushed to the user and their designated guardian, thus realizing a closed loop of the entire process from data collection and intelligent analysis to accurate early warning and in-depth interpretation.

[0073] Furthermore, in specific implementation, when the system determines that the user's physiological coupling regulation stability shows a declining trend, i.e., the moving average of the convex hull area over a continuous one-hour period... Continuously exceeding the individual alarm threshold Instead of simply outputting a general instability alert, the system automatically activates a deep qualitative analysis process to reveal the specific pathological patterns behind the instability decline. The core of this process is to finely interpret the distribution patterns of all coupling vectors accumulated within a specific time period (such as two hours) before the alert is triggered in its two-dimensional coordinate space.

[0074] In this two-dimensional coordinate system, the high-response region is defined as the rate of change of heart rate. The region exceeding 50% of the user's average resting heart rate variability represents an overactivation of the sympathetic nervous system in the autonomic nervous system. This indicates an excessively modulated response of the heart to minute movements, a direct projection of the body being in a state of regulation or hyperactivity. Conversely, a sluggish or disconnected region is defined as the heart rate variability... The absolute value of the vector is less than 5 percent of the average heart rate change rate at rest. Vectors in this region reveal no response or significant lag in heart rate regulation to motion disturbances. This usually points to abnormal vagal tone or obstruction of the cardiopulmonary regulatory pathway, and is a sign of fatigue or inhibition in the body.

[0075] To illustrate the value of this qualitative analysis, we used the aforementioned septuagenarian user as an example and observed him over two different time periods. In the first month of monitoring, the system detected his convex hull area. The number of vectors began to increase continuously and triggered an alert. In the subsequent qualitative analysis, the system counted the number of vectors falling into the sluggish or disconnected regions. The number is 18, while the number of vectors falling into the high response region is... There are only 2, according to the preset judgment rules, because The system accurately classified this instability event as inhibitory instability. The significance of this classification goes far beyond a simple alarm: it strongly suggests that the user's physiological state is trending towards a decline in regulatory function and a slow reaction. This may be a signal of age-related functional decline, chronic fatigue accumulation, or even early cardiac autonomic neuropathy. The health report generated by the system will clearly point out this inhibitory characteristic and recommend that the caregiver pay attention to the user's energy level and consider conducting relevant examinations targeting cardiac function or neural regulatory pathways.

[0076] In the subsequent second month of monitoring, the user's convex hull area... The threshold was repeatedly exceeded, but this time the qualitative analysis presented a completely different picture: the system counted the number of vectors falling into the high-response region. There are 21 vectors, while the number of vectors falling into sluggish or disconnected regions is 21. There are 4, because The system classified this instability as hyperactive instability, a conclusion that points to a completely different physiological state from the previous one. That is, the body may be in an unstable state caused by high regulation, high anxiety, or poor sleep quality, making it overreact to minor disturbances. Based on this classification, the health advice pushed by the system will focus on stress management, improving the sleep environment, or reviewing whether there has been any recent use of drugs that affect the nervous system.

[0077] Through the above methods, this invention successfully analyzes a single, quantitative indicator of decreased stability (increased convex hull area) into two pattern characteristics (inhibitory and hyperactive) with clearly different pathological orientations and intervention recommendations. This deepening from problem discovery to problem characterization makes the early warning information no longer a vague signal, but a health insight and early warning with preliminary diagnostic value that can be understood and acted upon.

[0078] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0079] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A health monitoring, analysis, and early warning method based on multimodal data fusion, characterized in that, The method includes the following steps: Step a: Synchronously collect a first physiological data stream and a second physiological data stream through a wearable device. The first physiological data stream contains user motion information, and the second physiological data stream contains user heart rate information. The collection is performed in a low-frequency non-uniform manner. Step b: Within a specific short time window, perform frequency domain analysis on the first physiological data stream to determine whether the frequency distribution of its motion energy conforms to physiological low-frequency characteristics. Step c: Only when the frequency distribution of exercise energy conforms to the physiological low-frequency characteristics, determine an exercise intensity scalar based on the first physiological data stream, and determine a heart rate scalar based on the second physiological data stream; Step d: Pair the exercise intensity scalar and the heart rate change rate scalar to form a coupling vector characterizing the physiological coupling response within the short time window; Step e: Over a specific long period of time, a series of coupling vectors are accumulated to form a coupling vector bundle, which reflects the intrinsic pattern of the user's physiological activities and heart rate response. Step f: Evaluate the morphological dispersion of the coupled vector bundle, and determine the stability of the user's physiological coupling regulation based on the changing trend of the morphological dispersion over a continuous time period. Step g: When the stability of physiological coupling regulation shows a declining trend, health risk warning information is generated; In step c, the exercise intensity scalar is the variance of the acceleration signal within a specific short time window, and the heart rate change rate scalar is the slope of the heart rate sequence within a specific short time window. In step b, the frequency domain analysis specifically involves performing a short-time Fourier transform on the first physiological data stream and determining whether the energy proportion of the high-frequency part in the short-time Fourier transform result is lower than a specific judgment threshold; in step f, the method for evaluating the morphological dispersion of the coupled vector bundle specifically involves calculating the convex hull area of ​​the coupled vector bundle in a two-dimensional coordinate system.

2. The health monitoring, analysis, and early warning method based on multimodal data fusion according to claim 1, characterized in that, In step a, the first physiological data stream is acquired by a triaxial accelerometer sensor, and the second physiological data stream is acquired by a photoplethysmography (PPG) sensor.

3. The health monitoring, analysis, and early warning method based on multimodal data fusion according to claim 1, characterized in that, The method also includes a personal baseline self-calibration step: continuously recording the convex hull area and calculating its statistical average over an ultra-long period of 24 to 72 hours, using this statistical average as a personal benchmark for judging the stability of physiological coupling regulation.

4. The health monitoring, analysis, and early warning method based on multimodal data fusion according to claim 2, characterized in that, In step g, the early warning information is generated when the convex hull area of ​​the current time window is higher than the alarm threshold determined by the individual benchmark for at least one hour.

5. The health monitoring, analysis, and early warning method based on multimodal data fusion according to claim 1, characterized in that, In the two-dimensional coordinate system of the coupled vector bundle, the X-axis represents exercise intensity and the Y-axis represents the rate of change of heart rate. It is divided into the following functional quadrants: the high-response region, where the rate of change of heart rate is higher than 50 percent of its average value at rest; and the sluggish or disjointed region, where the absolute value of the rate of change of heart rate is less than 5 percent of its average value at rest.

6. The health monitoring, analysis, and early warning method based on multimodal data fusion according to claim 5, characterized in that, In step g, when it is determined that the stability of physiological coupling regulation is showing a declining trend, it further includes: counting the number of coupling vectors falling into the high response region. and the number of coupling vectors falling into sluggish or disconnected regions By comparison and The relative magnitudes are used to determine the pattern characteristics of physiological coupling regulatory instability, and the criteria for judging the pattern characteristics are: if If so, it is determined to be hyperactive instability; if If it is, it is determined to be an inhibited instability, in which The preset ratio threshold is greater than 1.

7. The health monitoring, analysis, and early warning method based on multimodal data fusion according to claim 1, characterized in that, Data collection is performed in a low-frequency, non-uniform manner, specifically by randomly waking up the device for five to ten seconds every minute to collect data.

8. A health monitoring, analysis, and early warning system based on multimodal data fusion, executing the health monitoring, analysis, and early warning method based on multimodal data fusion as described in claims 1-7, wherein the system comprises: The data acquisition module is configured to simultaneously acquire a first physiological data stream of user motion information and a second physiological data stream of user heart rate information, and the acquisition is performed in a low-frequency non-uniform manner. The frequency analysis module is configured to perform frequency domain analysis on the first physiological data stream within a specific short time window to determine whether the frequency distribution of its motion energy conforms to physiological low-frequency characteristics. The coupling vector generation module is configured to determine an exercise intensity scalar based on the first physiological data stream and a heart rate change rate scalar based on the second physiological data stream only when the frequency distribution of exercise energy conforms to the physiological low-frequency characteristics. The exercise intensity scalar and the heart rate change rate scalar are then paired to form a coupling vector characterizing the physiological coupling response within the short time window. The Coupled Vector Bundle Construction Module is configured to accumulate a series of coupled vectors over a specific long period of time to form a Coupled Vector Bundle, which reflects the intrinsic pattern of user physiological activity and heart rate response. The stability assessment module is configured to evaluate the morphological dispersion of the coupled vector bundle and determine the stability of the user's physiological coupling regulation based on the trend of the morphological dispersion over a continuous time period. The early warning generation module is configured to generate health risk early warning information when the stability of physiological coupling regulation shows a declining trend.

Images