SCRM diabetes multi-terminal interconnection data processing method

Abstract

The invention relates to the technical field of data processing, in particular to an SCRM diabetes multi-terminal interconnection data processing method. The method comprises the following steps: acquiring multi-dimensional health data of a user; performing multi-terminal interconnection according to the user multi-dimensional health data to obtain user multi-dimensional interconnection data, and performing interconnection feature extraction according to the user multi-dimensional interconnection data to obtain user multi-dimensional interconnection feature data; performing health index mapping according to the user multi-dimensional interconnection feature data to obtain health index mapping data, and performing channel right and interest processing according to the health index mapping data to obtain health index weighted data; mapping the multi-dimensional health data of the user according to the health index weighted data to obtain multi-dimensional health mapping data of the user, and performing blood glucose control recommendation according to the multi-dimensional health mapping data of the user to obtain blood glucose control recommendation data so as to perform SCRM diabetes multi-terminal interconnection data processing operation. The diabetes management effect and the life quality of a patient are improved.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a SCRM diabetes multi-terminal interconnected data processing method. Background Art

[0002] Diabetes is a serious chronic disease, and its incidence rate is increasing year by year with the change of lifestyle. Diabetic patients need to monitor their blood sugar levels for a long time and control blood sugar through diet, exercise and medication. However, traditional diabetes management methods have many problems, such as scattered data, single monitoring methods, insufficient management, etc., which cannot meet the needs of modern diabetes management.

[0003] With the rapid development of information technology and communication technology, health management systems based on information and communication technology (ICT) have been widely used in diabetes management. In particular, the introduction of customer relationship management systems (SCRM) has provided new ideas and methods for diabetes management. SCRM can integrate health data from different devices and platforms, conduct multi-dimensional analysis and personalized management, thereby improving the effectiveness of diabetes management and patient compliance. The existing system still has limitations in providing individual health management recommendations and cannot fully consider the differences and dynamic changes in the medical data collection channels of each patient. Summary of the invention

[0004] In order to solve the above-mentioned technical problems, the present invention proposes a SCRM diabetes multi-terminal interconnected data processing method to solve at least one of the above-mentioned technical problems.

[0005] The present application provides a SCRM diabetes multi-terminal interconnected data processing method, the method comprising:

[0006] S1. Obtain multi-dimensional health data of users;

[0007] S2. Perform multi-terminal interconnection based on the user's multi-dimensional health data to obtain the user's multi-dimensional interconnection data, and perform interconnection feature extraction based on the user's multi-dimensional interconnection data to obtain the user's multi-dimensional interconnection feature data;

[0008] S3, mapping the health index according to the multi-dimensional interconnected feature data of the user to obtain health index mapping data, and processing the channel rights and interests according to the health index mapping data to obtain health index weighted data;

[0009] S4. Map the user's multidimensional health data according to the health index weighted data to obtain the user's multidimensional health mapping data, and make blood sugar control recommendations based on the user's multidimensional health mapping data to obtain blood sugar control recommendation data, so as to perform SCRM diabetes multi-terminal interconnected data processing operations.

[0010] The method in the present invention covers the entire process from data collection, interconnection, analysis to recommendation, and realizes the comprehensive management of health data of diabetic patients. Through the integration and analysis of multi-dimensional data, personalized health advice is provided, which improves the accuracy and effectiveness of health management. By utilizing real-time data collection and feedback mechanisms, health management strategies are dynamically adjusted to ensure the timeliness and adaptability of management plans. Through scientific and accurate health indexes and personalized recommendations, patients' compliance with and implementation of health management recommendations are improved. The burden of daily health monitoring of patients on medical personnel is reduced, the allocation of medical resources is optimized, and the efficiency of medical services is improved.

[0011] Optionally, S1 includes:

[0012] Obtaining real-time blood glucose data from the user's blood glucose monitoring device;

[0013] Acquire daily activity data from the user's mobile device, where the daily activity data includes step data, exercise volume data, and heart rate data;

[0014] Obtaining dietary data from the user’s food-logging app;

[0015] Obtain basic user data from the user's electronic medical system;

[0016] Integrate real-time blood sugar data, daily activity data, diet data and user basic data into multi-dimensional health data of users.

[0017] The present invention ensures the timeliness and accuracy of data by acquiring blood sugar data and other health data in real time. The acquisition and integration of multi-dimensional data realizes all-round monitoring and comprehensive analysis of the health status of users.

[0018] Optionally, S2 includes:

[0019] Perform multi-terminal interconnection based on the user's multi-dimensional health data to obtain the user's multi-dimensional interconnection data;

[0020] Extracting features based on the health-related data to obtain health-related feature data;

[0021] Perform multi-dimensional encoding according to the health-related characteristic data to obtain health-related characteristic encoding data for sending to the cloud platform;

[0022] The control cloud platform performs multiple decoding based on the health-related feature coding data to obtain the user's multi-dimensional interconnected feature data.

[0023] In the present invention, multi-terminal interconnection technology is used to achieve data integration and real-time synchronization, ensure data consistency and timeliness, and improve the comprehensiveness and scientificity of health management. Accurately extract key health characteristics, support personalized health analysis and management, and improve the accuracy and practicality of data analysis. Through multi-dimensional coding technology, efficient data compression and secure transmission are achieved, data transmission efficiency and security are improved, and data privacy and security are ensured. Utilize the powerful computing power of the cloud platform to perform multiple decoding, accurately restore data, and support real-time data analysis and processing, thereby improving the timeliness and effectiveness of health management.

[0024] Optionally, applied to a cloud platform, S3 includes:

[0025] Mapping is performed based on the user's multi-dimensional interconnected feature data and the preset diabetes knowledge graph data to obtain health index mapping data;

[0026] Channel equity processing is performed based on the diabetes health index mapping data to obtain health index weighted data.

[0027] The present invention uses diabetes knowledge graph for data mapping, which improves the scientificity and reliability of health index generation. Through personalized health index mapping and weighted processing, accurate health assessment and management suggestions are provided. Channel equity processing improves the efficiency and effect of data fusion, ensures the high quality of the final health index, and prevents data changes brought by a single channel from having too much impact on the overall results. The weighted processing process is fair and transparent, which enhances the credibility of data processing and user trust.

[0028] Optionally, applied to a cloud platform, S4 includes:

[0029] Mapping is performed based on the user's multi-dimensional health data and the preset user health profile data to obtain preliminary user health profile mapping data;

[0030] Mapping the preliminary user health portrait mapping data according to the health index weighted data to obtain the user multi-dimensional health mapping data;

[0031] Blood sugar control recommendations are made based on the user's multi-dimensional health mapping data to obtain blood sugar control recommendation data for SCRM diabetes multi-terminal interconnected data processing operations.

[0032] The present invention combines the user's multi-dimensional health data and preset health portrait data to generate an accurate health portrait and improve the personalization of health assessment. Through the secondary mapping of health index weighted data, the data weight distribution is optimized to ensure the scientificity and credibility of the assessment results. The multi-dimensional health mapping data integrates the user's multi-faceted health information and provides a comprehensive health assessment. Using real-time data and feedback mechanisms, blood sugar control recommendations are dynamically adjusted to ensure the effectiveness and adaptability of the management plan.

[0033] Optionally, the multidimensional encoding includes:

[0034] Performing feature decomposition according to the health-related feature data to obtain health-related feature decomposition data;

[0035] Normalizing the health-related feature decomposition data to obtain health-related feature normalized data;

[0036] Perform self-attention feature embedding on the normalized data of health-related features to obtain health-related feature embedding data;

[0037] Quantitatively process the health-related feature embedded data to obtain health-related feature quantified data;

[0038] The health-related feature quantified data is feature encoded to obtain health-related feature encoded data to be sent to the cloud platform.

[0039] In the present invention, feature decomposition and normalization are used to simplify the data structure and improve data processing efficiency. Self-attention feature embedding and quantization are used to generate high-quality feature representation, which improves the effect of data analysis and model training. Quantization and encoding processes reduce the cost of data storage and transmission and improve the efficiency and security of data transmission. Feature encoding protects data privacy and ensures the security of data during transmission. High-quality feature representation improves the model performance of downstream tasks (such as blood sugar control recommendations).

[0040] Optionally, the self-attention feature embedding includes:

[0041] Perform channel coding and position coding on the normalized health-related feature data to obtain feature channel coding data and feature position coding data;

[0042] Perform masked multi-head self-attention calculation on the feature position encoding data to obtain feature masked multi-head self-attention data;

[0043] Perform multi-head self-attention calculation on the feature channel encoding data to obtain feature multi-head attention data;

[0044] Perform layer normalization on the feature multi-head attention data to obtain feature layer normalized data;

[0045] Perform full connection layer processing on the normalized data of the feature layer to obtain feature full connection layer data;

[0046] Perform activation calculation on the feature fully connected layer data to obtain the feature feedforward neural network feature data;

[0047] According to the feature mask multi-head self-attention data and the feature feedforward neural network feature data, residual connection and layer normalization are performed to obtain the health-related feature embedding data.

[0048] The present invention realizes unified representation and processing of multi-source data through channel coding and position coding, and improves the efficiency of data integration. The multi-head self-attention mechanism can capture the global dependencies in the data and improve the accuracy and richness of feature representation. Technologies such as layer normalization and residual connection improve the training stability and performance of the model and reduce overfitting. Through multi-level processing and activation calculation, the resolution and expression ability of the features are enhanced, and the performance of the model in downstream tasks is improved. Through multi-level data processing and embedding, the security of data transmission and storage is enhanced.

[0049] Optionally, multiple decoding includes:

[0050] The control cloud platform performs multi-level feature decoding based on the health-related feature embedded data to obtain feature decoding data;

[0051] Performing inverse vector quantization processing on the characteristic decoded data to obtain characteristic inverse quantized data;

[0052] The feature reverse quantization data is de-embedded to obtain the user's multi-dimensional interconnected feature data.

[0053] The multi-level feature decoding and reverse vector quantization in the present invention ensure high accuracy and integrity of data decoding. The multi-level decoding process can handle redundancy and noise in the data, and improve the robustness and reliability of decoding. The computing power of the cloud platform is used to achieve rapid decoding and processing of large-scale data and improve efficiency. Reverse vector quantization and feature de-embedding ensure fine recovery and consistency of data and improve data quality. The de-embedded data is more interpretable, which is convenient for subsequent analysis and application, and improves the practical value of the data.

[0054] Optionally, multi-level feature decoding includes:

[0055] The control cloud platform performs Huffman decoding based on the health-related feature embedded data to obtain primary decoding data;

[0056] Performing entropy coding and decoding on the primary decoded data to obtain secondary decoded data;

[0057] performing error correction on the secondary decoded data and restoring it to obtain characteristic error-corrected data;

[0058] The characteristic error correction data is subjected to noise reduction and automatic encoding to obtain characteristic decoding data.

[0059] In the present invention, Huffman coding and entropy coding are used to achieve efficient data compression and decoding, thereby improving data processing efficiency and accuracy. Errors in the data transmission process are detected and corrected through error correction coding technology to ensure data reliability and integrity. A noise reduction autoencoder is used to remove data noise, improve the quality and accuracy of data restoration, and ensure high data quality. Multi-level decoding and noise reduction processing are used to enhance the robustness of the system when processing complex and noisy data, thereby improving the stability and reliability of the system.

[0060] Optionally, the present application further provides a SCRM diabetes multi-terminal interconnected data processing system, which is used to execute the SCRM diabetes multi-terminal interconnected data processing method as described above, and the SCRM diabetes multi-terminal interconnected data processing system includes:

[0061] User multi-dimensional health data collection module, used to obtain user multi-dimensional health data;

[0062] A user multi-dimensional interconnection feature extraction module is used to perform multi-terminal interconnection according to the user's multi-dimensional health data to obtain the user's multi-dimensional interconnection data, and to perform interconnection feature extraction according to the user's multi-dimensional interconnection data to obtain the user's multi-dimensional interconnection feature data;

[0063] A health index weighted calculation module is used to perform health index mapping according to the multi-dimensional interconnected characteristic data of the user to obtain health index mapping data, and to perform channel equity processing according to the health index mapping data to obtain health index weighted data;

[0064] The SCRM diabetes multi-terminal interconnected data processing module is used to map the user's multi-dimensional health data according to the health index weighted data to obtain the user's multi-dimensional health mapping data, and to make blood sugar control recommendations based on the user's multi-dimensional health mapping data to obtain blood sugar control recommendation data for SCRM diabetes multi-terminal interconnected data processing operations.

[0065] The purpose of the present invention is to collect and integrate multi-dimensional health data of users, monitor the health status of users in real time, and provide timely health management and intervention. Through standardized processing and feature extraction, the efficiency and accuracy of data processing are improved, providing a high-quality data foundation for subsequent analysis and management. By using the self-attention mechanism and health index mapping, intelligent feature extraction and personalized health assessment are realized, and accurate health management suggestions are provided. Through quantification and coding processing, the security and efficiency of data transmission and storage are improved, and the integrity and security of data are ensured. Through health index mapping and channel rights processing, the scientificity and reliability of the data processing process are ensured, and a reliable health management plan is provided. Using a dynamic adjustment mechanism, the health management plan is continuously optimized based on user feedback and real-time data to ensure the effectiveness and adaptability of the plan. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] Other features, objects and advantages of the present application will become more apparent by reading the detailed description of non-limiting implementations made with reference to the following drawings:

[0067] Figure 1 A flowchart showing a method for processing SCRM diabetes multi-terminal interconnected data according to an embodiment is shown;

[0068] Figure 2 A flowchart showing a method for collecting multi-dimensional health data of a user according to an embodiment of the present invention is shown;

[0069] Figure 3 A flowchart showing a method for extracting multi-dimensional interconnection features of users according to an embodiment of the present invention is shown;

[0070] Figure 4 A flowchart showing a method for recommending multi-dimensional health and blood sugar control for a user according to an embodiment of the present invention is shown;

[0071] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0072] The following is a clear and complete description of the technical method of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by technicians in this field without creative work are within the scope of protection of the present invention.

[0073] In addition, the accompanying drawings are only schematic illustrations of the present invention and are not necessarily drawn to scale. The same reference numerals in the figures represent the same or similar parts, and their repeated description will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities and do not necessarily correspond to physically or logically independent entities. The functional entities can be implemented in software form, or implemented in one or more hardware modules or integrated circuits, or implemented in different networks and / or processor methods and / or microcontroller methods.

[0074] It should be understood that, although the terms "first", "second", etc. may be used herein to describe various units, these units should not be limited by these terms. These terms are used only to distinguish one unit from another unit. For example, without departing from the scope of the exemplary embodiments, the first unit may be referred to as the second unit, and similarly the second unit may be referred to as the first unit. The term "and / or" used herein includes any and all combinations of one or more of the listed associated items.

[0075] See also Figures 1 to 4 The present application provides a SCRM diabetes multi-terminal interconnected data processing method, the method comprising:

[0076] S1. Obtain multi-dimensional health data of users;

[0077] Specifically, the Internet of Things (IoT) technology is used to synchronize the health monitoring devices worn by patients (such as blood glucose meters and smart bracelets) with the central database.

[0078] The patient's blood sugar, heart rate and other health data are collected in real time through wireless sensor networks (WSN).

[0079] Import patients’ historical health data from electronic health record (EHR) systems.

[0080] The patient's movement and activity data are collected through GPS and accelerometer sensors.

[0081] All the above data are aggregated and stored in the edge computing center to obtain the user's multi-dimensional health data.

[0082] S2. Perform multi-terminal interconnection based on the user's multi-dimensional health data to obtain the user's multi-dimensional interconnection data, and perform interconnection feature extraction based on the user's multi-dimensional interconnection data to obtain the user's multi-dimensional interconnection feature data;

[0083] Specifically, multimodal data fusion technology is used to fuse different types of health data to form multi-dimensional interconnected data.

[0084] Use association rule mining technology to associate different types of health data and generate associated features. Use feature engineering technology to extract key health feature data, such as average blood sugar value, exercise volume, etc., to obtain multi-dimensional interconnected feature data of users.

[0085] S3, mapping the health index according to the multi-dimensional interconnected feature data of the user to obtain health index mapping data, and processing the channel rights and interests according to the health index mapping data to obtain health index weighted data;

[0086] Specifically, health index mapping is performed through preset diabetes knowledge graph data and user multi-dimensional interconnected feature data to obtain health index mapping data; the health index is mapped to a specific health status through a mapping algorithm to obtain health mapping data.

[0087] Steps for building the preset diabetes knowledge graph data: Collect medical knowledge about diabetes, including pathology, physiology, diagnostic criteria, treatment options, risk factors, etc. Use a graph database (such as Neo4j) to build a diabetes knowledge graph and define nodes (such as symptoms, drugs, treatment options) and edges (such as causal relationships, association relationships). Each node and edge contains relevant attributes and weights, reflecting its importance in diabetes management.

[0088] The user's multi-dimensional interconnected feature data is mapped to the nodes and edges of the diabetes knowledge graph. According to the weights of the nodes and edges in the knowledge graph and the values ​​of the user's feature data, a weighted calculation is performed to calculate the health index and obtain the output health index mapping data.

[0089] The threshold range of health status is set according to medical standards (such as healthy, sub-healthy, high risk, etc.). The health index is mapped to a specific health status using a mapping algorithm.

[0090] Knowledge graphs such as: Nodes: blood sugar level (attributes: normal, high, low), exercise frequency (attributes: high, medium, low), eating habits (attributes: healthy, normal, unhealthy), symptoms (attributes: whether there are complications). Edges:

[0091] User multi-dimensional interconnected characteristic data: average blood sugar level: 160mg / dL, number of steps per day: 4000 steps, exercise frequency per week: 2 times, eating habits: high intake of sugary drinks.

[0092] Map the user's feature data to the knowledge graph. Blood sugar level node weight: 0.4, exercise frequency node weight: 0.3, eating habits node weight: 0.2, symptom node weight: 0.1.

[0093] Health index calculation: blood sugar level (high, value 160): 160×0.4=64, exercise frequency (low, value 4000): 0.2×0.3=0.06 (since the number of steps did not reach the standard, it was scored as 0.2), eating habits (unhealthy): 0.2×0.2=0.04, symptoms (with complications): 1×0.1=0.1 (assuming the complications are serious, score as 1), total health index: 64+0.06+0.04+0.1=64.2.

[0094] Health status thresholds: Healthy: health index <40, Sub-healthy: 40≤ health index <60, High risk: health index ≥60. A health index of 64.2 is mapped to a high-risk state.

[0095] Evaluate the credibility and weight of different health data sources, perform channel equity processing, and obtain channel equity data. Based on the channel equity data, perform weighted calculation on the health index to obtain the weighted data of the health index.

[0096] Specifically, list all sources of health data, including but not limited to: blood glucose monitoring devices, mobile devices (such as smart watches, mobile phones), diet record applications, and electronic medical systems. Quantify each indicator of each data source and score it on a scale of 0-1 (or 0-100), for example: data completeness: 0.9, data accuracy: 0.8, data timeliness: 0.7, data stability: 0.85.

[0097] According to the weight of each indicator (which can be set according to actual conditions, such as 0.25), the comprehensive credibility score of each data source is calculated.

[0098] Blood glucose monitoring equipment: data integrity: 0.9, data accuracy: 0.85, data timeliness: 0.8, data stability: 0.9, comprehensive credibility: 0.25×0.9+0.25×0.85+0.25×0.8+0.25×0.9=0.8625.

[0099] Mobile device: Data integrity: 0.8, Data accuracy: 0.75, Data timeliness: 0.7, Data stability: 0.8. Overall credibility: 0.25×0.8+0.25×0.75+0.25×0.7+0.25×0.8=0.7625.

[0100] Diet record application: Data completeness: 0.7, Data accuracy: 0.7, Data timeliness: 0.6, Data stability: 0.7. Overall credibility: 0.25×0.7+0.25×0.7+0.25×0.6+0.25×0.7=0.675.

[0101] Electronic medical system: Data integrity: 0.95, Data accuracy: 0.9, Data timeliness: 0.85, Data stability: 0.95. Overall credibility: 0.25×0.95+0.25×0.9+0.25×0.85+0.25×0.95=0.9125.

[0102] The composite credibility scores of each data source were normalized so that their sum was 1. Blood glucose monitoring devices: weight 0.287, mobile devices: weight 0.254, diet logging apps: weight 0.225, and electronic health systems: weight 0.304.

[0103] The health index of each data source is calculated based on their respective data. The health index of blood glucose monitoring devices is 65, the health index of mobile devices is 60, the health index of diet recording applications is 55, and the health index of electronic medical systems is 70. Through the above weight calculation, the weighted health index is 67.55.

[0104] S4. Map the user's multidimensional health data according to the health index weighted data to obtain the user's multidimensional health mapping data, and make blood sugar control recommendations based on the user's multidimensional health mapping data to obtain blood sugar control recommendation data, so as to perform SCRM diabetes multi-terminal interconnected data processing operations.

[0105] Specifically, the health index weighted data is mapped to specific health data points through a multidimensional mapping algorithm. The patient's blood sugar fluctuations are analyzed based on the health prediction data through a deep learning algorithm (such as a regression algorithm). A recommendation system algorithm is used to generate personalized blood sugar control recommendations for patients based on their blood sugar fluctuations.

[0106] The method in the present invention covers the entire process from data collection, interconnection, analysis to recommendation, and realizes the comprehensive management of health data of diabetic patients. Through the integration and analysis of multi-dimensional data, personalized health advice is provided, which improves the accuracy and effectiveness of health management. By utilizing real-time data collection and feedback mechanisms, health management strategies are dynamically adjusted to ensure the timeliness and adaptability of management plans. Through scientific and accurate health indexes and personalized recommendations, patients' compliance with and implementation of health management recommendations are improved. The burden of daily health monitoring of patients on medical personnel is reduced, the allocation of medical resources is optimized, and the efficiency of medical services is improved.

[0107] Optionally, S1 includes:

[0108] S11, obtaining real-time blood sugar data from the user's blood sugar monitoring device;

[0109] Specifically, the user's blood glucose monitoring device (such as Dexcom, FreeStyle Libre, etc.) is connected via Bluetooth or Wi-Fi. The blood glucose monitoring device automatically collects blood glucose data every 5 minutes. The collected data is transmitted to the user's mobile phone application or edge computing center in real time through a secure protocol (such as HTTPS).

[0110] S12, obtaining daily activity data from the user's mobile device, wherein the daily activity data includes step data, exercise volume data, and heart rate data;

[0111] Specifically, the user's mobile device (such as a smart watch, mobile phone application) is integrated through an API interface. The mobile device collects the user's step count, exercise volume, and heart rate data in real time through sensors (such as accelerometers and heart rate monitors). The mobile device periodically (such as every hour) synchronizes the collected data to the mobile phone application or edge computing center.

[0112] S13, obtaining dietary data from a user's dietary record application;

[0113] Specifically, the user's diet record application (such as MyFitnessPal, Lifesum, etc.) is integrated through the API interface. The user enters the daily diet record in the application, including food type, intake and time. The user's diet data is extracted regularly (such as daily) through the API interface or the edge computing center.

[0114] S14, obtaining basic user data from the user's electronic medical system;

[0115] Specifically, the user's electronic medical system (such as the hospital's EMR system, health management platform) is connected through the API interface. The user authorizes the system to obtain his or her basic health data, including medical history, allergy history, medication records, etc. The user's basic health data is extracted regularly (such as weekly) through the API interface.

[0116] S15. Integrate real-time blood sugar data, daily activity data, diet data and user basic data into multi-dimensional health data of the user.

[0117] Specifically, after the edge center receives real-time blood sugar data, daily activity data, diet data, and basic health data, it stores them in a distributed database.

[0118] The present invention ensures the timeliness and accuracy of data by acquiring blood sugar data and other health data in real time. The acquisition and integration of multi-dimensional data realizes all-round monitoring and comprehensive analysis of the health status of users.

[0119] Optionally, S2 includes:

[0120] S21, performing multi-terminal interconnection according to the multi-dimensional health data of the user to obtain the multi-dimensional interconnection data of the user;

[0121] Specifically, various health monitoring devices of users, such as blood glucose monitors and smart watches, are connected through API interfaces or data protocols. A data synchronization mechanism is set up to ensure that the data of all devices can be synchronized to the edge computing center in real time. The data integration platform is used to convert and integrate the data from different devices to form unified multi-dimensional interconnected data of users.

[0122] S22, extracting features based on the health-related data to obtain health-related feature data;

[0123] Specifically, through correlation analysis (such as Pearson correlation coefficient), characteristic variables with high correlation with health are selected, such as blood sugar fluctuations, exercise frequency, etc. Dimensionality reduction techniques such as principal component analysis (PCA) are used to reduce high-dimensional data to lower dimensions and extract main features.

[0124] S23, performing multi-dimensional encoding according to the health-related characteristic data to obtain health-related characteristic encoding data, and sending the health-related characteristic encoding data to the cloud platform;

[0125] Specifically, use PCA and other algorithms to decompose the health-related feature data to obtain feature decomposition data. The decomposed feature data is normalized to standardize the data to a uniform range. Apply a self-attention mechanism (such as the Transformer model) to embed features of the normalized data to capture the global dependencies between features. Quantize the embedded data to reduce the data size for easy transmission and storage. Use a coding algorithm (such as Huffman coding) to encode the quantized data to generate health-related feature encoded data. Send the encoded data to the cloud platform via a secure transmission protocol (such as HTTPS).

[0126] S24. The control cloud platform performs multiple decoding based on the health-related feature coding data to obtain the user's multi-dimensional interconnected feature data.

[0127] Specifically, multiple decoding is performed on the cloud platform to restore the original feature data. The received encoded data is decoded using the Huffman decoding algorithm to obtain preliminary decoded data. The decoded quantized data is restored to continuous feature data using the inverse vector quantization technology. The anti-self-attention mechanism is applied to restore the quantized feature data to the original high-dimensional feature space. The restored feature data is integrated to generate the user's multi-dimensional interconnected feature data for subsequent analysis and use.

[0128] In the present invention, multi-terminal interconnection technology is used to achieve data integration and real-time synchronization, ensure data consistency and timeliness, and improve the comprehensiveness and scientificity of health management. Accurately extract key health characteristics, support personalized health analysis and management, and improve the accuracy and practicality of data analysis. Through multi-dimensional coding technology, efficient data compression and secure transmission are achieved, data transmission efficiency and security are improved, and data privacy and security are ensured. Utilize the powerful computing power of the cloud platform to perform multiple decoding, accurately restore data, and support real-time data analysis and processing, thereby improving the timeliness and effectiveness of health management.

[0129] Optionally, applied to a cloud platform, S3 includes:

[0130] Mapping is performed based on the user's multi-dimensional interconnected feature data and the preset diabetes knowledge graph data to obtain health index mapping data;

[0131] Specifically, collect and organize medical knowledge related to diabetes, including symptoms, treatment methods, risk factors, etc. Use graph databases (such as Neo4j) to build a diabetes knowledge graph and define the attributes and relationships of nodes and edges.

[0132] Connect the user's multi-dimensional interconnected feature data with the nodes in the knowledge graph. Match the user data with the corresponding nodes in the knowledge graph through a similarity calculation algorithm (such as cosine similarity). Calculate the weights of the relevant nodes and edges in the graph based on the user feature data. Use algorithms such as graph convolutional networks (GCN) to perform convolution operations on the knowledge graph, comprehensively consider the weights of nodes and edges, and calculate the health index.

[0133] Channel equity processing is performed based on the diabetes health index mapping data to obtain health index weighted data.

[0134] Specifically, the source channel (such as device, application, manual input, etc.) of each health index mapping data is marked, and different weights are preset according to the credibility and importance of the data source.

[0135] The quality of each data source is evaluated using evaluation indicators (such as data completeness, accuracy, and timeliness). Based on the evaluation results, a multi-criteria decision-making method such as the analytic hierarchy process (AHP) is used to assign a weight to each data source.

[0136] Each health index mapping data is multiplied by its corresponding weight to obtain weighted health index data. The weighted data are integrated to generate health index weighted data.

[0137] The present invention uses diabetes knowledge graph for data mapping, which improves the scientificity and reliability of health index generation. Through personalized health index mapping and weighted processing, accurate health assessment and management suggestions are provided. Channel equity processing improves the efficiency and effect of data fusion, ensures the high quality of the final health index, and prevents data changes brought by a single channel from having too much impact on the overall results. The weighted processing process is fair and transparent, which enhances the credibility of data processing and user trust.

[0138] Optionally, applied to a cloud platform, S4 includes:

[0139] S41, mapping is performed according to the user's multi-dimensional health data and the preset user health portrait data to obtain preliminary user health portrait mapping data;

[0140] Specifically, the user's basic health data, historical medical records, living habits, etc. are collected to build a preliminary health profile. The health profile data is stored in a graph database for quick query and update.

[0141] Connect the user's multi-dimensional health data with the preset health profile data. Use a similarity calculation algorithm (such as cosine similarity or Euclidean distance) to calculate the similarity between the user's current data and the health profile. Update the feature values ​​in the health profile based on the similarity calculation results. Generate preliminary user health profile mapping data to reflect the user's current condition.

[0142] S42, mapping the preliminary user health portrait mapping data according to the health index weighted data to obtain the user multi-dimensional health mapping data;

[0143] Specifically, the health index weighted data is applied to the preliminary user health portrait mapping data, and a weighted value of each feature is calculated using a weighting algorithm (such as linear weighting or weighted average).

[0144] The weighted preliminary user health portrait mapping data is fused with the health index weighted data. Data fusion algorithms (such as weighted fusion or Bayesian fusion) are used to perform feature fusion to generate comprehensive user multidimensional health mapping data. Generate user multidimensional health mapping data to provide a more comprehensive health status analysis.

[0145] S43. Make blood sugar control recommendations based on the user's multi-dimensional health mapping data to obtain blood sugar control recommendation data for SCRM diabetes multi-terminal interconnected data processing operations.

[0146] Specifically, a rule-based expert system is constructed to define the recommended rules and strategies for blood sugar control. Based on the prediction results and expert system rules, personalized blood sugar control recommendations are generated. Blood sugar control recommendation data is generated and fed back to the user for SCRM diabetes multi-terminal interconnected data processing operations.

[0147] Gather expertise in the field of diabetes management, including medical literature, clinical guidelines, expert opinions, etc. Based on the collected knowledge, define a series of rules and strategies for blood sugar control. Use open source rule engines such as Drools to design and implement the workflow of the rule engine. Encode the defined rules in a rule engine language (such as the Drools rule language). Perform forward reasoning based on the user's current health data and rule base to derive blood sugar control recommendations. Based on user data, trigger the corresponding rules to generate personalized recommendations.

[0148] The present invention combines the user's multi-dimensional health data and preset health portrait data to generate an accurate health portrait and improve the personalization of health assessment. Through the secondary mapping of health index weighted data, the data weight distribution is optimized to ensure the scientificity and credibility of the assessment results. The multi-dimensional health mapping data integrates the user's multi-faceted health information and provides a comprehensive health assessment. Using real-time data and feedback mechanisms, blood sugar control recommendations are dynamically adjusted to ensure the effectiveness and adaptability of the management plan.

[0149] Optionally, the multidimensional encoding includes:

[0150] Performing feature decomposition according to the health-related feature data to obtain health-related feature decomposition data;

[0151] Specifically, principal component analysis (PCA) or factor analysis (FA) is used for feature decomposition. The health-related feature data is prepared in matrix form, where each row is a sample and each column is a feature. The covariance matrix of the data matrix is ​​calculated to represent the linear relationship between the features. The covariance matrix is ​​subjected to eigenvalue decomposition to obtain eigenvalues ​​and eigenvectors. The first k eigenvectors with larger eigenvalues ​​are selected as principal components to form a feature representation after dimensionality reduction. The original data is projected onto the selected principal components to generate health-related feature decomposition data.

[0152] Normalizing the health-related feature decomposition data to obtain health-related feature normalized data;

[0153] Specifically, the mean and standard deviation of each feature are calculated, and the data of each feature is standardized to a distribution with a mean of 0 and a standard deviation of 1. The data of each feature is normalized to the interval [0,1]. The normalized data is used as the normalized data of health-related features.

[0154] Perform self-attention feature embedding on the normalized data of health-related features to obtain health-related feature embedding data;

[0155] Specifically, the Transformer model is used for self-attention feature embedding. Position encoding is added to the normalized data to retain sequence information. The query, key, and value matrices of the input data are calculated. The attention weights are calculated, normalized using the Softmax function, and the weighted sum value matrix is ​​calculated. The attention representation is calculated in parallel using multiple independent attention heads, and the results are concatenated. The output of the multi-head attention is processed through a fully connected layer and an activation function to generate an embedded feature. The data processed by the self-attention embedding is embedded into the data as a health-related feature.

[0156] Quantitatively process the health-related feature embedded data to obtain health-related feature quantified data;

[0157] Specifically, the embedded feature data is converted into discrete value representation and quantized through a lookup table. The nearest quantization center is found for each feature vector. The quantization center is pre-calculated through a clustering algorithm (such as K-means). The quantized feature data is used as health-related feature quantization data.

[0158] The health-related feature quantified data is feature encoded to obtain health-related feature encoded data to be sent to the cloud platform.

[0159] Specifically, the frequency of occurrence of each value in the quantitative data is counted to construct a frequency table. A Huffman tree is constructed based on the frequency table, where each node represents a feature value and the path of the tree represents the code. A coding table is generated based on the Huffman tree, and a unique binary code is assigned to each feature value. The quantitative data is converted into Huffman codes to generate health-related feature coding data. The coded data is sent to the cloud platform via a secure transmission protocol (such as HTTPS).

[0160] In the present invention, feature decomposition and normalization are used to simplify the data structure and improve data processing efficiency. Self-attention feature embedding and quantization are used to generate high-quality feature representation, which improves the effect of data analysis and model training. Quantization and encoding processes reduce the cost of data storage and transmission and improve the efficiency and security of data transmission. Feature encoding protects data privacy and ensures the security of data during transmission. High-quality feature representation improves the model performance of downstream tasks (such as blood sugar control recommendations).

[0161] Optionally, the self-attention feature embedding includes:

[0162] Perform channel coding and position coding on the normalized health-related feature data to obtain feature channel coding data and feature position coding data;

[0163] Specifically, one-hot encoding is used for channel encoding, and position encoding is used for position encoding. One-hot encoding is performed on the data source channels, and each channel is represented as a vector.

[0164] Generate position encoding according to the position of the feature in the sequence and combine the position information with the normalized data.

[0165] Perform masked multi-head self-attention calculation on the feature position encoding data to obtain feature masked multi-head self-attention data;

[0166] Specifically, the query, key, and value matrices of the input data are calculated. The masked multi-head self-attention weights are calculated, and the invalid positions are masked using the mask matrix M, and the weighted sum value matrix is ​​performed. Multiple independent attention heads are calculated in parallel, and the results are concatenated to obtain the feature masked multi-head self-attention data.

[0167] Perform multi-head self-attention calculation on the feature channel encoding data to obtain feature multi-head attention data;

[0168] Specifically, the query, key, and value matrices of the input data are calculated. The multi-head self-attention weights are calculated and the weighted sum value matrix is ​​performed. Multiple independent attention heads are calculated in parallel, and the results are concatenated to obtain the feature multi-head attention data.

[0169] Perform layer normalization on the feature multi-head attention data to obtain feature layer normalized data;

[0170] Specifically, the multi-head attention data is normalized, and the mean and standard deviation are calculated for normalization.

[0171] Perform full connection layer processing on the normalized data of the feature layer to obtain feature full connection layer data;

[0172] Specifically, the layer normalized data is input into the fully connected layer for linear transformation.

[0173] Perform activation calculation on the feature fully connected layer data to obtain the feature feedforward neural network feature data;

[0174] Specifically, an activation function is applied to the fully connected layer data to perform a nonlinear transformation.

[0175] According to the feature mask multi-head self-attention data and the feature feedforward neural network feature data, residual connection and layer normalization are performed to obtain the health-related feature embedding data.

[0176] Specifically, the masked multi-head self-attention data is residually connected with the feedforward neural network feature data. The data after residual connection is layer normalized, the mean and standard deviation are calculated, and normalized to obtain the health-related feature embedding data.

[0177] The present invention realizes unified representation and processing of multi-source data through channel coding and position coding, and improves the efficiency of data integration. The multi-head self-attention mechanism can capture the global dependencies in the data and improve the accuracy and richness of feature representation. Technologies such as layer normalization and residual connection improve the training stability and performance of the model and reduce overfitting. Through multi-level processing and activation calculation, the resolution and expression ability of the features are enhanced, and the performance of the model in downstream tasks is improved. Through multi-level data processing and embedding, the security of data transmission and storage is enhanced.

[0178] Optionally, multiple decoding includes:

[0179] The control cloud platform performs multi-level feature decoding based on the health-related feature embedded data to obtain feature decoding data;

[0180] Specifically, the cloud platform receives health-related feature embedded data. By traversing the Huffman tree, the code is converted back to the original data. The binary code string is read and decoded according to the Huffman tree to obtain primary decoded data. The primary decoded data is entropy encoded and decoded to restore the secondary decoded data. Error correction is performed using Reed-Solomon encoding or other ECC methods. Error detection and correction are performed on the secondary decoded data to generate error-corrected feature decoded data.

[0181] Performing inverse vector quantization processing on the characteristic decoded data to obtain characteristic inverse quantized data;

[0182] Specifically, the quantization index is converted back to the original vector using the pre-stored quantization center table, the quantization index in the feature decoded data is mapped back to the corresponding original feature vector, and the data after the inverse quantization processing is used as the feature inverse quantization data.

[0183] The feature reverse quantization data is de-embedded to obtain the user's multi-dimensional interconnected feature data.

[0184] Specifically, the query, key, and value matrices of the reverse self-attention mechanism are calculated. The reverse attention weights are calculated and the weighted sum value matrix is ​​performed. The embedded features are restored to the original features through the reverse multi-head self-attention mechanism. The reverse quantized data is restored to a high-dimensional feature vector using the reverse self-attention mechanism. The reverse embedded data is used as the user's multi-dimensional interconnected feature data.

[0185] The multi-level feature decoding and reverse vector quantization in the present invention ensure high accuracy and integrity of data decoding. The multi-level decoding process can handle redundancy and noise in the data, and improve the robustness and reliability of decoding. The computing power of the cloud platform is used to achieve rapid decoding and processing of large-scale data and improve efficiency. Reverse vector quantization and feature de-embedding ensure fine recovery and consistency of data and improve data quality. The de-embedded data is more interpretable, which is convenient for subsequent analysis and application, and improves the practical value of the data.

[0186] Optionally, multi-level feature decoding includes:

[0187] The control cloud platform performs Huffman decoding based on the health-related feature embedded data to obtain primary decoding data;

[0188] Specifically, the binary string embedded in the health-related feature data is read, traversed and matched according to the Huffman tree, and the binary string is converted back to the original data.

[0189] Performing entropy coding and decoding on the primary decoded data to obtain secondary decoded data;

[0190] Specifically, an entropy coding and decoding algorithm is used to decode the primary decoding data, calculate the probability distribution of the data, and map the encoded value back to the original value.

[0191] performing error correction on the secondary decoded data and restoring it to obtain characteristic error-corrected data;

[0192] Specifically, Reed-Solomon coding is used for error detection and correction. Error data is detected and corrected according to coding rules.

[0193] The characteristic error correction data is subjected to noise reduction and automatic encoding to obtain characteristic decoding data.

[0194] Specifically, a denoising autoencoder model is trained, including an encoder and a decoder part. The model is trained using noisy data so that it can restore the original data without noise. The feature error correction data is input into the denoising autoencoder. The encoder part maps the input data to the hidden layer representation. The decoder part maps the hidden layer representation back to the original data space, removes the noise, and obtains the feature decoded data.

[0195] In the present invention, Huffman coding and entropy coding are used to achieve efficient data compression and decoding, thereby improving data processing efficiency and accuracy. Errors in the data transmission process are detected and corrected through error correction coding technology to ensure data reliability and integrity. A noise reduction autoencoder is used to remove data noise, improve the quality and accuracy of data restoration, and ensure high data quality. Multi-level decoding and noise reduction processing are used to enhance the robustness of the system when processing complex and noisy data, thereby improving the stability and reliability of the system.

[0196] Optionally, the present application further provides a SCRM diabetes multi-terminal interconnected data processing system, which is used to execute the SCRM diabetes multi-terminal interconnected data processing method as described above, and the SCRM diabetes multi-terminal interconnected data processing system includes:

[0197] User multi-dimensional health data collection module, used to obtain user multi-dimensional health data;

[0198] A user multi-dimensional interconnection feature extraction module is used to perform multi-terminal interconnection according to the user's multi-dimensional health data to obtain the user's multi-dimensional interconnection data, and to perform interconnection feature extraction according to the user's multi-dimensional interconnection data to obtain the user's multi-dimensional interconnection feature data;

[0199] A health index weighted calculation module is used to perform health index mapping according to the multi-dimensional interconnected characteristic data of the user to obtain health index mapping data, and to perform channel equity processing according to the health index mapping data to obtain health index weighted data;

[0200] The SCRM diabetes multi-terminal interconnected data processing module is used to map the user's multi-dimensional health data according to the health index weighted data to obtain the user's multi-dimensional health mapping data, and to make blood sugar control recommendations based on the user's multi-dimensional health mapping data to obtain blood sugar control recommendation data for SCRM diabetes multi-terminal interconnected data processing operations.

[0201] The purpose of the present invention is to collect and integrate multi-dimensional health data of users, monitor the health status of users in real time, and provide timely health management and intervention. Through standardized processing and feature extraction, the efficiency and accuracy of data processing are improved, providing a high-quality data foundation for subsequent analysis and management. By using the self-attention mechanism and health index mapping, intelligent feature extraction and personalized health assessment are realized, and accurate health management suggestions are provided. Through quantification and coding processing, the security and efficiency of data transmission and storage are improved, and the integrity and security of data are ensured. Through health index mapping and channel rights processing, the scientificity and reliability of the data processing process are ensured, and a reliable health management plan is provided. Using a dynamic adjustment mechanism, the health management plan is continuously optimized based on user feedback and real-time data to ensure the effectiveness and adaptability of the plan.

[0202] Therefore, from any point of view, the embodiments should be regarded as illustrative and non-restrictive, and the scope of the present invention is limited by the attached application documents rather than the above description, and it is intended that all changes falling within the meaning and scope of equivalent elements of the application documents are included in the present invention.

[0203] The above description is only a specific embodiment of the present invention, so that those skilled in the art can understand or implement the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but should conform to the widest scope consistent with the principles and novel features invented herein.

Claims

1. A SCRM diabetes multi-terminal interconnected data processing method, It is characterized in that The method comprises: S1. Obtain multi-dimensional health data of users; S2. Perform multi-terminal interconnection based on the user's multi-dimensional health data to obtain the user's multi-dimensional interconnection data, and perform interconnection feature extraction based on the user's multi-dimensional interconnection data to obtain the user's multi-dimensional interconnection feature data; S3, mapping the health index according to the multi-dimensional interconnected characteristic data of the user to obtain health index mapping data, and processing the channel rights and interests according to the health index mapping data to obtain health index weighted data; S4. Map the user's multidimensional health data according to the health index weighted data to obtain the user's multidimensional health mapping data, and make blood sugar control recommendations based on the user's multidimensional health mapping data to obtain blood sugar control recommendation data, so as to perform SCRM diabetes multi-terminal interconnected data processing operations.

2. The method according to claim 1, It is characterized in that S1 includes: Obtaining real-time blood glucose data from the user's blood glucose monitoring device; Acquire daily activity data from the user's mobile device, where the daily activity data includes step data, exercise volume data, and heart rate data; Obtaining dietary data from the user’s food-logging app; Obtain basic user data from the user's electronic medical system; Integrate real-time blood sugar data, daily activity data, diet data and user basic data into multi-dimensional health data of users.

3. The method according to claim 1, It is characterized in that S2 include: Perform multi-terminal interconnection based on the user's multi-dimensional health data to obtain the user's multi-dimensional interconnection data; Extracting features based on the health-related data to obtain health-related feature data; Perform multi-dimensional encoding according to the health-related characteristic data to obtain health-related characteristic encoding data for sending to the cloud platform; The control cloud platform performs multiple decoding based on the health-related feature coding data to obtain the user's multi-dimensional interconnected feature data.

4. The method according to claim 1, It is characterized in that Applied to cloud platforms, S3 includes: Mapping is performed based on the user's multi-dimensional interconnected feature data and the preset diabetes knowledge graph data to obtain health index mapping data; Channel equity processing is performed based on the diabetes health index mapping data to obtain health index weighted data.

5. The method according to claim 1, It is characterized in that Applied to the cloud platform, S4 includes: Mapping is performed based on the user's multi-dimensional health data and the preset user health profile data to obtain preliminary user health profile mapping data; Mapping the preliminary user health portrait mapping data according to the health index weighted data to obtain the user multi-dimensional health mapping data; Blood sugar control recommendations are made based on the user's multi-dimensional health mapping data to obtain blood sugar control recommendation data for SCRM diabetes multi-terminal interconnected data processing operations.

6. The method according to claim 3, It is characterized in that Multidimensional encoding includes: Performing feature decomposition according to the health-related feature data to obtain health-related feature decomposition data; Normalizing the health-related feature decomposition data to obtain health-related feature normalized data; Perform self-attention feature embedding on the normalized data of health-related features to obtain health-related feature embedding data; Quantitatively process the health-related feature embedded data to obtain health-related feature quantified data; The health-related feature quantified data is feature encoded to obtain health-related feature encoded data to be sent to the cloud platform.

7. The method according to claim 6, It is characterized in that Self-attention feature embedding includes: Perform channel coding and position coding on the normalized health-related feature data to obtain feature channel coding data and feature position coding data; Perform masked multi-head self-attention calculation on the feature position encoding data to obtain feature masked multi-head self-attention data; Perform multi-head self-attention calculation on the feature channel encoding data to obtain feature multi-head attention data; Perform layer normalization on the feature multi-head attention data to obtain feature layer normalized data; Perform full connection layer processing on the normalized data of the feature layer to obtain feature full connection layer data; Perform activation calculation on the feature fully connected layer data to obtain the feature feedforward neural network feature data; According to the feature mask multi-head self-attention data and the feature feedforward neural network feature data, residual connection and layer normalization are performed to obtain the health-related feature embedding data.

8. The method according to claim 3, It is characterized in that Multiple decoding includes: The control cloud platform performs multi-level feature decoding based on the health-related feature embedded data to obtain feature decoding data; Performing inverse vector quantization processing on the characteristic decoded data to obtain characteristic inverse quantized data; The feature reverse quantization data is de-embedded to obtain the user's multi-dimensional interconnected feature data.

9. The method according to claim 8, It is characterized in that Multi-level feature decoding includes: The control cloud platform performs Huffman decoding based on the health-related feature embedded data to obtain primary decoding data; Performing entropy coding and decoding on the primary decoded data to obtain secondary decoded data; performing error correction on the secondary decoded data and restoring it to obtain characteristic error-corrected data; The characteristic error correction data is subjected to noise reduction and automatic encoding to obtain characteristic decoding data.

10. A SCRM diabetes multi-terminal interconnected data processing system, It is characterized in that Used to execute the SCRM diabetes multi-terminal interconnected data processing method according to claim 1, the SCRM diabetes multi-terminal interconnected data processing system comprises: User multi-dimensional health data collection module, used to obtain user multi-dimensional health data; A user multi-dimensional interconnection feature extraction module is used to perform multi-terminal interconnection according to the user's multi-dimensional health data to obtain the user's multi-dimensional interconnection data, and to perform interconnection feature extraction according to the user's multi-dimensional interconnection data to obtain the user's multi-dimensional interconnection feature data; A health index weighted calculation module is used to perform health index mapping according to the multi-dimensional interconnected characteristic data of the user to obtain health index mapping data, and to perform channel equity processing according to the health index mapping data to obtain health index weighted data; The SCRM diabetes multi-terminal interconnected data processing module is used to map the user's multi-dimensional health data according to the health index weighted data to obtain the user's multi-dimensional health mapping data, and to make blood sugar control recommendations based on the user's multi-dimensional health mapping data to obtain blood sugar control recommendation data for SCRM diabetes multi-terminal interconnected data processing operations.

Images