Intelligent chip-based online health management method and system

Abstract

This invention discloses an online health management method and system based on a smart chip, relating to the field of data processing technology. The method includes: online collection of user characteristics of a target user, and acquisition of a health processing plan provided by a first person and a management processing plan provided by a second person; inputting the user characteristics into a management demand analysis agent configured within the smart chip, and outputting a predicted management demand information distribution; calculating the management matching degree distribution between the management processing plan and the predicted management demand information distribution, and processing and acquiring the online degree based on the target user's online records; adjusting and optimizing the management processing plan based on the online degree and management matching degree distribution to obtain the optimal management processing plan, and then pushing online health management information. This invention effectively improves the accuracy of non-medical aspects of online health management for chronic diseases.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and more specifically to an online health management method and system based on smart chips. Background Technology

[0002] With the continuous rise in the prevalence of chronic diseases and the accelerating aging of the population, chronic diseases have become a major threat to residents' health. Current chronic disease health management technologies mainly rely on physical examinations, screenings, and drug treatments conducted by physicians in medical institutions.

[0003] However, existing online health management for chronic diseases still faces some technical challenges in practical implementation: unclear responsibilities and lack of service standards make it difficult to effectively distinguish and integrate physician technical factors with non-physician service factors; insufficient intervention in non-medical factors such as lifestyle and psychological state leads to inaccurate management of non-medical aspects, making it difficult to motivate patients to actively participate and affecting the actual effectiveness of health management and user compliance. Summary of the Invention

[0004] This invention provides an online health management method and system based on smart chips, aiming to solve the technical problem of insufficient accuracy in the non-medical aspects of existing online health management for chronic diseases.

[0005] In view of the above problems, the present invention provides an online health management method and system based on smart chips.

[0006] In a first aspect, the present invention provides an online health management method based on a smart chip, comprising:

[0007] Collect user characteristics of target users online and obtain health treatment plans provided by the first person and management treatment plans provided by the second person.

[0008] The user characteristics are input into the management demand analysis agent configured in the smart chip, and the output is the predicted distribution of management demand information.

[0009] Calculate the management matching degree distribution of the management processing scheme and the distribution of predicted management demand information, and process and obtain the online degree based on the online records of the target users;

[0010] Based on the distribution of online status and management matching degree, the management processing scheme is adjusted and optimized to obtain the optimal management processing scheme and to push online health management information.

[0011] Secondly, the present invention provides an online health management system based on a smart chip, comprising:

[0012] The user feature collection module is used to collect the user features of the target user online and obtain the health treatment plan given by the first person and the management treatment plan given by the second person.

[0013] The management demand analysis module is used to input the user characteristics into the management demand analysis intelligent agent configured in the intelligent chip, and output the predicted distribution of management demand information.

[0014] The matching degree and online degree calculation module is used to calculate the management matching degree distribution of the management processing scheme and the distribution of predicted management demand information, and to process and obtain the online degree based on the online records of the target users;

[0015] The solution optimization and push module is used to adjust and optimize the management and processing solution based on the online degree and management matching degree distribution, obtain the optimal management and processing solution, and push the online health management solution.

[0016] One or more technical solutions provided in this invention have at least the following technical effects or advantages:

[0017] This invention provides an online health management method and system based on a smart chip. By collecting user characteristics online and integrating multi-dimensional intervention plans from both a first-person and a second-person perspective, the system relies on an intelligent agent embedded in the smart chip to accurately predict the distribution of users' health management needs. It then calculates the matching degree between the management solution and the demand distribution, and dynamically obtains online performance indicators based on user online behavior data. On this basis, the management solution is adaptively adjusted and optimized according to the online performance and matching degree distribution, ultimately generating and pushing the optimal comprehensive management solution. This invention achieves dynamic, accurate optimization and efficient online push of management solutions, effectively improving the adaptability of management solutions to user needs, solving the problem of insufficient accuracy in non-medical health management, enhancing the personalization level and real-time response capability of health management solutions, and providing an efficient implementation path for home-based online health management of chronic diseases. Attached Figure Description

[0018] Figure 1 A flowchart illustrating the online health management method based on a smart chip provided in an embodiment of the present invention;

[0019] Figure 2 A schematic diagram of the structure of an online health management system based on a smart chip provided in an embodiment of the present invention;

[0020] The components represented by each number in the attached diagram are explained below:

[0021] User feature collection module 11, management requirements analysis module 12, matching degree and online degree calculation module 13, and solution optimization and push module 14. Detailed Implementation

[0022] This invention provides an online health management method and system based on smart chips, which addresses the technical problem of insufficient accuracy in the non-medical aspects of existing online health management for chronic diseases.

[0023] Example 1, as Figure 1 As shown, this invention provides an online health management method based on a smart chip, the method comprising:

[0024] S100: Collects user characteristics of target users online and obtains health treatment plans provided by the first person and management treatment plans provided by the second person.

[0025] In this embodiment of the invention, user characteristics of the target user are collected online, and a health treatment plan provided by a first person and a management treatment plan provided by a second person are obtained. Online health management of chronic diseases requires accurate individual user information as a foundation, while integrating medical technical interventions by physicians and non-medical service interventions by non-physicians to achieve personalized health management goals. Home-based chronic disease management requires the online platform to first collect user health information and simultaneously obtain professional medical treatment plans and non-medical management plans, achieving preliminary integration of medical technical factors and non-medical service factors, thus laying a solid data and plan foundation for subsequent full-process online health management.

[0026] Step S100 in the method provided in this embodiment of the invention includes:

[0027] The system receives user characteristics uploaded by target users online, including user vital signs data.

[0028] The system receives health treatment plans and management treatment plans from the first and second personnel based on the uploaded user characteristics. The health treatment plans include medical treatment content, and the management treatment plans include non-medical treatment content.

[0029] First, the system receives user characteristics uploaded by target users online. These characteristics include user vital signs data. Target users refer to patients with chronic diseases requiring health management, such as those with hypertension, diabetes, or chronic respiratory diseases. User characteristics are a multi-dimensional set of information reflecting the target user's health status, including at least vital signs data, and can be extended to lifestyle habits, psychological state, and past medical history. Vital signs data refer to physiological indicators obtained through wearable devices, home medical instruments, or manual input, such as blood pressure, blood sugar, heart rate, blood oxygen saturation, weight, and body fat percentage.

[0030] Specifically, the platform receives user characteristic data proactively uploaded by target users or automatically synchronized through smart devices via its front-end interface or API. The platform supports multiple data access methods: users can manually fill out questionnaires, such as dietary records, sleep duration, and emotional state, or automatically upload real-time vital signs data via Bluetooth or Wi-Fi connection to devices such as smart bracelets, blood pressure monitors, and blood glucose meters. All data is encrypted during transmission and stored in a cloud database, with timestamps for dynamic analysis.

[0031] For example, the target patient is a person with hypertension. This patient binds a home smart blood pressure monitor to a mobile app, measures their blood pressure every morning and evening, and automatically uploads the data to the platform. Simultaneously, the patient completes a weekly questionnaire on lifestyle habits sent by the platform, recording recent steps, dietary preferences (salt or salt), sleep quality, and mood fluctuations. The above data collectively constitute the patient's user characteristics, with blood pressure values ​​such as systolic pressure of 145 mmHg and diastolic pressure of 95 mmHg representing the user's vital signs.

[0032] Secondly, the system receives health and management treatment plans provided by both the first and second personnel based on the uploaded user characteristics. The health treatment plans include medical interventions, while the management treatment plans include non-medical interventions. The first personnel refers to professionals with medical prescription rights, typically licensed physicians, responsible for developing medical intervention plans based on user characteristics, focusing on medications, medical devices, and clinical examinations. The second personnel refers to non-physician health service professionals, including but not limited to exercise therapists, nutritionists, psychologists, and health managers, responsible for developing non-medical management plans based on user characteristics, focusing on lifestyle interventions. The health treatment plan includes medical interventions provided by the first personnel, such as adjusting the type or dosage of antihypertensive medication, recommending the use of specific medical devices, and arranging regular follow-up examinations. The management treatment plan includes non-medical interventions provided by the second personnel, such as customizing personalized exercise plans, dietary adjustments, and psychological relaxation training.

[0033] Specifically, the online health management platform anonymizes user characteristic data and distributes it to the designated first and second personnel. The first personnel logs into the platform's physician interface to view user characteristic trends, such as blood pressure fluctuation curves, and, based on online consultations or historical medical records, issues an electronic health management plan. The plan clearly specifies medication names, usage and dosage, precautions, etc. The second personnel logs into the corresponding professional interface and, based on the user's lifestyle and vital sign data, develops a personalized management plan. For example, for hypertensive patients, this might involve designing a daily 30-minute brisk walking plan, recommending a low-sodium diet, or providing mindfulness-based stress reduction audio. Both types of plans are entered through the platform's structured forms, with the assigned personnel type clearly indicated, facilitating subsequent weighting and integration.

[0034] For example, the first person reviewed the patient's blood pressure data for the past two weeks and found that the patient's morning peak blood pressure was high, accompanied by a complaint of mild dizziness. Therefore, a management plan was proposed: the original amlodipine 5mg / day was adjusted to amlodipine 5mg combined with valsartan 80mg / day, and a follow-up ambulatory blood pressure check was recommended in two weeks. The second person, based on the patient's exercise habits questionnaire and body mass index, proposed a management plan: an exercise plan of 30 minutes of brisk walking daily, maintaining a heart rate of 110-120 beats / minute, and three low-salt recipes were recommended. Simultaneously, a psychological counselor, based on the patient's self-assessment of emotional state, provided 10 minutes of mindfulness meditation audio daily and recommended weekly online psychological counseling. Both plans were entered into the platform and stored in association with the target patient's user characteristics.

[0035] In this embodiment of the invention, user characteristics are collected and verified through a multi-channel, standardized online approach, ensuring the accuracy, completeness, and standardization of basic data for chronic disease health management. This provides reliable data support for subsequent management needs analysis and solution optimization. Simultaneously, the platform enables professional division of labor between primary and secondary personnel, allowing for the simultaneous acquisition of medical and non-medical management solutions tailored to user characteristics. This completes the initial integration of medical technology factors and non-medical service factors, resolving the existing problem of separation between medical and non-medical solutions in chronic disease health management and clarifying the professional compatibility of the two solutions. Furthermore, the fully online operation caters to the needs of home-based chronic disease management, improving the ease of use for both users and professionals, and enhancing the timeliness and accuracy of health management.

[0036] S200: Input the user characteristics into the management demand analysis agent configured in the smart chip, and output the predicted management demand information distribution.

[0037] In this embodiment of the invention, the user characteristics are input into a management demand analysis agent configured within a smart chip, and the output is a predicted distribution of management demand information. In chronic disease health management, the health needs of different users exhibit significant individual differences and dynamic changes, encompassing both medical treatment needs and non-medical behavioral intervention needs. Therefore, this step pre-configures a trained management demand analysis agent within the smart chip. This agent, composed of multiple parallel analysis networks, can deeply mine the management demand information implicit in user characteristics from multiple perspectives, outputting a predicted demand distribution covering both medical and non-medical dimensions. This provides a quantitative basis for subsequent solution matching and optimization, thereby improving the accuracy and comprehensiveness of health management.

[0038] Step S200 in the method provided in this embodiment of the invention includes:

[0039] A management demand analysis agent that has been pre-configured and trained is configured in the smart chip, wherein the management demand analysis agent includes multiple management demand analysis networks;

[0040] The user characteristics are input into the management demand analysis agent, which outputs multiple predicted management demand information as the predicted management demand information distribution.

[0041] First, a management demand analysis agent that has been trained is pre-configured in the smart chip, wherein the management demand analysis agent includes multiple management demand analysis networks.

[0042] The training and configuration steps for the management demand analysis agent include:

[0043] Based on historical user data of health management within the online platform, a set of sample user characteristics is collected, and sample management processing solutions that have passed management verification under different sample user characteristics are collected, and a set of sample management demand information is obtained by labeling.

[0044] The sample user feature set and the sample management requirement information set are randomly divided multiple times to obtain multiple sets of management requirement training data.

[0045] Based on deep learning, multiple management demand analysis networks are constructed, and each network is trained and validated using the multiple sets of management demand training data. After convergence, these networks are configured in the smart chip.

[0046] First, based on historical user data from health management on the online platform, a sample user characteristic set is collected. Then, management solutions that have passed verification under different sample user characteristics are collected, and a sample management demand information set is obtained. The sample user characteristic set refers to a large amount of user characteristic data collected from historical health management cases, including vital signs, lifestyle habits, past medical history, medication records, etc., covering different types of chronic diseases and different disease stages. A management solution that has passed verification refers to an effective management plan that has been validated in practice and is jointly or separately issued by the first and second personnel. The sample management demand information set refers to standardized demand labels or quantitative indicators extracted from the effective management plans, such as the need for adjustment of antihypertensive medication, the need to increase aerobic exercise duration, and the need for psychological relaxation training, which serve as training labels.

[0047] Specifically, the online platform extracts a large amount of user data from historical databases that have completed health management cycles, and filters out cases where management effectiveness has met standards. For each case, user characteristics and corresponding effective management plans are collected. Then, professional annotators or automated rules analyze the core needs elements in the plan and transform them into a structured set of needs tags. For example, for hypertensive patients, if the plan includes medication dosage adjustment, it is labeled "Medical Intervention: Optimization of Antihypertensive Drugs"; if it includes a daily brisk walking plan, it is labeled "Non-Medical Intervention: Exercise Guidance". All annotation results and corresponding user characteristics together constitute a sample management needs information set.

[0048] For example, suppose the platform's historical data includes a hypertensive patient similar to the target patient. After three months of health management, this patient's blood pressure decreased from 160 / 100 mmHg to 130 / 85 mmHg, with no adverse drug reactions. The platform extracts the patient's user characteristics and management plan. The annotation team transforms the user characteristics and management plan information into a sample management needs information set: {Medical needs: combined antihypertensive drug therapy; Non-medical needs: ≥150 minutes of aerobic exercise per week; Non-medical needs: low-salt diet guidance}.

[0049] Secondly, the sample user feature set and sample management demand information set are randomly partitioned multiple times to obtain multiple sets of management demand training data. Multiple random partitioning refers to using methods such as cross-validation or bootstrapping to randomly divide the sample data into multiple parts. Each part contains user features and corresponding demand labels, used to train different sub-networks to enhance the model's diversity and generalization ability. The sample user feature set and sample management demand information set are randomly sampled multiple times with or without replacement, with a portion selected each time as the training set and the remainder as the validation set. Alternatively, multi-fold cross-validation can be used, dividing the data into training and validation sets each time, for a total of K times, resulting in K different training data subsets, each used to train multiple independent management demand analysis networks.

[0050] For example, the platform has 100,000 historical sample data. Five random partitions are performed, with 80,000 data points randomly selected as the training set and 20,000 data points as the validation set each time, resulting in five different training data sets, denoted as training set 1, training set 2, ..., training set 5.

[0051] Furthermore, based on deep learning, multiple management demand analysis networks are constructed, each undergoing supervised training and validation using the aforementioned multiple sets of management demand training data. After convergence, these networks are configured within the intelligent chip. The management demand analysis network refers to a prediction model built upon a deep neural network, with user feature vectors as input and probability distributions or multi-label prediction results for corresponding demand categories as output. Supervised training involves using sample user features from the training data as input and sample management demand information as labels, optimizing network parameters through a backpropagation algorithm to make the network prediction results approximate the true labels. The intelligent chip refers to a dedicated chip integrated into the health management terminal device, possessing low power consumption and high-performance inference capabilities, capable of running the trained model locally.

[0052] Specifically, a deep neural network architecture is used to construct the management demand analysis network. Each network structure can be flexibly designed according to the task complexity, typically including an input layer, multiple hidden layers, and an output layer. The number of nodes in the input layer corresponds to the dimension of user features, and the number of nodes in the output layer equals the preset total number of management demand categories. The number of hidden layers and neurons is adjusted experimentally to balance fitting ability and generalization performance. The ReLU activation function is chosen to introduce nonlinearity and alleviate the gradient vanishing problem. The output layer uses the Softmax activation function to convert the prediction results into probability distributions for each demand category. During training, a supervised learning paradigm is adopted, using user features as input and corresponding demand labels as supervision signals. The difference between the predicted probability distribution and the true label distribution is quantified using the cross-entropy loss function, and the network weights are iteratively updated using optimizers such as Adam or SGD based on mini-batch stochastic gradient descent. Each network is trained independently using a training set, and its performance is periodically evaluated on the corresponding validation set, monitoring changes in loss values ​​and accuracy metrics. Training stops when the validation set loss no longer decreases or shows no improvement for several consecutive cycles, ensuring model convergence and preventing overfitting. After training, the weight parameters of multiple networks are solidified and burned into the storage unit of the smart chip. The chip's built-in inference engine supports parallel computing of multiple models, enabling online real-time inference.

[0053] For example, five identical three-layer fully connected neural networks are constructed. Each network contains an input layer, two hidden layers, and an output layer. The input layer has 20 nodes, corresponding to 20 user characteristics such as age, gender, systolic blood pressure, diastolic blood pressure, heart rate, steps taken, sleep duration, and salt intake score. The first hidden layer has 64 neurons with ReLU activation function; the second hidden layer has 32 neurons with ReLU activation function. The output layer has 10 nodes, corresponding to 10 management needs such as medication adjustment, exercise guidance, dietary intervention, and psychological counseling, with Softmax activation function. The training data comes from 100,000 sample cases of management effectiveness achieved in the platform's historical database. Each case includes user characteristics and their corresponding need labels. The sample data is randomly divided into five parts, each containing 80,000 training samples and 20,000 validation samples, used for training the five networks respectively. Training uses mini-batch gradient descent with a batch size of 128, cross-entropy as the loss function, Adam as the optimizer, and an initial learning rate of 0.001. Each network was trained for 100 epochs, with accuracy and loss calculated on the validation set after each epoch. Training was stopped early if the loss did not decrease for five consecutive validation epochs. After training, the accuracy of the five management demand analysis networks on the validation set all reached over 85%, indicating model convergence and no overfitting. Finally, the weight parameters of the five management demand analysis networks were embedded into the smart chip of the home smart health management terminal. The smart chip supports parallel invocation of the five management demand analysis networks for real-time inference.

[0054] Finally, the user characteristics are input into the management demand analysis agent, which outputs multiple predicted management demand information items, forming the predicted management demand information distribution. Predicted management demand information refers to the demand vector output by each management demand analysis network based on the input user characteristics, representing the probability or confidence level of the user in different demand categories. The predicted management demand information distribution is a set of outputs from multiple networks, comprehensively representing the diverse possibilities of user management demands. Subsequent steps can calculate the matching degree based on the predicted management demand information distribution.

[0055] For example, the target user characteristics obtained by S100 are: age 58, systolic blood pressure 145 mmHg, diastolic blood pressure 95 mmHg, heart rate 78 bpm, daily exercise 2000 steps, sleep 6 hours, and high salt intake score. After inputting the user characteristics into the smart chip, five management demand analysis networks output demand prediction probabilities respectively. The summation yields the following distribution: for the demand for optimizing antihypertensive medication, the five management demand analysis networks give probabilities of 0.9, 0.85, 0.88, 0.92, and 0.87 respectively; similarly, for the other nine types of management demands, the five management demand analysis networks provide corresponding probabilities. Each network outputs a probability vector of length 10, representing the user's matching probability for the 10 types of demands. The outputs of all networks are summed to form a 5×10 prediction matrix, which serves as the distribution of predicted management demand information.

[0056] In this embodiment of the invention, multiple independently trained deep learning networks are used to perform parallel analysis of user features, outputting multi-angle predicted distributions of management demand information. This effectively overcomes the bias and overfitting problems that may exist in a single model, improving the robustness and accuracy of demand prediction. Simultaneously, the multiple trained networks are deployed in a smart chip, enabling localized, low-latency real-time inference, ensuring the privacy and security of user data, and providing quantitative input for subsequent dynamic optimization of management solutions. This step shifts health management from passive response to proactive prediction, accurately identifying users' potential needs in both medical and non-medical dimensions, laying a scientific foundation for the generation of personalized comprehensive management solutions.

[0057] S300: Calculate the management matching degree distribution of the management processing scheme and the distribution of predicted management demand information, and process and obtain the online degree based on the online records of the target users.

[0058] In this embodiment of the invention, the management matching degree distribution of the management processing scheme and the distribution of predicted management demand information is calculated, and the online degree is obtained based on the online records of the target users. The optimization of online health management schemes for chronic diseases requires simultaneous reliance on both the suitability of the scheme to the user's actual needs and the user's health management participation status. The quantitative results of both are the main basis for subsequent accurate scheme optimization. In existing chronic disease management, there is a lack of multi-dimensional quantitative matching analysis between non-medical management schemes and user management needs. The suitability judgment of the scheme relies heavily on subjective experience, and there is no scientific quantitative evaluation method for the online participation status of users in home health management, which cannot provide an objective and accurate reference for scheme optimization. In home chronic disease management, the user's online participation directly affects the implementation effect of the management scheme. If the scheme optimization only considers the suitability of needs and ignores the user's participation ability, it is easy to make the optimized scheme difficult to implement. Therefore, this step calculates the management matching degree distribution through scientific algorithms and obtains the online degree based on the user's online records, realizing dual quantification of scheme suitability and user participation status, ensuring that the final pushed scheme not only meets the user's needs but also has a good execution foundation.

[0059] Step S300 in the method provided in this embodiment of the invention includes:

[0060] Calculate the similarity between the management processing scheme and multiple predicted management demand information within the predicted management demand information distribution, and use this as the management matching degree to obtain the management matching degree distribution;

[0061] Obtain the online records of target users within a preset time range in the past, and analyze and process them to obtain the online status.

[0062] First, the similarity between the management processing plan and multiple predicted management demand information within the predicted management demand information distribution is calculated as the management matching degree, resulting in the management matching degree distribution. The management processing plan refers to the non-medical intervention plan provided by the second person based on user characteristics in step S100. It typically includes specific suggestions across multiple dimensions, such as exercise plans, dietary guidance, and psychological relaxation training, and can be converted into a structured vector representation. The predicted management demand information distribution refers to the set of prediction results output in parallel by multiple management demand analysis networks within the intelligent chip in step S200. Each network outputs a probability vector representing the intensity of the user's demand in different demand categories. Similarity is a quantitative indicator measuring the closeness between the management processing plan vector and each predicted demand vector, and can be calculated using methods such as cosine similarity, Euclidean distance, or dot product.

[0063] First, the management and handling plan provided by the second person is analyzed according to the platform's preset demand categories to determine the intervention intensity of the plan for each demand category. For example, if the plan includes exercise guidance, a higher score is assigned to that category; if the plan does not involve psychological counseling, a zero score is assigned to that category.

[0064] Then, for each network's output of predicted demand information within the intelligent chip, it is compared item by item with the parsed management processing scheme, and the degree of matching between the two is calculated comprehensively. Specifically, the score of each scheme is compared with the probability value predicted by the network; the closer the two are, the higher the degree of matching for that item. Finally, the matching results for all categories are weighted and averaged to obtain a comprehensive matching degree value. This process is repeated for each network to obtain multiple matching degree values, forming a management matching degree distribution.

[0065] For example, the management and treatment plan provided by the second person for the target patient includes: 30 minutes of brisk walking daily, a low-salt diet, and 10 minutes of mindfulness meditation daily. The platform's preset non-medical demand categories include exercise guidance, dietary intervention, psychological counseling, and sleep management. The plan is parsed as follows: exercise guidance score 1.0, dietary intervention score 1.0, psychological counseling score 1.0, and sleep management score 0. The predicted non-medical demand information output by the five networks in step S200 is as follows: Network 1: Exercise guidance 0.7, dietary intervention 0.6, psychological counseling 0.3, sleep management 0.2; ...; Network 5: Exercise guidance 0.8, dietary intervention 0.62, psychological counseling 0.35, sleep management 0.2. By comparing each category and calculating the cosine similarity, the matching degree between each network and the management processing scheme is obtained as follows: Network 1 matching degree 0.85, Network 2 matching degree 0.82, Network 3 matching degree 0.79, Network 4 matching degree 0.84, Network 5 matching degree 0.88, that is, the management matching degree distribution is [0.85, 0.82, 0.79, 0.84, 0.88].

[0066] Secondly, obtain the target user's online records within a preset time range and analyze and process them to obtain the online status.

[0067] This includes obtaining online records of target users within a preset time range, analyzing and processing them to obtain online status, including:

[0068] Obtain the target user's online records within a preset time range and count the number of times they were online;

[0069] Obtain the baseline number of online users within a preset time range on the online platform, where the baseline number of online users is the maximum number of online users within the preset time range;

[0070] The online frequency is calculated by comparing the online frequency with the baseline online frequency.

[0071] First, the system retrieves the target user's online records within a preset time range and counts the number of online sessions. Online records refer to the target user's behavioral logs on the health management platform, including interactive events such as login, browsing, data upload, plan viewing, and online consultations. Each record is timestamped. The preset time range is a fixed duration window used to assess the user's recent activity, such as the past 7 days, 30 days, or 90 days, which can be flexibly set according to the management cycle. The platform extracts all online event records of the target user within the specified time range from the user behavior database and counts the number of days in which online behavior occurred, as the number of online sessions.

[0072] For example, for this target patient, the platform counted the patient's online records over the past 30 days: logging in and uploading blood pressure data through the app for 18 days, viewing program details 8 times, and interacting with the exercise therapist 3 times. Based on the number of days with activity, the total online activity was 18 days.

[0073] Secondly, a baseline online count is obtained within a preset time range on the online platform. This baseline online count represents the maximum number of times multiple users are online within that time range. The baseline online count serves as a reference value for normalizing user online activity, typically selecting the maximum online count among all active users within the same time range to reflect the upper limit of the most active user's behavior. Multiple users refer to all users participating in health management on the platform, or groups of chronic disease patients similar to the target user. Within the same preset time range, the platform counts the online counts of all active users and identifies the maximum as the baseline online count, used for subsequent calculations of relative activity. For example, if the platform counts the online days of all hypertension patients in the past 30 days and finds that one user logs in and records data every day for 30 days, then the baseline online count is 30.

[0074] Finally, the ratio of the number of online users to the baseline number of online users is calculated to obtain the online score. The online score is a normalized indicator of user online activity, ranging from [0,1]. A higher online score indicates a greater user dependence on the platform, increasing the weight of the management scheme predicted by the platform model in subsequent optimizations. The calculation formula is: Online Score = Number of User Online Users / Baseline Number of Online Users. If the baseline number of online users is 0, meaning no users are online, the online score is set to 0. For example, if the target user's online users are 18 and the baseline number of online users is 30, the online score is 18 / 30 = 0.6.

[0075] In this embodiment of the invention, by dimensionally decomposing the distribution of management processing schemes and predicted management needs information and using standardized algorithms to calculate similarity, a multi-dimensional quantitative analysis of the suitability of the schemes to user needs is achieved. The resulting management matching degree distribution reflects both the overall suitability level of the schemes and accurately identifies the shortcomings in the suitability of each non-medical intervention dimension, providing precise guidance for subsequent scheme optimization. Simultaneously, by screening valid online records and calculating online frequency based on platform benchmarks, a scientific quantitative assessment of users' online participation in home health management is achieved, objectively reflecting users' participation enthusiasm and actual participation ability. This provides dual important data for subsequent optimization of management processing schemes, ensuring that the optimized schemes accurately match user management needs while also considering users' actual execution capabilities, effectively improving the targeting, scientific nature, and implementability of the scheme optimization.

[0076] S400: Based on the online degree and management matching degree distribution, adjust and optimize the management processing scheme to obtain the optimal management processing scheme and push online health management information.

[0077] In this embodiment of the invention, the management processing plan is adjusted and optimized based on the online degree and management matching degree distribution to obtain the optimal management processing plan, which is then pushed to users for online health management. Optimization of non-medical management processing plans for chronic diseases needs to consider the adaptability of the plan to predicted management needs, the professional rationality of the original plan, and the user's actual online participation ability. Optimization based on a single dimension can easily lead to insufficient accuracy or high difficulty in implementation. In existing chronic disease management, plan optimization is mostly done manually and subjectively, lacking a standardized quantitative optimization process. It fails to combine user participation status with plan suitability and lacks a scientific adaptability evaluation system to judge the quality of the plan. This can easily result in optimized plans that, while adaptable to the needs, are difficult for users to implement, or that are user-friendly but deviate from professional requirements. Therefore, this step dynamically adjusts plan parameters and combines predicted matching degree and plan difference degree for weighted evaluation, ultimately generating the optimal management processing plan and pushing it to users, thereby improving the accuracy of health management and user compliance.

[0078] Step S400 in the method provided in this embodiment of the invention includes:

[0079] Based on the management matching degree distribution, calculate the initial management matching degree and configure the optimization step size;

[0080] Using the optimized step size, the non-medical treatment parameters within the management and treatment scheme are adjusted to obtain the first management and treatment scheme;

[0081] Calculate the average similarity between the first management processing scheme and multiple predicted management demand information within the predicted management demand information distribution, and use it as the first prediction matching degree;

[0082] Calculate the difference between the first management processing scheme and the management processing scheme, and calculate the first suggested matching degree;

[0083] Based on the online degree, the first predicted matching degree and the first suggested matching degree are fused and calculated to obtain the first management fitness degree;

[0084] Continue iteratively optimizing the management and processing solution until convergence is achieved, obtaining the optimal management and processing solution with the highest management adaptability, and then push online health management information.

[0085] First, based on the management matching degree distribution, the initial management matching degree is calculated, and the optimization step size is configured.

[0086] Specifically, based on the management matching degree distribution, the initial management matching degree is calculated, and the optimization step size is configured, including:

[0087] Get the preset step size;

[0088] Calculate the mean of the management matching degree distribution to obtain the initial management matching degree;

[0089] Based on the initial management matching degree, the configuration step size adjustment coefficient is calculated, and the preset step size is adjusted to obtain the optimized step size.

[0090] First, the preset step size is obtained. The preset step size refers to the basic range set in advance for adjusting various non-medical treatment parameters in the management plan, such as adjusting exercise duration by 5 minutes each time or adjusting dietary salt intake by 0.5 grams each time. This serves as the initial step size for optimization searches. Based on the management need category and common intervention intensity range, the platform sets a default adjustment step size for each type of non-medical treatment parameter, which is stored in the system configuration. The step size can be customized according to different types of chronic diseases or user groups.

[0091] For example, for hypertension management of target patients, the platform presets the step size of non-medical treatment parameters as follows: exercise duration step size 5 minutes / day, dietary salt intake step size 0.5 grams / day, and psychological training duration step size 2 minutes / day.

[0092] Next, the mean of the management matching degree distribution is calculated to obtain the initial management matching degree. The management matching degree distribution refers to the set of matching degrees between the multiple network outputs calculated in step S300 and the initial management processing scheme. The initial management matching degree is the arithmetic mean of this management matching degree distribution, reflecting the overall fit between the initial scheme and the user's multi-angle prediction needs. The multiple management matching degree values ​​obtained in S300 are added together and divided by the number of networks to obtain the mean.

[0093] For example, the distribution of the management matching degree of the target patient obtained in S300 is [0.85, 0.82, 0.79, 0.84, 0.88], and (0.85+0.82+0.79+0.84+0.88) / 5=0.836, that is, the initial management matching degree is 0.836.

[0094] Further, based on the initial management matching degree, a step size adjustment coefficient is calculated to adjust the preset step size, resulting in an optimized step size. The step size adjustment coefficient is a proportional factor that dynamically adjusts the step size based on the initial matching degree. A higher matching degree indicates a better solution, allowing for a smaller step size for finer searching; a lower matching degree requires a larger step size for faster exploration. A mapping relationship is established: Step size adjustment coefficient = 1 - Initial management matching degree. The calculation formula is: Optimized step size = Preset step size × Step size adjustment coefficient. This ensures a larger step size when the matching degree is low and a smaller step size when the matching degree is high.

[0095] For example, given an initial management matching degree of 0.836, the step size adjustment coefficient = 1 - 0.836 = 0.164. Multiplying the preset step size by 0.164 yields the optimized step sizes: optimized step size for exercise duration = 5 × 0.164 = 0.82 minutes, optimized step size for dietary salt intake = 0.5 × 0.164 = 0.082 grams, and optimized step size for psychological training duration = 2 × 0.164 = 0.328 minutes.

[0096] Secondly, using the optimized step size, the non-medical treatment parameters within the management treatment plan are adjusted to obtain the first management treatment plan. Non-medical treatment parameters refer to quantifiable intervention indicators in the management treatment plan, such as daily exercise duration, salt intake, and frequency of psychological training. The first management treatment plan is a new plan obtained by modifying the parameters according to the optimized step size and a preset adjustment direction based on the initial plan. The adjustment direction is determined according to the nature of the parameter; for indicators beneficial to health, such as exercise duration and psychological training duration, a positive increase is adopted; for indicators that need to be controlled, such as salt intake, a negative decrease is adopted. According to the preset direction rules, the optimized step size is added to or subtracted from each non-medical treatment parameter in the initial plan to generate a set of candidate plans. For example, the optimized step size is increased for exercise duration and psychological training duration, and decreased for salt intake.

[0097] For example, the initial management plan for the target patient is: 30 minutes of brisk walking daily, salt intake controlled at 5 grams / day, and 10 minutes of meditation daily. Based on the optimized step length and adjustment direction, the non-medical treatment parameters are adjusted as follows: exercise duration = 30 + 0.82 = 30.82 minutes ≈ 31 minutes, dietary salt intake = 5 - 0.082 = 4.918 ≈ 4.9 grams, and mental training duration = 10 + 0.328 = 10.382 minutes ≈ 10.3 minutes. Therefore, the first management plan is obtained: 31 minutes of brisk walking daily, 4.9 grams of salt intake daily, and 10.3 minutes of meditation daily.

[0098] Next, the average similarity between the first management processing scheme and multiple predicted management demand information within the predicted management demand information distribution is calculated as the first prediction matching degree. The first prediction matching degree refers to the average matching degree between the new first management processing scheme and the predicted demand of each network, reflecting the degree of fit between the first management processing scheme and user needs. The first management processing scheme is parsed into a demand intensity vector using the same method as S300, and then the similarity is calculated with the predicted demand vector output by each management demand analysis network to obtain a set of matching degrees, and then their arithmetic mean is calculated.

[0099] For example, suppose that after calculating the cosine similarity between the first management processing scheme and the predicted demand vectors output by the five management demand analysis networks, the matching degrees obtained are 0.86, 0.83, 0.80, 0.85 and 0.89, respectively, and the average value is 0.846, that is, the first predicted matching degree is 0.846.

[0100] Furthermore, the difference between the first management treatment plan and the initial plan is calculated, and the first recommended matching degree is obtained. The difference refers to the magnitude of change between the first management treatment plan and the initial plan, which is comprehensively measured by the changes in each parameter. The first recommended matching degree reflects the rationality of the plan adjustment; generally, the smaller the difference, the more stable the plan, and the higher the recommended matching degree. The absolute change of the first management treatment plan and the initial plan for each non-medical parameter is calculated, divided by the initial value to obtain the relative change rate. If the initial value is 0, the absolute change is taken, and then the average of the change rates of all parameters is taken as the difference. First recommended matching degree = 1 - difference.

[0101] For example, the initial plan: 30 minutes of exercise, 5 grams of salt intake, and 10 minutes of meditation. The first management treatment plan: 31 minutes of exercise, 4.9 grams of salt intake, and 10.3 minutes of meditation. Calculate the relative rates of change for each parameter: Exercise rate of change = |31-30| / 30 ≈ 0.033, Salt intake rate of change = |4.9-5| / 5 = 0.02, Meditation rate of change = |10.3-10| / 10 = 0.03. Average difference = (0.033+0.02+0.03) / 3 ≈ 0.028. First recommendation fit = 1-0.028 = 0.972.

[0102] Then, based on the online degree, the first predicted matching degree and the first suggested matching degree are fused and calculated to obtain the first management fitness degree.

[0103] Specifically, based on the online degree, the first predicted matching degree and the first suggested matching degree are fused and calculated to obtain the first management fitness degree, including:

[0104] The online degree is labeled as the prediction weight, and the suggested weight is calculated;

[0105] The first predicted matching degree and the first suggested matching degree are weighted and calculated using the predicted weight and the suggested weight to obtain the first management fitness degree.

[0106] First, the online activity level is labeled as the prediction weight, and the suggestion weight is calculated. The prediction weight is a coefficient used to weight the predicted matching degree, reflecting the importance of user online activity to the suitability of needs. A higher online activity level indicates that the user is more likely to actively participate and accept a solution that meets their needs. The suggestion weight is a coefficient used to weight the suggested matching degree, reflecting the importance of the stability of the solution. The online activity level obtained in step S300 is directly used as the prediction weight. Suggestion weight = 1 - Predicted weight = 1 - Online activity level. For example, if the patient's online activity level is known to be 0.6, then the prediction weight = 0.6, and the suggestion weight = 1 - 0.6 = 0.4.

[0107] Secondly, the predicted matching degree and the first suggested matching degree are weighted and calculated to obtain the first management fitness degree using the predicted weight and suggested weight. Management fitness degree is an evaluation index that comprehensively considers the fit between the solution and user needs, as well as the stability of the solution, and is used to compare the merits of different candidate solutions. The calculation formula is: First Management Fitness Degree = First Predicted Matching Degree × Predicted Weight + First Suggested Matching Degree × Suggested Weight. For example, if the first predicted matching degree is 0.846 and the first suggested matching degree is 0.972, the weighted calculation is: 0.846 × 0.6 + 0.972 × 0.4 ≈ 0.896, that is, the first management fitness degree is 0.896.

[0108] Finally, the management processing scheme is iteratively optimized until convergence, yielding the optimal management processing scheme with the highest management fitness, which is then pushed to the user for online health management. Iterative optimization involves repeatedly executing the above steps within the parameter space, generating a new candidate scheme each time, calculating its management fitness, and comparing it with the current best scheme, until convergence conditions are met, such as no further improvement in fitness after multiple consecutive iterations or reaching the maximum number of iterations. The optimal management processing scheme is the one with the highest management fitness obtained during the iteration process, i.e., the personalized scheme ultimately pushed to the user.

[0109] Specifically, in each iteration, multiple candidate solutions can be generated based on the current optimal solution. The fitness of each solution is calculated, and the optimal solution is retained. Then, based on this optimal solution, the step size is adjusted to continue the search until the fitness change is less than a preset threshold or the maximum number of iterations is reached. Finally, the solution with the highest fitness is output and pushed to the user through the platform.

[0110] For example, based on the first management plan, a second management plan is generated using a reduced optimization step size: 31.5 minutes of exercise, 4.85 grams of salt intake, and 10.5 minutes of meditation. The calculated second predicted fit is 0.85, the second suggested fit is 0.96, and the second management fitness is 0.85 × 0.6 + 0.96 × 0.4 = 0.894, slightly lower than the first plan's 0.896. Trying to adjust in the opposite direction, such as reducing exercise, might yield an even lower value. Assuming that after several iterations, the first plan still has the highest fitness, it is determined as the optimal management plan. Finally, the platform pushes the plan of "31 minutes of brisk walking daily, 4.9 grams of salt intake control, and 10.3 minutes of meditation daily" to the target user.

[0111] In this embodiment of the invention, a standardized and quantified management and processing solution optimization process is constructed to achieve dynamic iterative optimization of the initial management and processing solution. This process comprehensively considers the matching degree between the solution and the user's predicted needs, as well as the stability of the solution itself, and uses user online activity as a weight for fusion evaluation. The optimization process can adaptively adjust the step size, finding the optimal balance between demand alignment and solution continuity, ultimately generating a personalized and easily executable optimal management and processing solution. Pushing the optimal management and processing solution to users can improve user acceptance and compliance, thereby more effectively improving the management outcomes of chronic diseases.

[0112] Through the specific implementation methods described above, the embodiments of the present invention achieve the following technical effects:

[0113] This invention provides an online health management method and system based on a smart chip. First, by collecting user characteristics online and simultaneously acquiring multi-dimensional intervention plans from both physicians and non-physicians, it breaks the limitations of the traditional model where medical and non-medical factors are separated, laying a data foundation for subsequent quantitative integration. Based on this, multiple deep learning networks configured within the smart chip are used to perform parallel analysis of user characteristics, outputting multi-angle predictive management demand information distribution, realizing a shift from passive response to proactive prediction, and improving the accuracy and robustness of demand identification. Next, by calculating the matching degree distribution between management plans and predicted demands, and combining it with user online behavior data to generate an online degree index, it objectively quantifies the degree of plan fit and user participation potential, providing a scientific basis for dynamic optimization. Finally, based on the online degree and management matching degree distribution, an iterative optimization algorithm adaptively adjusts non-medical processing parameters, seeking the optimal balance between demand fit and plan stability, generating a personalized and easily executable optimal management processing plan and pushing it to the user. This invention effectively improves the accuracy of non-medical aspects of online health management for chronic diseases, effectively optimizes the overall effect of chronic disease health management, and provides an efficient implementation path for scientific online home management of chronic diseases.

[0114] Example 2, as Figure 2 As shown, the present invention provides an online health management system based on a smart chip, the system comprising:

[0115] User feature collection module 11 is used to collect user features of target users online and obtain health processing solutions given by a first person and management processing solutions given by a second person.

[0116] The management demand analysis module 12 is used to input the user characteristics into the management demand analysis intelligent agent configured in the intelligent chip, and output the predicted management demand information distribution.

[0117] The matching degree and online degree calculation module 13 is used to calculate the management matching degree distribution of the management processing scheme and the distribution of predicted management demand information, and to process and obtain the online degree based on the online records of the target user;

[0118] The scheme optimization and push module 14 is used to adjust and optimize the management and processing scheme according to the online degree and management matching degree distribution, obtain the optimal management and processing scheme, and push online health management.

[0119] In one embodiment, the user feature acquisition module 11 is further configured to:

[0120] The system receives user characteristics uploaded by target users online, including user vital signs data.

[0121] The system receives health treatment plans and management treatment plans from the first and second personnel based on the uploaded user characteristics. The health treatment plans include medical treatment content, and the management treatment plans include non-medical treatment content.

[0122] In one embodiment, the management requirements analysis module 12 is further configured to:

[0123] A management demand analysis agent that has been pre-configured and trained is configured in the smart chip, wherein the management demand analysis agent includes multiple management demand analysis networks;

[0124] The user characteristics are input into the management demand analysis agent, which outputs multiple predicted management demand information as the predicted management demand information distribution.

[0125] The training and configuration steps for the management demand analysis agent include:

[0126] Based on historical user data of health management within the online platform, a set of sample user characteristics is collected, and sample management processing solutions that have passed management verification under different sample user characteristics are collected, and a set of sample management demand information is obtained by labeling.

[0127] The sample user feature set and the sample management requirement information set are randomly divided multiple times to obtain multiple sets of management requirement training data.

[0128] Based on deep learning, multiple management demand analysis networks are constructed, and each network is trained and validated using the multiple sets of management demand training data. After convergence, these networks are configured in the smart chip.

[0129] In one embodiment, the matching degree and online degree calculation module 13 is further used for:

[0130] Calculate the similarity between the management processing scheme and multiple predicted management demand information within the predicted management demand information distribution, and use this as the management matching degree to obtain the management matching degree distribution;

[0131] Obtain the online records of target users within a preset time range in the past, and analyze and process them to obtain the online status.

[0132] This includes obtaining online records of target users within a preset time range, analyzing and processing them to obtain online status, including:

[0133] Obtain the target user's online records within a preset time range and count the number of times they were online;

[0134] Obtain the baseline number of online users within a preset time range on the online platform, where the baseline number of online users is the maximum number of online users within the preset time range;

[0135] The online frequency is calculated by comparing the online frequency with the baseline online frequency.

[0136] In one embodiment, the scheme optimization push module 14 is further configured to:

[0137] Based on the management matching degree distribution, calculate the initial management matching degree and configure the optimization step size;

[0138] Using the optimized step size, the non-medical treatment parameters within the management and treatment scheme are adjusted to obtain the first management and treatment scheme;

[0139] Calculate the average similarity between the first management processing scheme and multiple predicted management demand information within the predicted management demand information distribution, and use it as the first prediction matching degree;

[0140] Calculate the difference between the first management processing scheme and the management processing scheme, and calculate the first suggested matching degree;

[0141] Based on the online degree, the first predicted matching degree and the first suggested matching degree are fused and calculated to obtain the first management fitness degree;

[0142] Continue iteratively optimizing the management and processing solution until convergence is achieved, obtaining the optimal management and processing solution with the highest management adaptability, and then push online health management information.

[0143] Specifically, based on the management matching degree distribution, the initial management matching degree is calculated, and the optimization step size is configured, including:

[0144] Get the preset step size;

[0145] Calculate the mean of the management matching degree distribution to obtain the initial management matching degree;

[0146] Based on the initial management matching degree, the configuration step size adjustment coefficient is calculated, and the preset step size is adjusted to obtain the optimized step size.

[0147] Specifically, based on the online degree, the first predicted matching degree and the first suggested matching degree are fused and calculated to obtain the first management fitness degree, including:

[0148] The online degree is labeled as the prediction weight, and the suggested weight is calculated;

[0149] The first predicted matching degree and the first suggested matching degree are weighted and calculated using the predicted weight and the suggested weight to obtain the first management fitness degree.

[0150] It should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An online health management method based on smart chips, characterized in that, The method includes: Collect user characteristics of target users online and obtain health treatment plans provided by the first person and management treatment plans provided by the second person. The user characteristics are input into the management demand analysis agent configured in the smart chip, and the output is the predicted distribution of management demand information. Calculate the management matching degree distribution of the management processing scheme and the distribution of predicted management demand information, and process and obtain the online degree based on the online records of the target users; Based on the distribution of online status and management matching degree, the management processing scheme is adjusted and optimized to obtain the optimal management processing scheme and to push online health management information.

2. The online health management method based on a smart chip according to claim 1, characterized in that, The system collects user characteristics of target users online and obtains health management solutions provided by a first-level personnel and management solutions provided by a second-level personnel, including: The system receives user characteristics uploaded by target users online, including user vital signs data. The system receives health treatment plans and management treatment plans from the first and second personnel based on the uploaded user characteristics. The health treatment plans include medical treatment content, and the management treatment plans include non-medical treatment content.

3. The online health management method based on a smart chip according to claim 1, characterized in that, The user characteristics are input into a management demand analysis agent configured within a smart chip, and the output is a predicted distribution of management demand information, including: A management demand analysis agent that has been pre-configured and trained is configured in the smart chip, wherein the management demand analysis agent includes multiple management demand analysis networks; The user characteristics are input into the management demand analysis agent, which outputs multiple predicted management demand information as the predicted management demand information distribution.

4. The online health management method based on a smart chip according to claim 3, characterized in that, The training and configuration steps for the management requirements analysis agent include: Based on historical user data of health management within the online platform, a set of sample user characteristics is collected, and sample management processing solutions that have passed management verification under different sample user characteristics are collected, and a set of sample management demand information is obtained by labeling. The sample user feature set and the sample management requirement information set are randomly divided multiple times to obtain multiple sets of management requirement training data. Based on deep learning, multiple management demand analysis networks are constructed, and each network is trained and validated using the multiple sets of management demand training data. After convergence, these networks are configured in the smart chip.

5. The online health management method based on a smart chip according to claim 1, characterized in that, Calculate the management matching degree distribution of the management processing scheme and the distribution of predicted management demand information, and process and obtain the online degree based on the online records of the target users, including: Calculate the similarity between the management processing scheme and multiple predicted management demand information within the predicted management demand information distribution, and use this as the management matching degree to obtain the management matching degree distribution; Obtain the online records of target users within a preset time range in the past, and analyze and process them to obtain the online status.

6. The online health management method based on a smart chip according to claim 5, characterized in that, Obtain the target user's online records within a preset time range, analyze and process them to obtain online status, including: Obtain the target user's online records within a preset time range and count the number of times they were online; Obtain the baseline number of online users within a preset time range on the online platform, where the baseline number of online users is the maximum number of online users within the preset time range; The online frequency is calculated by comparing the online frequency with the baseline online frequency.

7. The online health management method based on a smart chip according to claim 1, characterized in that, Based on the online degree and management matching degree distribution, the management processing scheme is adjusted and optimized to obtain the optimal management processing scheme, and online health management push is performed, including: Based on the management matching degree distribution, calculate the initial management matching degree and configure the optimization step size; Using the optimized step size, the non-medical treatment parameters within the management and treatment scheme are adjusted to obtain the first management and treatment scheme; Calculate the average similarity between the first management processing scheme and multiple predicted management demand information within the predicted management demand information distribution, and use it as the first prediction matching degree; Calculate the difference between the first management processing scheme and the management processing scheme, and calculate the first suggested matching degree; Based on the online degree, the first predicted matching degree and the first suggested matching degree are fused and calculated to obtain the first management fitness degree; Continue iteratively optimizing the management and processing solution until convergence is achieved, obtaining the optimal management and processing solution with the highest management adaptability, and then push online health management information.

8. The online health management method based on a smart chip according to claim 7, characterized in that, Based on the management matching degree distribution, calculate the initial management matching degree and configure the optimization step size, including: Get the preset step size; Calculate the mean of the management matching degree distribution to obtain the initial management matching degree; Based on the initial management matching degree, the configuration step size adjustment coefficient is calculated, and the preset step size is adjusted to obtain the optimized step size.

9. The online health management method based on a smart chip according to claim 7, characterized in that, Based on the online degree, the first predicted matching degree and the first suggested matching degree are fused and calculated to obtain the first management fitness degree, including: The online degree is labeled as the prediction weight, and the suggested weight is calculated; The first predicted matching degree and the first suggested matching degree are weighted and calculated using the predicted weight and the suggested weight to obtain the first management fitness degree.

10. An online health management system based on a smart chip, characterized in that, The system is used to implement the online health management method based on a smart chip according to any one of claims 1-9, the system comprising: The user feature collection module is used to collect the user features of the target user online and obtain the health treatment plan given by the first person and the management treatment plan given by the second person. The management demand analysis module is used to input the user characteristics into the management demand analysis intelligent agent configured in the intelligent chip, and output the predicted distribution of management demand information. The matching degree and online degree calculation module is used to calculate the management matching degree distribution of the management processing scheme and the distribution of predicted management demand information, and to process and obtain the online degree based on the online records of the target users; The solution optimization and push module is used to adjust and optimize the management and processing solution based on the online degree and management matching degree distribution, obtain the optimal management and processing solution, and push the online health management solution.

Images