Real-time health status monitoring system and method based on biosensor

Abstract

The present application relates to the technical field of health state monitoring, in particular to a real-time health state monitoring system and method based on a biological sensor. The present application combines posture, heart rate, blood oxygen and other physiological data collected by the biological sensor with body posture and environmental video collected by the image collector through multi-modal data fusion, solves the limitations of single data source monitoring, avoids false positives caused by normal actions such as bending over in traditional fall early warning, or the problem that high altitude reaction early warning only prompts hypoxia but cannot associate with fall risk. The pre-trained risk assessment model outputs a graded risk level, triggers targeted early warning such as low-risk prompt deceleration and high-risk sound and light alarm, provides more accurate and timely health monitoring services for the elderly in plateau areas, and effectively reduces the fall risk caused by the interaction between environment and physiological state.

Description

Technical Field

[0001] This invention relates to the field of health status monitoring technology, specifically a real-time health status monitoring system and method based on biosensors. Background Technology

[0002] Real-time health monitoring systems using biosensors are widely used in home health management and outdoor activity safety. They integrate sensors such as accelerometers, gyroscopes, and pulse oximeters to collect physiological data like posture, heart rate, and blood oxygen saturation. These data are then combined with image analysis techniques, including posture estimation and behavior recognition, to assess the user's health status. Primarily used for fall prevention and identifying sudden health risks such as altitude sickness, these systems provide real-time health assurance for the elderly and those active in high-altitude environments. However, existing health monitoring systems for the elderly in high-altitude environments often focus solely on fall risk or altitude sickness, failing to adequately consider the dynamic impact of the low-oxygen environment on balance and muscle control. This results in a high false alarm rate for traditional fall warning systems due to gait abnormalities caused by hypoxia at high altitudes, while altitude sickness warning systems only indicate the degree of hypoxia without correlating it with actual fall risk. These systems struggle to meet the precise health monitoring needs of this specific group of elderly people in high-altitude areas. Therefore, a real-time health monitoring system and method that integrates biosensor and image data is urgently needed to address these issues. Summary of the Invention

[0003] This application provides a real-time health status monitoring system and method based on biosensors, which solves the limitations of monitoring from a single data source and avoids the technical problems of traditional fall warnings that may falsely report due to normal movements such as bending over, or altitude sickness warnings that only indicate hypoxia but cannot be associated with fall risk.

[0004] To achieve the above objectives, the embodiments of this application disclose the following technical solutions:

[0005] On the one hand, this solution discloses a real-time health status monitoring method based on biosensors, including the following steps: S1: Collecting physiological data of the user through biosensors, including a posture sensor, heart rate sensor, blood oxygen sensor, and respiratory rate sensor integrated into a wearable device; S2: Preprocessing the physiological data to extract altitude sickness features and fall risk features; S3: Collecting user posture videos and environmental videos through an image acquisition device; S4: Extracting features from the posture videos and environmental videos to obtain abnormal posture features and environmental risk features; S5: Multimodal fusion of the altitude sickness features, fall risk features, abnormal posture features, and environmental risk features to construct a risk correlation matrix; S6: Inputting the risk correlation matrix into a pre-trained risk assessment model to output the current health risk level; S7: Triggering a corresponding warning based on the health risk level; S8: Storing the physiological data, video data, feature extraction results, health risk level, and warning information in a local memory and transmitting them to a remote terminal through a wireless communication module.

[0006] On the other hand, this solution discloses a real-time health status monitoring system based on biosensors, including:

[0007] The system includes a data acquisition module for collecting the user's physiological data via biosensors, including a posture sensor, heart rate sensor, blood oxygen sensor, and respiratory rate sensor integrated into the wearable device; a sensor data processing module for preprocessing the physiological data and extracting altitude sickness features and fall risk features; an image acquisition module for acquiring user posture videos and environmental videos via an image acquisition device; an image data processing module for extracting features from the posture videos and environmental videos to obtain abnormal posture features and environmental risk features; and a multimodal fusion module for integrating the altitude sickness features, fall risk features, and abnormal posture features. The system integrates common features and environmental risk features in a multimodal manner to construct a risk correlation matrix; a risk assessment module, which inputs the risk correlation matrix into a pre-trained risk assessment model and outputs the current health risk level; an early warning triggering module, which triggers a corresponding early warning based on the health risk level; a result output module, which stores the physiological data, video data, feature extraction results, health risk level, and early warning information in a local memory and transmits them to a remote terminal via a wireless communication module; a storage medium, which stores the pre-trained risk assessment model and historical data; and a processor, which executes the calculation process of the pre-trained risk assessment model.

[0008] This invention relates to a real-time health monitoring system and method based on biosensors. This solution utilizes multimodal data fusion, combining physiological data such as posture, heart rate, and blood oxygen collected by biosensors with image-based data on body posture and environmental video. This overcomes the limitations of single-data source monitoring and avoids the problems of false alarms from traditional fall warnings due to normal movements like bending over, or altitude sickness warnings that only indicate hypoxia without linking it to fall risk. By dynamically extracting features such as SpO2 decrease rate and center of gravity shift variance and constructing a risk correlation matrix, the inhibitory effect of high-altitude hypoxia on the balance ability of the elderly is quantified. Environmental factors such as terrain slope and walking speed are deeply correlated with physiological states such as hypoxia degree and gait stability, improving the accuracy of risk identification in complex environments. Combined with a pre-trained risk assessment model, a graded risk level is output, triggering targeted warnings such as slowing down for low-risk situations and audible and visual alarms for high-risk situations. This provides more accurate and timely health monitoring services for the elderly in high-altitude areas, effectively reducing the risk of falls caused by the interaction between the environment and physiological state. Attached Figure Description

[0009] Figure 1 This is a flowchart of the method according to Embodiment 1 of the present invention; Figure 2 This is an overall block diagram of the system according to Embodiment 2 of the present invention; Figure 3 This is an interaction diagram of the system in Embodiment 2 of the present invention. Detailed Implementation

[0010] Specific embodiments of the invention will now be described in detail. Although the invention is described in conjunction with these specific embodiments, it should be understood that the invention is not intended to be limited to these specific embodiments. Rather, these embodiments are intended to cover alternative, modified, or equivalent embodiments that may be included within the spirit and scope of the invention as defined by the claims. In the following description, numerous specific details are set forth in order to provide a thorough understanding of the invention. The invention may be practiced without some or all of these specific details. In other instances, well-known processes have not been described in detail so as not to unnecessarily obscure the invention.

[0011] When used in conjunction with the terms "comprising," "method comprising," or similar language in this specification and appended claims, the singular forms "a," "some," and "the" include plural references unless the context clearly indicates otherwise. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0012] Application Overview: Current technologies for health monitoring of the elderly in high-altitude environments largely rely on single-type sensors or independent fall warning and altitude sickness monitoring systems, failing to fully consider the dynamic interaction between the hypoxic environment of high altitudes and the physiological state and behavioral posture of the elderly. For example, traditional fall warning systems only identify gait abnormalities through posture sensors; however, in hypoxic environments at high altitudes, hypoxia can cause slight abnormalities in the elderly's normal gait, such as a shift in the center of gravity, leading to false alarms. While altitude sickness monitoring systems can indicate a decrease in blood oxygen saturation, they cannot correlate this with the actual impact of decreased balance due to hypoxia on fall risk, resulting in a disconnect between warnings and actual risks.

[0013] To address the aforementioned challenges, this application first considers combining biosensors with image acquisition technology to capture the correlation between the high-altitude environment, physiological state, and behavioral posture through multimodal data fusion. Physiological data is collected by integrating biosensors such as posture sensors and blood oxygen sensors, while body posture and environmental videos are captured using privacy-preserving cameras and depth cameras to extract features of altitude sickness, fall risk, abnormal body posture, and environmental risk. To resolve the interaction between the environment and physiological state, this application constructs a risk correlation matrix to quantify the inhibitory effect of hypoxia on balance ability and combines it with a pre-trained risk assessment model to output graded risk levels, triggering targeted warnings such as slowing down for low-risk situations and audible and visual alarms for high-risk situations.

[0014] Example 1

[0015] A real-time health status monitoring method based on biosensors includes the following steps:

[0016] S1: Collect physiological data of users through biosensors, including posture sensors, heart rate sensors, blood oxygen sensors and respiratory rate sensors integrated into wearable devices;

[0017] Integrating posture sensors, heart rate sensors, blood oxygen sensors, and respiratory rate sensors into wearable devices, such as sensors that fit against the skin or are attached to clothing, allows for real-time capture of changes in the user's posture, such as body tilt angle, heart rate fluctuations, blood oxygen saturation (SpO2), and respiratory rate. During data acquisition, the sensors obtain data at a frequency of 10-20 times per second, ensuring continuous dynamic monitoring.

[0018] S2: Preprocess the physiological data to extract characteristics of altitude sickness and fall risk;

[0019] The collected physiological data were denoised by removing motion interference through moving average filtering and standardizing the data units. For altitude sickness characteristics, real-time data was compared with baseline values ​​to extract the decreasing trend of SpO2 over time, the degree of deviation of respiratory rate from the normal range, and the magnitude of sudden increases and decreases in heart rate. For fall risk characteristics, changes in body angle recorded by posture sensors were used to calculate the sudden change in angle when leaning forward, as well as the degree of difference in gait amplitude between the left and right sides, i.e., the decrease in gait symmetry.

[0020] S3: Acquire user posture video and environmental video through image acquisition equipment;

[0021] Privacy-preserving cameras and depth cameras are deployed in the user activity area. The former is used to capture body posture video, and the latter is used to acquire 3D environmental information. During data acquisition, the devices and wearable sensors maintain time consistency via wireless synchronization signals such as Bluetooth to ensure that the timestamps of physiological data and video data match. S4: Feature extraction is performed on the body posture video and environmental video to obtain abnormal body posture features and environmental risk features;

[0022] For body posture videos, algorithms are used to identify user silhouettes and extract changes in the distance between the feet and the trajectory of the body's center of gravity. For environmental videos, the degree of terrain inclination and the user's movement speed are identified and converted into quantifiable risk features. S5: The altitude sickness features, fall risk features, abnormal body posture features, and environmental risk features are fused using a multimodal method to construct a risk correlation matrix.

[0023] By integrating physiological and video features, the impact of hypoxic environments on balance and the effects of terrain and walking status on physical strength are analyzed. A risk correlation matrix is ​​constructed to visually present the interactions between various factors. S6: Based on a pre-trained risk assessment model, the risk correlation matrix is ​​input into the model to output the current health risk level.

[0024] The risk correlation matrix is ​​input into the pre-trained model, which outputs the current risk level based on feature combinations, reflecting the user's likelihood of experiencing altitude sickness or a fall. S7: Trigger a corresponding warning based on the stated health risk level;

[0025] Depending on the risk level, at low risk, only a prompt message is displayed on the wearable device; at medium risk, the device vibrates and emits a low-volume alarm; at high risk, in addition to a local alarm, a distress signal is simultaneously sent to a remote terminal. S8: The physiological data, video data, feature extraction results, health risk level, and warning information are stored in the local memory and sent to the remote terminal via the wireless communication module.

[0026] All data is first stored in the device's local memory to prevent loss during network interruptions. Once the network is stable, it is sent to a remote terminal via 4G, 5G, or Wi-Fi for medical staff or family members to view historical records.

[0027] This embodiment further proposes that the altitude sickness characteristics mentioned in step S2 include the rate of SpO2 decrease, abnormal respiratory rate, and sudden heart rate; the fall risk characteristics include sudden forward tilt angle and gait symmetry decrease; and the baseline physiological data are obtained through statistical analysis of the user's daily activity data.

[0028] Baseline data establishment: During users' daily activities, such as living in plain areas, their physiological data are continuously collected, and the normal fluctuation range of heart rate, respiratory rate, and SpO2, as well as the baseline values ​​of walking posture angle and gait symmetry, are statistically analyzed as a reference for subsequent judgment of abnormalities.

[0029] Altitude sickness characteristic extraction: SpO2 decrease rate is measured by real-time monitoring of SpO2 values ​​and calculation of the decrease per unit time (e.g., per minute). If the decrease exceeds the baseline fluctuation range (e.g., SpO2 is stable at 95%-98% at sea level, and the rate of decrease exceeds 2% / minute at altitude), it is marked as an abnormal characteristic. Abnormal respiratory rate: Real-time respiratory rate is compared with the baseline value. If the respiratory rate at rest is consistently 1.5 times higher than the baseline, it is considered abnormal.

[0030] Heart rate mutation value is captured by a heart rate sensor to detect instantaneous changes in heart rate. If the heart rate suddenly increases or decreases in a short period of time, it is recorded as a mutation feature.

[0031] Fall risk feature extraction, forward tilt angle mutation value: The posture sensor records the forward tilt angle of the body in real time, which is about 5°-15° during normal walking. If the angle suddenly increases by more than 20° within 1 second, it indicates a tendency of body imbalance. Gait symmetry decline value: Analyze the difference in stride amplitude and time between the left and right legs. If the difference increases by more than 30% from the baseline and the real-time difference is more than 8cm, it is judged as a decline in gait symmetry.

[0032] This embodiment further proposes that the image acquisition device in step S3 includes a privacy-protecting camera and a depth camera, wherein the privacy-protecting camera adopts infrared imaging technology; the body posture video extracts the distance between the user's feet through an edge detection algorithm and extracts the center of gravity offset trajectory through a centroid coordinate algorithm; the environmental video extracts the terrain slope through Canny edge detection and Hough transform and extracts the walking speed through optical flow.

[0033] Privacy-protecting cameras use infrared imaging technology to capture only human outlines and movement trajectories, without recording facial or other private information, making them suitable for use in private settings at home or outdoors. Depth cameras acquire environmental depth information by emitting infrared light to determine terrain elevation differences. The devices should be installed at a high point in the user's activity area, such as in a corner or among tree branches, ensuring coverage of the user's walking path, and the lens angle should avoid directly shining into private areas.

[0034] Foot spacing extraction: An edge detection algorithm is used to identify the user's foot contours, calculate the straight-line distance between the outermost points of the two feet, and continuously record the changes in this distance to determine whether there is a lack of balance when the feet are together.

[0035] The center of gravity offset trajectory extraction uses a centroid coordinate algorithm to treat the human body as a point mass and calculates the center of gravity coordinates at different times. Based on the positions of various parts of the body obtained by a depth camera, the coordinate points are connected to form a trajectory, which reflects whether the center of gravity is stable.

[0036] The terrain slope extraction process applies Canny edge detection to environmental video frames to identify ground contour edges. Then, Hough transform is used to convert the edges into straight lines. The angle between the straight line and the horizontal line is calculated, which is the terrain slope. If the angle exceeds 30°, it is determined to be a steep slope.

[0037] Walking speed extraction uses optical flow to track the displacement of feature points of the user's silhouette, such as shoulders and feet, in consecutive frames in the video. The movement speed is calculated by combining the frame interval time to reflect whether the walking is fast or slow.

[0038] This embodiment further proposes that the abnormal body posture features mentioned in step S4 include the reduction in the distance between the feet and the variance of the center of gravity shift; the environmental risk features include steep slope signs and rapid walking signs.

[0039] The reduction in the distance between the feet is based on the distance between the user's feet when walking normally. The difference between the real-time distance and the baseline value is calculated. If the difference is negative and the absolute value exceeds 20cm, it is judged as a reduction in the distance, indicating a decrease in balance ability.

[0040] The center of gravity offset variance is calculated by statistically analyzing the coordinate data of the center of gravity offset trajectory. The variance value reflects the degree of trajectory dispersion. If the variance increases by more than 50% compared with the normal state, it is marked as an abnormal center of gravity offset.

[0041] The steep slope marker is combined with the terrain slope extraction results. When the slope exceeds 30°, it is automatically marked as a steep slope, indicating that there is a risk of falling.

[0042] The "fast walking" indicator compares walking speed to the user's normal walking speed. If the speed exceeds 1.5 times and lasts for more than 10 seconds, and this is combined with terrain slope such as walking quickly on a steep slope, it is marked as "fast walking," indicating that physical strength is being depleted too quickly.

[0043] This embodiment further proposes that the risk correlation matrix in step S5 is constructed in the following way: calculating the hypoxia balance correlation value to quantify the impact of the hypoxic environment on the user's balance ability; calculating the environmental behavior correlation value to quantify the comprehensive impact of terrain and walking speed on the user's physical exertion.

[0044] The risk correlation matrix uses physiology, physical condition, and environment as three-dimensional coordinate axes. Each dimension contains corresponding features, and the elements in the matrix are the correlation values ​​between the features, intuitively presenting the mutual influence between hypoxic environment, physical condition, and environmental factors.

[0045] The hypoxia-balance correlation value is used to measure the impact of low-oxygen environments, such as high-altitude environments, on a user's balance ability. It integrates the SpO2 decline rate to reflect the degree of hypoxia, the variance of the center of gravity shift to reflect the balance state, and the abrupt change in the forward lean angle to reflect body stability, comprehensively determining whether hypoxia exacerbates balance impairment. When SpO2 drops rapidly, if a significant shift in the center of gravity and a sudden forward lean occur simultaneously, the correlation value increases, indicating that hypoxia has significantly affected balance.

[0046] The environmental behavior correlation value is used to measure the combined impact of terrain and walking speed on physical strength. It integrates the steep slope indicator to reflect the terrain difficulty and the rapid walking indicator to reflect the intensity of exercise. It determines whether the user is overexerting physical strength in a high-difficulty environment. When walking rapidly on a steep slope, the correlation value increases significantly, indicating that the physical strength consumption is far beyond the normal level, which may easily lead to fatigue falls or aggravate altitude sickness.

[0047] Using the hypoxia balance correlation value and the environmental behavior correlation value as the core elements of the matrix, supplemented by the values ​​of each original feature, a complete risk correlation matrix is ​​formed, providing comprehensive input for subsequent risk assessment.

[0048] This embodiment further proposes that the formula for calculating the hypoxia balance correlation value is:

[0049] ;

[0050] in, This indicates the correlation value for hypoxia balance. Indicates the rate of decrease of SpO2. Indicates the variance of the centroid shift. Indicates the abrupt change in the lean angle;

[0051] In the formula, the rate of decrease of SpO2 The highest weight is 0.5, because hypoxia is a core characteristic of the high-altitude environment, directly affecting bodily functions; center of gravity shift variance. The second-highest weight is 0.3, because decreased balance is a direct cause of falls; the sudden change in forward tilt angle... The weight is low at 0.2 because it is a momentary manifestation of imbalance.

[0052] The higher the value, the greater the negative impact of oxygen deficiency on balance. The formula for calculating the environmental behavior correlation value is:

[0053] ;

[0054] in, Indicates the correlation value between environmental behavior. Indicates a steep slope. A sign indicating hurried walking.

[0055] The steep slope indicator P (0.6) has a higher weight than the brisk walking indicator W (0.4) because the terrain difficulty has a more lasting impact on physical exertion. Here, P and W are Boolean values ​​converted from their corresponding values; for example, a steep slope indicator is 1, and a non-steep slope indicator is 0; brisk walking is 1, and normal walking is 0.

[0056] The higher the value, the higher the overall risk of the environment and behavior.

[0057] This embodiment further proposes that the pre-trained risk assessment model in step S6 is a logistic regression model, with the input being the feature vector of the risk correlation matrix and the output being the health risk level; the health risk level includes low risk, medium risk, and high risk.

[0058] We collect extensive physiological data, video features, and actual health events such as falls and altitude sickness incidents from users in high-altitude environments. These data are then labeled as low, medium, or high risk: no event indicates low risk, mild discomfort indicates medium risk, and a fall or severe reaction indicates high risk. Using the feature vectors of the risk correlation matrix (A and B values) and the original feature values ​​as inputs, and the risk level as the output, we train the model using a logistic regression algorithm until the prediction accuracy reaches a target of >90%.

[0059] After acquiring the risk correlation matrix in real time, feature vectors such as A value, B value, and SpO2 decrease rate are extracted and input into the pre-trained model. The model calculates the weighted sum of each feature and compares it with preset thresholds such as low risk < 0.3, 0.3 ≤ medium risk < 0.7, and high risk ≥ 0.7 to output the corresponding risk level.

[0060] The system periodically feeds back newly collected user data, including actual risk results, to the model for retraining and parameter adjustment, ensuring that the model adapts to individual differences among users, such as age, health status, and environmental changes, such as different altitudes.

[0061] Example 2

[0062] A real-time health status monitoring system based on biosensors includes:

[0063] The data acquisition module is used to collect the user's physiological data through biosensors, including a posture sensor, heart rate sensor, blood oxygen sensor, and respiratory rate sensor integrated into the wearable device.

[0064] The blood oxygen sensor illuminates the finger with red and infrared light and calculates SpO2 based on the amount of light absorbed; the posture sensor captures changes in body angle using an accelerometer and a gyroscope. The sensor data processing module is used to preprocess the physiological data and extract characteristics of altitude sickness and fall risk.

[0065] After receiving the raw physiological signals, the system first filters out motion interference and amplifies weak signals. Then, it calls the feature extraction algorithm to separate the features of altitude sickness and fall risk, and transmits the results to the multimodal fusion module.

[0066] The image acquisition module is used to acquire user posture videos and environmental videos through image acquisition devices;

[0067] The privacy protection camera and the depth camera start simultaneously. The former outputs infrared body movement video, and the latter outputs environmental depth video. The video data is compressed and then transmitted to the image data processing module.

[0068] The image data processing module is used to extract features from the body posture video and the environmental video to obtain abnormal body posture features and environmental risk features;

[0069] After decoding the video data, edge detection and centroid coordinates are used to extract abnormal body features, and Canny edge detection and optical flow are used to extract environmental risk features. The results are then sent to the multimodal fusion module.

[0070] The multimodal fusion module is used to fuse the altitude sickness characteristics, fall risk characteristics, abnormal physical characteristics, and environmental risk characteristics in a multimodal manner to construct a risk correlation matrix;

[0071] After receiving physiological and video features, the system calculates hypoxia balance correlation values ​​and environmental behavior correlation values ​​according to preset logic, constructs a risk correlation matrix, and transmits it to the risk assessment module. The risk assessment module, based on a pre-trained risk assessment model, inputs the risk correlation matrix into the model and outputs the current health risk level.

[0072] The pre-trained model is loaded from the storage medium, the risk correlation matrix is ​​input, calculations are performed, and the risk level is output and sent to the early warning triggering module. The early warning triggering module is used to trigger a corresponding early warning based on the stated health risk level.

[0073] The system activates corresponding early warning mechanisms based on risk levels. For example, low-risk areas only display warnings, while high-risk areas receive both audible and visual alarms plus remote notifications. Simultaneously, the warning information is sent to the result output module. The result output module stores the physiological data, video data, feature extraction results, health risk level, and warning information in a local memory and transmits them to a remote terminal via a wireless communication module.

[0074] All data, including physiological, video, feature, risk level, and early warning classifications, are stored locally on a memory card such as an SD card and transmitted to remote terminals such as a mobile app or hospital monitoring platform via wireless communication modules (4G, Bluetooth). The storage medium stores the pre-trained risk assessment model and historical data; the processor executes the calculation process of the pre-trained risk assessment model.

[0075] This embodiment further proposes that the sensor data processing module includes: a high altitude reaction feature extraction unit, used to calculate the SpO2 decrease rate, abnormal respiratory rate values, and sudden heart rate values; and a fall risk feature extraction unit, used to calculate the forward tilt angle sudden value and gait symmetry decrease value.

[0076] The image data processing module includes: a body posture abnormality feature extraction unit, used to calculate the reduction value of the distance between the two feet and the variance of the center of gravity shift; and an environmental risk feature extraction unit, used to identify steep slope signs and rapid walking signs.

[0077] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them; although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications can still be made to the specific implementation methods of the present invention or equivalent substitutions can be made to some technical features without departing from the spirit of the technical solutions of the present invention, and all such modifications should be covered within the scope of the technical solutions claimed in the present invention.

Claims

1. A method for real-time health status monitoring based on biosensors, characterized in that, Includes the following steps: S1: Collect physiological data of users through biosensors, including posture sensors, heart rate sensors, blood oxygen sensors and respiratory rate sensors integrated into wearable devices; S2: Preprocess the physiological data to extract altitude sickness features and fall risk features; the altitude sickness features include SpO2 decrease rate, abnormal respiratory rate, and heart rate mutation; the fall risk features include forward tilt angle mutation and gait symmetry decrease; baseline physiological data are obtained through statistical analysis of users' daily activity data; S3: Acquire user posture video and environmental video through image acquisition equipment; S4: Extract features from the body posture video and environmental video to obtain abnormal body posture features and environmental risk features; the abnormal body posture features include the reduction in the distance between the feet and the variance of the center of gravity shift; the environmental risk features include steep slope signs and rapid walking signs; S5: Multimodal fusion of the aforementioned altitude sickness characteristics, fall risk characteristics, abnormal physical characteristics, and environmental risk characteristics to construct a risk correlation matrix; S6: Based on the pre-trained risk assessment model, input the risk correlation matrix into the model and output the current health risk level; S7: Trigger the corresponding warning based on the stated health risk level; S8: Store the physiological data, video data, feature extraction results, health risk level and early warning information in the local memory, and send them to the remote terminal through the wireless communication module; The risk correlation matrix is ​​constructed by calculating the hypoxia balance correlation value, which is used to quantify the impact of hypoxic environment on the user's balance ability. Calculate environmental behavior correlation values ​​to quantify the combined impact of terrain and walking speed on user physical exertion; The formula for calculating the hypoxia balance correlation value is as follows: ； in, This indicates the correlation value for hypoxia balance. Indicates the rate of decrease of SpO2. Indicates the variance of the centroid shift. Indicates the abrupt change in the lean angle; The formula for calculating the environmental behavior correlation value is as follows: ； in, Indicates the correlation value between environmental behavior. Indicates a steep slope. A sign indicating hurried walking.

2. The real-time health status monitoring method based on biosensors according to claim 1, characterized in that, The image acquisition device in step S3 includes a privacy-protecting camera and a depth camera. The privacy-protecting camera uses infrared imaging technology. The body posture video extracts the distance between the user's feet using an edge detection algorithm and extracts the center of gravity offset trajectory using a centroid coordinate algorithm. The environmental video extracts the terrain slope using Canny edge detection and Hough transform and extracts the walking speed using optical flow.

3. The real-time health status monitoring method based on biosensors according to claim 1, characterized in that, The pre-trained risk assessment model mentioned in step S6 is a logistic regression model, with the input being the feature vector of the risk correlation matrix and the output being the health risk level; The health risk levels are categorized as low risk, medium risk, and high risk.

4. A real-time health status monitoring system based on biosensors, characterized in that, include: The data acquisition module is used to collect the user's physiological data through biosensors, including a posture sensor, heart rate sensor, blood oxygen sensor, and respiratory rate sensor integrated into the wearable device. The sensor data processing module is used to preprocess the physiological data and extract characteristics of altitude sickness and fall risk. The image acquisition module is used to acquire user posture videos and environmental videos through image acquisition devices; The image data processing module is used to extract features from the body posture video and the environmental video to obtain abnormal body posture features and environmental risk features; The multimodal fusion module is used to fuse the altitude sickness characteristics, fall risk characteristics, abnormal physical characteristics, and environmental risk characteristics in a multimodal manner to construct a risk correlation matrix; The risk assessment module is used to input the risk correlation matrix into a pre-trained risk assessment model and output the current health risk level. The early warning triggering module is used to trigger corresponding early warnings based on the health risk level. The result output module is used to store the physiological data, video data, feature extraction results, health risk level and early warning information in the local memory, and send them to the remote terminal through the wireless communication module. Storage medium for storing the pre-trained risk assessment model and historical data; A processor for executing the computational process of the pre-trained risk assessment model; Baseline data unit: Baseline physiological data is obtained through statistical analysis of users' daily activity data; The sensor data processing module includes: a high altitude reaction feature extraction unit, used to calculate the SpO2 decrease rate, abnormal respiratory rate values, and sudden heart rate values; and a fall risk feature extraction unit, used to calculate the forward tilt angle change value and the gait symmetry decrease value. The image data processing module includes: a body posture abnormality feature extraction unit, used to calculate the reduction value of the distance between the two feet and the variance of the center of gravity shift; and an environmental risk feature extraction unit, used to identify steep slope signs and rapid walking signs.

Images