Diabetic foot ulcer early warning system and method based on multi-mode physiological signal fusion
Through a diabetic foot ulcer warning system that combines multi-dimensional sensors and edge computing with cloud-based collaborative computing, the problems of insufficient detection dimensions of existing equipment, poor model interpretability and limited real-time response capabilities are solved, and personalized health management and efficient early warning are achieved.
Patent Information
- Application Number
- CN202510521734.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-04-24
- Publication Date
- 2025-07-29
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The existing diabetic foot monitoring equipment has insufficient detection dimensions, poor interpretability of the model, limited real-time response capabilities, and weak personalized adaptability, which cannot meet the needs of timely clinical intervention.
The multi-dimensional sensor component is used to combine edge computing and cloud-based collaborative computing, and integrate temperature, pressure, blood oxygen and humidity sensors to perform data preprocessing and personalized evaluation through lightweight models, and combine random forest algorithms and time series fusion algorithms to provide personalized health assessments and real-time early warnings.
It significantly improves monitoring performance and user experience, improves the accuracy and real-time nature of disease assessment, reduces medical management costs, and realizes personalized adaptation and clinical practicality.
Smart Images
Figure CN120388740A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of medical and health technologies, and particularly to a diabetic foot ulcer early warning system and method based on multi-modal physiological signal fusion. Background Art
[0002] Diabetic foot is a common and serious complication of diabetic patients, mainly manifested as foot vascular lesions and neuropathy, which may lead to ulcers, infections, and even amputations. Many diabetic patients will experience diabetic foot ulcers in their lifetime. At present, most of the diabetic foot monitoring devices on the market adopt traditional algorithms combined with single or multi-modal sensors, and have the following limitations: (1) Insufficient detection dimensions: Existing devices can only provide local temperature and pressure monitoring, and cannot comprehensively evaluate the blood supply of the foot.
[0003] (2) Poor model interpretability: Some products adopt "black box" deep learning models, and it is difficult for clinicians to understand the prediction basis. Among them, the "black box model" refers to an algorithm model whose decision logic between input and output is unexplainable or difficult to trace, and its internal operation mechanism is opaque to users (even developers).
[0004] (3) Limited real-time response ability: The cloud centralized computing mode leads to early warning delays and fails to meet the needs of clinical timely intervention.
[0005] (4) Weak personalized adaptation ability: Lack of feature adaptive mechanisms for different patients. Summary of the Invention
[0006] The main purpose of the present invention is to overcome the deficiencies in the prior art and provide a diabetic foot ulcer early warning system and method based on multi-modal physiological signal fusion.
[0007] The technical solution adopted by the present invention to achieve its technical purpose is: A diabetic foot ulcer early warning system based on multi-modal physiological signal fusion, including a multi-dimensional sensor component, a data preprocessing system, a communication system, and a personalized health assessment system.
[0008] The multi-dimensional sensor component is used to perform daily health real-time monitoring on the diabetic foot, and the data preprocessing system is communicatively connected to the multi-dimensional sensor component and collects and preprocesses the data monitored by the multi-dimensional sensor in real time.
[0009] Preferably, the multi-dimensional sensor component includes a temperature sensor, a pressure sensor, a blood oxygen saturation sensor, and a microenvironment humidity sensor; The temperature sensor monitors the temperature changes of the diabetic foot in real time, reflecting and identifying local inflammation or abnormal blood supply conditions. The pressure sensor detects the deviation of the plantar force line and the pressure concentration area. The blood oxygen saturation sensor evaluates the local tissue oxygenation status, reflecting and detecting microcirculation disorders. The microenvironment humidity sensor monitors the humidity inside the shoe to prevent fungal infections.
[0010] Preferably, the data preprocessing system includes a data acquisition unit and an edge computing board; The data acquisition unit collects in real time the various data monitored by the multi-dimensional sensor component in real time. The edge computing board collects and preprocesses the data in real time based on the edge-cloud collaborative computing architecture, compresses and uploads the feature vectors to the cloud platform, and provides personalized prediction services.
[0011] Specifically, the edge computing board collects and preprocesses the data in real time through the local edge node, conducts preliminary analysis on the device side through a lightweight model, compresses and uploads the feature vectors to the cloud platform, then performs deep learning model training and updating through a remote server, and provides personalized prediction services.
[0012] The communication system uploads the data information processed by the data preprocessing system to the cloud platform. The personalized health assessment system establishes an information connection with the cloud platform, conducts real-time assessment and prediction on the current diabetic foot data, provides reasonable recommended solutions, and then feeds back the proposed recommended solutions to the diabetic foot patients through the cloud platform.
[0013] Preferably, the communication system uses Bluetooth or low-power wide area network technology to upload the compressed data to the cloud platform.
[0014] Preferably, the personalized health assessment system calculates and analyzes the individual characteristics of the patient, such as the course of the disease, the severity of neuropathy, etc., through the cloud platform, can automatically generate personalized monitoring frequency recommendations and nursing plan recommendations, and provides a remote medical consultation entrance to timely link professional doctors for intervention.
[0015] Preferably, the specific assessment steps of the personalized health assessment system are as follows: S1. Collect and clean the basic information of diabetic foot patients, collecting basic information such as the diabetic history, foot complication history, medication situation, etc. of diabetic foot patients; S2. Individual characteristic modeling: Use the random forest algorithm to identify the main factors affecting the ulcer risk; S3. Prediction model optimization: Adjust the weight coefficients of the model according to the individual parameters of diabetic foot patients to improve the prediction accuracy; S4, Real-time feedback and suggestion generation: Automatically generate personalized nursing plan prompts (such as suggesting to change shoes and socks, enhancing foot moisturization, etc.).
[0016] S5, When the risk score reaches the warning threshold, automatically send a medical reminder.
[0017] Preferably, it further includes a time series fusion algorithm system. The time series fusion algorithm system sets the warning threshold and hierarchical response criteria based on a medical expert knowledge base, adopts a dynamic weight allocation strategy, adaptively adjusts the model's attention degree according to the importance of different physiological parameters, and introduces an attention mechanism to optimize the ability to capture long-term dependencies.
[0018] The present invention also provides a method for warning diabetic foot ulcers based on multi-modal physiological signal fusion. It uses the warning system for diabetic foot ulcers based on multi-modal physiological signal fusion as described above. The specific steps are as follows: S1, Extract the key features of the data collected by the multi-dimensional sensor component 1; including the temperature difference between different areas of the foot collected by the temperature sensor, the temperature change rate data; the pressure center offset and peak pressure area data collected by the pressure sensor; the dynamic fluctuation amplitude and frequency data of blood oxygen collected by the blood oxygen saturation sensor; the local relative humidity and its change trend data in a specific time window collected by the microenvironment humidity sensor; S2, Preprocess the data collected by the sensors in the multi-dimensional sensor component 1. Denoise and smooth the temperature, pressure, blood oxygen saturation, and humidity signals respectively, and align the signals with different sampling frequencies to the same time reference through interpolation; S3, Build a fusion model. Use the LSTM network to capture long-term dependencies in the time series, introduce the Self-attention mechanism, and dynamically allocate the attention weights of different features; establish a hierarchical warning scoring system, which is divided into three levels: low risk, medium risk, and high risk; S4, Adjust the weight coefficient of the model according to the patient's individual parameters to improve the prediction accuracy and automatically generate personalized nursing plan prompts.
[0019] S5, When the predicted risk score reaches the warning threshold, automatically send a medical reminder.
[0020] Compared with the prior art, the beneficial effects of the present invention are: The warning system and method for diabetic foot ulcers based on multi-modal physiological signal fusion significantly improve the monitoring performance and user experience by integrating multi-modal sensors and innovative algorithms. Multi-modal data fusion significantly improves the monitoring performance. Edge computing optimizes the response speed and energy consumption to support personalized adaptation, significantly improving the accuracy and real-time nature of the condition assessment. Moreover, considering the physiological differences of different patients, the overall scheme design focuses on clinical practicability and user experience.
[0021] In summary, the system and method not only improve the effect of foot health management for diabetic patients, but also reduce the cost of long-term follow-up management in medical institutions. Through continuous learning and optimization of massive clinical data, the intelligent level and service ability of the system are further improved. BRIEF DESCRIPTION OF THE DRAWINGS
[0022] Figure 1 FIG. is a structural flowchart of a diabetic foot ulcer early warning system based on multi-modal physiological signal fusion.
[0023] Figure 2 FIG. is a flowchart of a diabetic foot ulcer early warning method based on multi-modal physiological signal fusion.
[0024] Wherein: 1 - includes a multi-dimensional sensor component; 2 - a data preprocessing system; 3 - a communication system; 4 - a personalized health assessment system; 5 - a cloud platform; 6 - a time series fusion algorithm system. DETAILED DESCRIPTION OF THE EMBODIMENTS
[0025] In the description of the present invention, it should be noted that the orientation or positional relationship indicated by the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc. is based on the orientation or positional relationship shown in the drawings, or the orientation or positional relationship in which the product of the present invention is usually placed during use. It is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and thus cannot be understood as a limitation to the present invention. In addition, the terms "first", "second", "third", etc. are only used for distinguishing descriptions and cannot be understood as indicating or implying relative importance.
[0026] In the description of the present invention, it should also be noted that unless otherwise clearly specified and defined, the terms "set", "installed", "connected", "connected" should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be directly connected or indirectly connected through an intermediate medium, and it can be the communication inside two elements. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific situations.
[0027] To make the purpose, technical solution and advantages of the present invention clearer, the present invention will be further described in detail below through the drawings and embodiments. However, it should be understood that the specific embodiments described herein are only used to explain the present invention and do not limit the scope of the present invention. In addition, in the following description, the description of well-known structures and technologies is omitted to avoid unnecessarily confusing the concepts of the present invention. Example 1
[0028] Please refer to Figure 1 - A diabetic foot ulcer warning system based on multi-modal physiological signal fusion, including a multi-dimensional sensor component 1, a data preprocessing system 2, a communication system 3, a personalized health assessment system 4, a cloud platform 5, and a time series fusion algorithm system 6; The multi-dimensional sensor component 1 includes a temperature sensor, a pressure sensor, a blood oxygen saturation sensor, and a microenvironment humidity sensor; the temperature sensor monitors the temperature change of the diabetic foot in real time to reflect and identify local inflammation or abnormal blood supply, the pressure sensor detects the deviation of the plantar force line and the pressure concentration area, the blood oxygen saturation sensor evaluates the local tissue oxygenation state to reflect and detect microcirculation disorders, and the microenvironment humidity sensor monitors the humidity inside the shoe to prevent fungal infections.
[0029] The data preprocessing system 2 includes a data acquisition unit and an edge computing board; the data acquisition unit collects the data monitored by the multi-dimensional sensor component 1 in real time; the edge computing board collects and preprocesses the data in real time based on the edge-cloud collaborative computing architecture, compresses and uploads the feature vectors to the cloud platform 5, and provides personalized prediction services.
[0030] Specifically, the edge computing board collects and preprocesses the data through the local edge node, performs preliminary analysis on the device side through a lightweight model, compresses and uploads the feature vectors to the cloud platform 5, then performs deep learning model training and update through a remote server, and provides personalized prediction services.
[0031] The communication system 3 uploads the compressed data to the cloud platform 5 using Bluetooth or low-power wide area network technology.
[0032] The personalized health assessment system 4 calculates and analyzes the individual characteristics of the patient, such as the course of the disease, the severity of neuropathy, etc., through the cloud platform 5, can automatically generate personalized monitoring frequency suggestions and nursing plan recommendations, and provides a remote medical consultation entrance to promptly link professional doctors for intervention.
[0033] Furthermore, in this embodiment, the specific evaluation steps of the personalized health assessment system 4 are as follows: S1. Collect and clean the basic information of diabetic foot patients, collecting basic information such as the diabetic history, foot complication history, and medication situation of diabetic foot patients; S2. Individual characteristic modeling: Use the random forest algorithm to identify the main factors affecting the ulcer risk; S3. Prediction model optimization: Adjust the weight coefficient of the model according to the individual parameters of diabetic foot patients to improve the prediction accuracy; S4, Real-time feedback and suggestion generation: Automatically generate personalized nursing plan prompts (such as suggesting to change shoes and socks, strengthen foot moisturization, etc.).
[0034] S5, When the risk score reaches the warning threshold, automatically send a medical reminder.
[0035] The time series fusion algorithm system 6 sets the warning threshold and hierarchical response standard based on the medical expert knowledge base, adopts a dynamic weight allocation strategy, adaptively adjusts the model's attention degree according to the importance of different physiological parameters, and introduces an attention mechanism to optimize the ability to capture long-term dependencies.
[0036] Through the multi-dimensional sensor component 1, daily health real-time monitoring of diabetic feet is carried out. The data preprocessing system 2 is communicatively connected to the multi-dimensional sensor component 1, and the data monitored by the multi-dimensional sensor is collected and preprocessed in real time; The communication system 3 uploads the data information processed by the data preprocessing system 2 to the cloud platform 5. The personalized health assessment system 4 establishes an information connection with the cloud platform, conducts real-time evaluation and prediction on the current diabetic foot data and provides a reasonable suggestion plan, and then feeds back the proposed suggestion plan to the diabetic foot patient through the cloud platform 5.
[0037] Specifically, when in use, the temperature sensor is used to monitor the temperature change of the diabetic foot in real time, the pressure sensor detects the plantar force line deviation and pressure concentration area, the blood oxygen saturation sensor evaluates the local tissue oxygenation state, and the microenvironment humidity sensor monitors the humidity inside the shoes. Then, the data acquisition unit collects the data monitored by the multi-dimensional sensor component 1 in real time; the edge computing board card collects and preprocesses the data in real time based on the edge-cloud collaborative computing architecture, compresses the feature vector and uploads it to the cloud platform 5 through the communication system 3, and provides personalized prediction services. The communication system 3 uploads the compressed data to the cloud platform 5 using Bluetooth or low-power wide area network technology. Then, through the personalized health assessment system 4, calculating and analyzing patient individual characteristics such as disease course, severity of neuropathy, etc., it can automatically generate personalized monitoring frequency suggestions and nursing plan recommendations, and provide a remote medical consultation entrance to timely link professional doctors for intervention. Embodiment 2
[0038] Please refer to Figure 2 , based on the above embodiment, the embodiment of the present invention further provides a diabetic foot ulcer warning method based on multi-modal physiological signal fusion, which uses the diabetic foot ulcer warning system based on multi-modal physiological signal fusion as described above, and its specific steps are as follows: S1. Extract the key features of the data collected by the multi-dimensional sensor component 1, including the temperature difference between different areas of the foot collected by the temperature sensor, the temperature change rate data, the pressure center offset and peak pressure area data collected by the pressure sensor, the dynamic fluctuation amplitude and frequency data of blood oxygen saturation collected by the blood oxygen saturation sensor, and the local relative humidity and its change trend data within a specific time window collected by the microenvironment humidity sensor; S2. Preprocess the data collected by the sensors in the multi-dimensional sensor component 1, denoise and smooth the temperature, pressure, blood oxygen saturation, and humidity signals respectively, and align the signals with different sampling frequencies to the same time reference through interpolation; S3. Construct a fusion model, use the LSTM network to capture the long-term dependencies in the time series, introduce the Self-attention mechanism to dynamically allocate the attention weights of different features, and establish a hierarchical early warning scoring system, which is divided into three levels: low risk, medium risk, and high risk; Among them, the division of the three levels of low risk, medium risk, and high risk is based on a dual determination mechanism of dynamic threshold + state machine migration to set the risk score, and the risk score corresponds to the current medical and biological state. The specific method is as follows: When the risk score is greater than or equal to "0" and less than 0.4, it is classified as low risk. At this time, the corresponding medical state is that the biological parameter fluctuations are within the normal range of the individual baseline, and intervention can be carried out by adopting daily care education (such as foot cleaning guidance); When the risk score is greater than or equal to "0.4" and less than 0.7, it is classified as medium risk. At this time, the corresponding medical state is that at least two parameters are abnormal but do not reach the pathological migration condition, and intervention can be carried out by adopting remote medical consultation + local physical therapy; When the risk score is greater than or equal to "0.7" and less than or equal to 1.0, it is classified as high risk. At this time, the corresponding medical state is that it meets the state machine migration to the "inflammation / ischemia" stage, and intervention can be carried out by adopting an emergency outpatient service + antibiotic / vascular intervention treatment.
[0039] S4. Adjust the weight coefficients of the model according to the patient's individual parameters to improve the prediction accuracy and automatically generate personalized nursing plan prompts.
[0040] S5. When the predicted risk score reaches the early warning threshold, automatically send a medical reminder.
[0041] The solution in this embodiment can be selectively combined and used with the solutions in other embodiments. Embodiment 3
[0042] Please refer to the figure. Based on the above Embodiment 1 and Embodiment 2, the present invention specifically implements and applies the diabetes foot ulcer early warning system and method based on multi-modal physiological signal fusion as follows: S1. Perform hardware configuration; Temperature sensor: DS18B20 sensor, sampling frequency 1Hz, error range ±0.5°C; Pressure sensor: FSR402 sensor, high sensitivity, suitable for long-term wear; Humidity sensor: HDC2080 sensor, with dust and waterproof characteristics; Blood oxygen sensor: MAX30102 module, measuring by combining infrared and red light.
[0043] S2. Data fusion process; Signal acquisition and synchronization: Synchronous acquisition of multi-modal data to ensure consistent timestamps; Use accelerometer to assist in calibrating data quality; Feature extraction method: Temperature: Calculate local temperature difference and change rate; Pressure: Analyze pressure distribution and concentrated areas; Humidity: Monitor skin moisture level and evaluate the risk of bacterial growth; Blood oxygen: Continuously monitor tissue perfusion status; S3. Data acquisition of the edge computing board based on the edge-cloud collaborative computing architecture; Use Raspberry Pi 4B for real-time data processing; Integrate lightweight neural network models (such as MobileNet); Implement preliminary risk assessment to reduce the load on the cloud platform.
[0044] S4. Cloud platform: Store historical data for in-depth analysis, long-term data storage and trend analysis; Regularly update the monitoring model and optimize parameters; Provide a user interface for patients and doctors to view the results.
[0045] S5. Experimental design: Control group setting: Control device: Existing mainstream monitoring products in the market, single-dimensional monitoring; Samples: 200 diabetic patients, randomly divided into two groups of 100 each; Time span: 3 months of continuous monitoring.
[0046] S6. Clinical validation results: Sensitivity: 92%, an increase of about 24% Specificity: 90%, false alarm rate reduced to an average of 0.7 times per day Accuracy: 91%, superior to existing methods (about 85%) Response time: The processing delay of the edge computing unit is < 3 seconds.
[0047] S7. Conclusions: 1. Improvement in accuracy: The accuracy of comprehensive risk assessment reaches 92%, an increase of about 24 percentage points compared to traditional single-modal solutions.
[0048] The average time for early symptom detection is advanced by 7 days.
[0049] 2. Optimization of response speed: Based on the edge computing architecture, the core processing delay is reduced to the millisecond level (< 100ms).
[0050] 3. Enhancement of clinical interpretability: Provide a visual interface for multi-modal attention weights, facilitating understanding and verification by medical staff.
[0051] 4. Improvement in personalized adaptability: Each user has an independent baseline model and warning threshold, enabling precise health management.
[0052] It should be noted that although the above embodiments have been described in this article, the patent protection scope of the present invention is not limited thereby. Therefore, based on the innovative concept of the present invention, any changes and modifications made to the embodiments described in this article, or equivalent structural, equivalent process, or equivalent functional transformations made using the content of the specification and drawings of the present invention, directly or indirectly applying the above technical solutions to other related technical fields, are all included in the protection scope of the present invention's patent.
Claims
1. A diabetic foot ulcer warning system based on multimodal physiological signal fusion, characterized in that: It includes a multi-dimensional sensor component (1), a data preprocessing system (2), a communication system (3), and a personalized health assessment system (4); The multi-dimensional sensor component (1) is used to monitor the daily health of the diabetic foot in real time. The data preprocessing system (2) is communicatively connected to the multi-dimensional sensor component (1), and collects and preprocesses the data monitored by the multi-dimensional sensor in real time; The communication system (3) uploads the data information processed by the data preprocessing system (2) to the cloud platform (5). The personalized health assessment system (4) establishes an information connection with the cloud platform, evaluates and predicts the data of the current diabetic foot in real time, provides a reasonable suggestion plan, and then feeds back the proposed suggestion plan to the diabetic foot patient through the cloud platform (5).
2. The diabetes foot ulcer early warning system based on multi-modal physiological signal fusion according to claim 1, wherein: The multi-dimensional sensor component (1) includes a temperature sensor, a pressure sensor, a blood oxygen saturation sensor, and a microenvironment humidity sensor; The temperature sensor monitors the temperature change of the diabetic foot in real time, reflects and identifies local inflammation or abnormal blood supply conditions. The pressure sensor detects the deviation of the plantar force line and the pressure concentration area. The blood oxygen saturation sensor evaluates the local tissue oxygenation state and reflects the discovery of microcirculation disorders. The microenvironment humidity sensor monitors the humidity inside the shoe to prevent fungal infections.
3. The diabetes foot ulcer early warning system based on multimodal physiological signal fusion according to claim 1, characterized in that: The data preprocessing system (2) includes a data acquisition unit and an edge computing board; The data acquisition unit collects the data monitored by the multi-dimensional sensor component (1) in real time. The edge computing board collects and preprocesses the data in real time based on the edge-cloud collaborative computing architecture, compresses and uploads the feature vectors to the cloud platform (5), and provides personalized prediction services.
4. The diabetic foot ulcer warning system based on multi-modal physiological signal fusion according to claim 1, characterized in that: The communication system (3) uploads the compressed data to the cloud platform (5) using Bluetooth or low-power wide area network technology.
5. The diabetes foot ulcer warning system based on multi-modal physiological signal fusion according to claim 1, wherein: The personalized health assessment system (4) calculates and analyzes the individual characteristics of the patient through the cloud platform (5), can automatically generate personalized monitoring frequency suggestions and nursing plan recommendations, and provides a remote medical consultation entrance to timely link professional doctors for intervention.
6. The diabetes foot ulcer warning system based on multimodal physiological signal fusion according to claim 1 or 5, characterized in that: It further includes a time series fusion algorithm system (6). The time series fusion algorithm system (6) sets warning thresholds and hierarchical response criteria based on a medical expert knowledge base, adopts a dynamic weight allocation strategy, adaptively adjusts the model's attention degree according to the importance of different physiological parameters, and introduces an attention mechanism to optimize the ability to capture long-term dependencies.
7. The diabetic foot ulcer warning system based on multimodal physiological signal fusion according to claim 5, characterized in that: The specific assessment steps of the personalized health assessment system (4) are as follows: S1. Collect and clean the basic information of diabetic foot patients; S2. Individual feature modeling: Use the random forest algorithm to identify the main factors affecting the ulcer risk; S3. Prediction model optimization: Adjust the weight coefficient of the model according to the individual parameters of diabetic foot patients to improve the prediction accuracy; S4. Real-time feedback and suggestion generation: Automatically generate personalized nursing plan prompts; S5. When the risk score reaches the warning threshold, automatically send a medical reminder.
8. A method for warning of diabetic foot ulcers based on multimodal physiological signal fusion, which uses the warning system for diabetic foot ulcers based on multimodal physiological signal fusion described in any one of claims 1-7, and the specific steps are as follows: S1. Extract the key features of the data collected by the multi-dimensional sensor component (1); S2. Preprocess the data collected by the sensors in the multi-dimensional sensor component (1), denoise and smooth the temperature, pressure, blood oxygen saturation and humidity signals respectively, and align the signals with different sampling frequencies to the same time reference through interpolation; S3. Construct a fusion model, use an LSTM network to capture the long-term dependencies in the time series, introduce a Self-attention mechanism, and dynamically allocate the attention weights of different features; establish a hierarchical warning scoring system, which is divided into three levels: low risk, medium risk and high risk; S4. Adjust the weight coefficients of the model according to the individual parameters of the patient to improve the prediction accuracy and automatically generate personalized nursing plan prompts; S5. When the predicted risk score reaches the warning threshold, automatically send a medical reminder.