A health service and management system based on a "double platform cooperation" architecture

Abstract

This invention discloses a health service and management system based on a "dual-platform collaborative" architecture, belonging to the field of intelligent health service technology. It includes two interconnected platforms: a user-end platform and a service-end platform, respectively serving the healthcare recipients. The user-end platform pushes intelligent pre-diagnosis questionnaires for hepatobiliary diseases to the recipients and dynamically adjusts the question path according to the pre-set screening and analysis strategies in the service-end platform. Based on the responses submitted by the recipients on the user-end platform, the service-end platform automatically calculates the health risk level, generates personalized health prescriptions, and recommends matching specialist physicians. This invention constructs a military-grade dual-platform collaborative architecture, realizing a closed loop of hepatobiliary disease screening, emergency early warning, and intervention. It relies on a domestically developed security foundation and national cryptographic encryption to ensure data reliability, and enhances early detection and emergency response capabilities through dynamic evaluation and green channels, thereby strengthening the level of precise support for grassroots medical services.

Description

Technical Field

[0001] This invention relates to the field of intelligent health service technology, and in particular to a health service and management system based on a "dual-platform collaborative" architecture. Background Technology

[0002] In recent years, with the widespread adoption of smartphones and cloud services, personal health management applications have developed rapidly in the civilian market, and some military medical information systems have also gradually introduced pre-diagnosis, remote consultation, and electronic medical record sharing mechanisms. However, existing technologies mostly employ one-way data collection or static rule engines, lacking the ability to perceive and adaptively respond to dynamic health states. Furthermore, their security architecture generally relies on general-purpose commercial cloud platforms, making it difficult to meet the military's high-level security requirements for data sovereignty, communication encryption, device trustworthiness, and emergency response.

[0003] Existing technologies generally suffer from the following limitations: First, health risk assessment and intervention strategies are disconnected. Most current systems handle risk scoring, prescription generation, and physician recommendations in a fragmented manner, failing to form a closed-loop management process of "assessment-intervention-reassessment," making it difficult to support early warning and green channel linkage for high-incidence acute diseases such as hepatobiliary diseases. Second, security collaboration mechanisms are weak. User terminals are often distributed through public application markets, and identity authentication relies solely on a single password or SMS verification. Device binding and communication encryption strength are insufficient, and audit log retention periods are short and coarse-grained, failing to meet the military's Level 3 and above compliance requirements for full-chain traceability, prevention of screenshot leaks, and mandatory application of national cryptographic algorithms. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides a health service and management system based on a "dual-platform collaboration" architecture, which solves the problems of disconnect between health risk assessment and intervention strategies and weak security collaboration mechanisms.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0007] In a first aspect, the present invention provides a health service and management system based on a "dual-platform collaborative" architecture, which includes,

[0008] The intelligent consultation module is used to configure two interconnected platforms, one for the user end and one for the service end, respectively targeting the healthcare recipients.

[0009] The risk assessment module is used by the user-end platform to push intelligent pre-diagnosis questionnaires for hepatobiliary diseases to health care subjects, and dynamically adjust the question path according to the screening and analysis strategies preset in the server-end platform.

[0010] The doctor collaboration module is used to automatically calculate the health risk level, generate personalized health prescriptions, and recommend matching specialist physicians based on the response data submitted by the health care recipients on the user-end platform.

[0011] The emergency response module is used to load response data, health risk levels, and personalized health prescriptions to the doctor's end of the server platform, and automatically trigger the green channel when emergency signs are identified, sending admission arrangement instructions to offline medical institutions, and building and maintaining personal health records.

[0012] The health record follow-up module is used to automatically push follow-up reminders and appropriate science popularization content to the user platform based on personal health records and preset follow-up plans, while the server platform generates individualized risk assessment results based on historical data in personal health records.

[0013] As a preferred embodiment of the health service and management system based on the "dual-platform collaboration" architecture described in this invention, the configuration of two interconnected platforms, namely a user-end platform and a service-end platform respectively oriented towards the health care recipient, includes:

[0014] A microservice cluster based on domestic Linux is deployed within a data center that has passed the Level 3 certification of the military's network security level protection. User-end applications are only allowed to be installed through the military's unified application distribution platform. Upon first launch, two-factor authentication is performed. After successful authentication, the user-end generates an encrypted device fingerprint and completes the binding with the user's identity. At the same time, the functions of screenshot, screen recording, and data export are forcibly disabled. Communication adopts the HTTPS protocol. The server calls the military's PKI system for user authentication and verifies the legitimacy of the terminal through the device whitelist. Both parties negotiate the session key based on the ECDH protocol and use the national cryptographic SM4 algorithm to encrypt and transmit subsequent communication data. After the secure channel is established, the server issues collaborative configuration parameters, and the user-end automatically keeps the configuration synchronized through a heartbeat mechanism.

[0015] As a preferred embodiment of the health service and management system based on the "dual-platform collaboration" architecture described in this invention, the user-side platform pushes intelligent pre-diagnosis questionnaires for hepatobiliary diseases to the health-care recipients, and dynamically adjusts the question path according to the preset screening and analysis strategies in the server-side platform, including:

[0016] The user client requests and receives SM4-encrypted questionnaire data returned by the server through a secure channel. After decryption, a local questionnaire flowchart is constructed. The interaction begins with the initial question, and the next question is dynamically determined based on the healthcare recipient's answers and preset jump rules to achieve path adaptation. During the answering process, emergency keywords are detected in the free text in real time. Once a keyword is detected, the questionnaire is terminated and an emergency warning is reported. After the questionnaire is completed, the response data is uploaded to the server via SM4 encryption.

[0017] As a preferred embodiment of the health service and management system based on the "dual-platform collaboration" architecture described in this invention, wherein: based on the response data submitted by the healthcare recipient on the user-end platform, the server-end platform automatically calculates the health risk level, generates a personalized health prescription, and recommends matching specialist physicians, including:

[0018] The server receives and decrypts the response data uploaded by the user, extracts features through structured and unstructured dual-channel processing, and calculates a comprehensive risk score based on a preset screening strategy, classifying it into three risk levels: low, medium, and high. Based on the risk level and suspected disease type, it calls the health intervention rule base to generate a personalized health prescription, and matches and recommends the best physician from the specialist physician pool, integrating the risk level, health prescription, and recommended physician information into the evaluation result.

[0019] As a preferred embodiment of the health service and management system based on the "dual-platform collaboration" architecture described in this invention, the step of loading response data, health risk levels, and personalized health prescriptions onto the doctor's end of the server platform includes:

[0020] After the server generates the assessment results, it notifies the recommending physician through an asynchronous task. The physician initiates a request, and after identity and qualification verification, the server aggregates the patient questionnaire, risk assessment, health prescription, and health record to form an assessment package. The package is encrypted using SM4 and returned through a secure channel. The physician decrypts the package and loads it onto the workbench interface.

[0021] As a preferred embodiment of the health service and management system based on the "dual-platform collaboration" architecture described in this invention, the step of automatically triggering a green channel and sending admission arrangement instructions to offline medical institutions upon identification of emergency indications includes:

[0022] After the doctor loads the assessment data, the system immediately performs emergency rule matching: the free text is segmented and standardized in terms of terminology. If two keywords related to hepatobiliary emergencies are matched or "confusion" is matched alone, it is judged as a suspected emergency. For clinical features from structured question items, if the risk level is high and three features are present simultaneously—jaundice, high fever ≥38.5℃, and abdominal pain lasting 48 hours—it is judged as a highly suspected emergency. Once any rule is triggered, a high-priority warning immediately pops up on the doctor's workbench interface. After the doctor clicks to activate the green channel, a structured admission instruction object is generated. After the server verifies the signature and patient status, it matches the designated military tertiary hospital and sends the admission instruction to the target hospital through the military network using a standard medical interface. The system automatically reserves a bed and notifies the emergency department to prepare for admission. After the instruction is successfully sent, the server returns a receipt to the doctor's end, pushes an emergency notification to the user's end, and writes all key operations in the blockchain audit log.

[0023] As a preferred embodiment of the health service and management system based on the "dual-platform collaboration" architecture described in this invention, the construction and maintenance of personal health records includes:

[0024] Upon receiving feedback from the green channel and new health events, an update is triggered. Based on historical risk scores, risk change trends and dynamic health indices are calculated. After the update is completed, a synchronization data package is pushed to the user's device, and full lifecycle management of health records is implemented.

[0025] As a preferred embodiment of the health service and management system based on the "dual-platform collaboration" architecture described in this invention, the user-end platform automatically pushes follow-up reminders and appropriate science popularization content based on personal health records and preset follow-up plans, including:

[0026] The server triggers a follow-up assessment after the health record is updated. It determines the follow-up level based on the current risk level, historical follow-ups, prescription execution, and green channel records, and generates the next follow-up time based on the most recent health event or the last follow-up time.

[0027] Once the set time is reached, a follow-up task is automatically created and a push notification is triggered. Based on the disease type, risk characteristics, task environment, and behavioral compliance tags in the health record, appropriate popular science content is matched, merged with the reminder into a push package, encrypted with SM4, and sent to the user's terminal through a secure channel. The user's terminal then automatically pushes the content to the health care recipient.

[0028] As a preferred embodiment of the health service and management system based on the "dual-platform collaboration" architecture described in this invention, the server platform generates individualized risk assessment results based on historical data in the individual's health record, including:

[0029] After the health records are updated, the server retrieves the health management records of the health care recipients, standardizes key information and weights it according to time, calculates individualized risk scores based on pre-set hepatobiliary disease risk scoring rules, writes the assessment results including risk level, risk value and key risk factors into the health records, updates the dynamic health index simultaneously, pushes a complete report to the doctor and a summary to the user for follow-up and popular science, and stores the key assessment information in the blockchain audit log.

[0030] The beneficial effects of this invention are as follows: By constructing a dual-platform collaborative health service architecture for military scenarios, it achieves closed-loop management of the entire process of hepatobiliary disease screening, risk assessment, emergency early warning, physician collaboration, and follow-up intervention. On the one hand, relying on a domestically developed security platform and national cryptographic encryption communication, it ensures terminal trustworthiness, data controllability, and operational auditability, meeting the high-level network security requirements of the military. On the other hand, through dynamic questionnaire paths, multi-source feature fusion scoring, parallel identification of emergency rules, and automatic triggering mechanisms for green channels, it significantly improves the early detection rate of high-risk cases and the efficiency of emergency response. Simultaneously, based on the continuous updating and risk reassessment of personal health records, it supports precise and personalized health interventions, effectively enhancing the grassroots health support capabilities. Attached Figure Description

[0031] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0032] Figure 1 This is a structural diagram of a health service and management system based on a "dual-platform collaboration" architecture, as shown in Example 1.

[0033] Figure 2 This is a flowchart of the health service and management method in Example 1.

[0034] Figure 3 This is a flowchart of the smart health collaborative management service in Example 1. Detailed Implementation

[0035] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0036] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0037] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0038] Example 1, referring to Figures 1-3This is the first embodiment of the present invention, which provides a health service and management system based on a "dual-platform collaboration" architecture, including the following steps:

[0039] S1. Configure two interconnected platforms, one for the user and one for the service provider.

[0040] S1.1: Deploy a microservice cluster based on domestic Linux within a data center that has passed the Level 3 certification of the military's network security level protection.

[0041] Specifically, server clusters are deployed in data centers that have passed the Level 3 certification of the military's network security level protection. The servers use domestically produced Linux distribution software, and business components based on microservice architecture are deployed on the server clusters.

[0042] The business components include: screening engine service, health record service, doctor's workbench service, and access control service;

[0043] At the same time, full-link audit logs are enabled on the server-side platform to record all data access behaviors. The logs are retained for no less than 180 days. The server-side platform is deployed only in the military private network environment and does not adopt public cloud or hybrid cloud deployment methods.

[0044] S1.2: User-end applications are only allowed to be installed through the military's unified application distribution platform. Upon first launch, two-factor authentication is performed. After successful authentication, the user-end generates an encrypted device fingerprint and completes the binding with the user's identity. At the same time, the functions of screenshot, screen recording, and data export are forcibly disabled.

[0045] Specifically, the application will be distributed to authorized devices through the military application distribution platform, and installation through third-party app stores will be prohibited.

[0046] The client platform executes a two-factor authentication process upon initial startup, including:

[0047] Enter the military's unified identification code and read the digital certificate from the device's built-in military security TF card;

[0048] Once identity authentication is successful, the user-side platform generates a unique device fingerprint:

[0049] ;

[0050] in, For a pre-set random string, For terminal device identification code, This is the device's network interface address. For hash functions;

[0051] Generated device fingerprint After encryption, the data is uploaded to the server platform for registration and binding. At the same time, the functions of screenshot, screen recording, and data export are disabled in the application security policy.

[0052] S1.3: Communication uses the HTTPS protocol. The server calls the military PKI system for user authentication and verifies the legitimacy of the terminal through the device whitelist.

[0053] Specifically, the user platform initiates an HTTPS request to the server platform, carrying the device fingerprint. and user identification code;

[0054] The server-side platform invokes the military PKI identity authentication system to perform verification operations, including: verifying whether the user's identity code is valid and whether the user is in active duty; and verifying the device fingerprint. Is it already registered in the device whitelist?

[0055] Once the verification is successful, the server platform generates a temporary session token. The session token is valid for 30 minutes.

[0056] S1.4: Both parties negotiate the session key based on the ECDH protocol, use the national cryptographic SM4 algorithm to encrypt and transmit subsequent communication data, and after the secure channel is established, the server sends out collaborative configuration parameters, and the user end automatically keeps the configuration synchronized through the heartbeat mechanism.

[0057] Specifically, the two parties negotiate and generate a session key based on the Elliptic Curve Diffie-Hellman (ECDH) key exchange protocol. The SM4 national cryptographic algorithm is used to encrypt subsequent data transmission.

[0058] The SM4 encryption process is as follows:

[0059] ;

[0060] in, For plaintext data to be transmitted, For encrypted ciphertext data, A 128-bit session key;

[0061] After the communication channel is established, the server platform sends the platform collaboration configuration parameters to the user platform. When the server platform updates the configuration object, the user platform automatically synchronizes the latest configuration data through a heartbeat mechanism.

[0062] The user-end platform and the service-end platform refer to the Great Wall Internet Platform and the Eastern Hepatobiliary Internet Hospital, respectively.

[0063] S2. The user-side platform pushes intelligent pre-diagnosis questionnaires for liver and gallbladder diseases to healthcare recipients and dynamically adjusts the question path according to the pre-set screening and analysis strategies in the server-side platform.

[0064] S2.1: The user client requests and receives SM4 encrypted questionnaire data returned by the server through a secure channel. After decryption, a local questionnaire flowchart is constructed. The interaction starts from the initial question, and the next question is dynamically determined based on the content of the healthcare recipient's answers and the preset jump rules to achieve path adaptation.

[0065] Specifically, after completing the platform collaborative configuration synchronization, the user-side platform sends a questionnaire request through the established secure communication channel;

[0066] The requested data includes user identification code and device fingerprint. Session token and the current questionnaire version number;

[0067] After receiving the request, the server platform calls the screening engine service to perform the questionnaire loading operation and returns the corresponding version of the pre-diagnosis questionnaire structure.

[0068] The questionnaire uses a structured data model, including question number, question content, question type, option set, and path control rules;

[0069] The returned data is encrypted with SM4 and sent to the client platform. The client platform parses the decrypted questionnaire JSON data, extracts all question nodes, and constructs a local questionnaire question set.

[0070] ;

[0071] in, Indicates the first One problem node, ;

[0072] Each issue node contains path rules:

[0073] ;

[0074] When the problem The answer is At that time, jump to the question ;

[0075] Based on the set and its internal navigation relationships, the user-side platform constructs a complete questionnaire decision graph. :

[0076] ;

[0077] in, For the set of problem nodes, This is a set of path redirection rules;

[0078] The client-side platform initializes the local questionnaire process based on the structure and determines the first question node based on the questionnaire decision graph. ;

[0079] The user-side platform displays the problem in the interface. And wait for the recipient to input their response; the response data is as follows:

[0080] ;

[0081] in, It is the first step in the pre-consultation process A structured response record, Number the questions. Answering for users, To answer with a timestamp;

[0082] When the user submits an answer Then, the client platform performs dynamic path calculation based on the path rules:

[0083] ;

[0084] Among them, the function For path selection functions;

[0085] The path selection function Defined as:

[0086] ;

[0087] The path rule set R is pre-set in the questionnaire structure data and is distributed by the server platform along with the questionnaire. This is the preset default subsequent issue node;

[0088] If the answer matches the path rule, the user will be redirected to the corresponding question node; otherwise, the user will proceed to the default question node.

[0089] S2.2: During the response process, emergency keywords are detected in the free text in real time. Once a keyword is found, the questionnaire is terminated and an emergency warning is reported. After the questionnaire is completed, the response data is encrypted with SM4 and uploaded to the server.

[0090] Specifically, the user-side platform subsequently displayed the problem. During the questionnaire response process, the user-side platform performs emergency keyword detection on the response text. If any keywords are detected... If so, the questionnaire process will be terminated early;

[0091] The client platform loads the set of emergency keywords from the synchronized platform collaboration configuration parameters:

[0092] ;

[0093] During the questionnaire response process, each time a healthcare participant submits a free-text response... The user-side platform then executes the keyword matching function:

[0094] ;

[0095] like If so, an emergency alert event will be immediately sent to the server platform;

[0096] The keyword matching uses a multi-mode string matching algorithm. First, the input text is preprocessed by Chinese word segmentation, and then combined with a fuzzy matching strategy to efficiently identify multiple preset keywords in a single scan, thereby significantly improving the recall rate while ensuring processing efficiency.

[0097] When the questionnaire reaches its termination point:

[0098] ;

[0099] The data is packaged, encrypted using SM4, and then uploaded to the server platform.

[0100] ;

[0101] After receiving the data, the server platform writes it into the health record service and triggers the risk assessment process.

[0102] S3. Based on the response data submitted by the health care recipient on the user-end platform, the server-side platform automatically calculates the health risk level, generates a personalized health prescription, and recommends a matching specialist physician.

[0103] S3.1: The server receives and decrypts the response data uploaded by the user, extracts features through structured and unstructured dual-channel processing, and calculates a comprehensive risk score based on the preset screening strategy after fusion, classifying it into three risk levels: low, medium, and high.

[0104] Specifically, the user-side platform completes the pre-consultation questionnaire and generates questionnaire response data. and using session keys The response data is encrypted and uploaded to the server platform through the established secure communication channel;

[0105] After the server platform decrypts the received encrypted data, it sends the response data. Write the temporary assessment data into the health record service and perform dual-channel structured processing;

[0106] The dual channels include a structured channel and an unstructured channel, where the structured channel is used to parse single-choice and multiple-choice questions, mapping them to numerical feature vectors. The unstructured channel is used to perform Chinese clinical semantic parsing on free text responses, extracting symptom entities, duration, and severity, and generating supplementary feature vectors. ;

[0107] Numerical feature vectors and supplementary feature vectors Fusion into a risk feature vector for hepatobiliary diseases :

[0108] ;

[0109] in, Refers to the first A health risk characteristic value, , Refers to feature dimension;

[0110] The server platform performs risk score calculations based on preset disease screening and analysis strategies:

[0111] ;

[0112] in, For the Sigmoid function, The task status is adaptively weighted, with 0 indicating no risk and 100 indicating extremely high risk.

[0113] It should be noted that weight The range of values ​​is The specific process is as follows: A team of military experts specializing in the prevention and treatment of hepatobiliary diseases, in conjunction with grassroots military doctors, screened out cases based on data on high-incidence cases in the military over the past five years. An initial risk feature with clinical discriminative power was generated based on the historical pre-consultation records, follow-up results, and clinical diagnosis labels of 4218 officers and soldiers, forming an labeled training set, in which each sample contains a feature vector. The model is trained using a Lasso logistic regression model. For different military mission scenarios, a mode identifier is introduced to dynamically scale the basic weights. The model is evaluated on the validation set and is required to simultaneously meet the following conditions: sensitivity ≥ 86% and monthly high-risk warning volume ≤ 30 cases / 10,000 people. If these conditions are not met, the expert group will retrospectively adjust the upper limit of high-weight features to form a three-stage closed loop of "data training, resource simulation, and expert review". Finally, the weight vector is solidified into the server-side screening strategy configuration library after review and is prohibited from being modified at runtime.

[0114] The server-side platform classifies health risk levels based on risk scores:

[0115] ;

[0116] in, The low-risk threshold This is a high-risk threshold;

[0117] It should be noted that, The range of values ​​is , The range of values ​​is Based on data from a military pilot program for screening hepatobiliary diseases, and optimized using receiver operating characteristic (ROC) curve analysis combined with the Youden index, the optimal value was determined after review and calibration by a military medical expert group considering grassroots medical resources and tolerance for missed diagnoses. , Through a pilot program involving 4,218 people, screening for hepatobiliary diseases in troops undergoing high-intensity training, setting the low-risk threshold at 30 and the high-risk threshold at 70 ensures a sensitivity of ≥86% while keeping the number of specialist referrals within the capacity of primary healthcare. The thresholds were determined through three stages: initial selection via ROC analysis, resource pressure simulation, and review by a military medical expert group, and have been incorporated into the service-side screening strategy configuration.

[0118] S3.2: Based on the risk level and suspected disease type, generate personalized health prescriptions by calling the health intervention rule base, and match and recommend the best physician from the specialist physician pool, integrating the risk level, health prescription and recommended physician information into the evaluation result.

[0119] Specifically, the server-side platform calls upon the health intervention rule base to generate personalized health prescriptions:

[0120] ;

[0121] in, Refers to the risk level. Refers to the current risk feature vector. Refers to personal health record data;

[0122] The health prescription includes recommendations for examinations, medications, diet, and exercise.

[0123] The server platform matches doctors with specialists from the available pool of doctors based on risk level and suspected disease type. It possesses three attributes: specialty field, professional title level, and current patient load. ;

[0124] Calculate the recommendation score for each doctor:

[0125] ;

[0126] in, For the degree matching, Doctor's experience points , , Let be the weighting coefficient, satisfying ;

[0127] It should be noted that, The range of values ​​is , The range of values ​​is , The range of values ​​is The specific process is as follows: Based on historical data of the dispatch of military hepatobiliary specialists, an expert judgment matrix was constructed using the analytic hierarchy process (AHP) to initially determine the weight direction. This was then validated through multi-objective backtesting using 2847 real referral records from the past year. Finally, the optimal value was determined after review and calibration by a military medical expert group based on the allocation of grassroots medical personnel, the workload limit of key physicians, and the priority of medical support for officers and soldiers. = 0.50, = 0.35, = 0.15, combined in the pilot screening of hepatobiliary diseases for troops undergoing high-intensity training, has been verified by 4,218 people in actual operation. It can ensure that high-risk patients are accurately matched with specialist physicians (matching accuracy rate ≥91%), while reducing the average daily overload of backbone doctors by 32%, effectively maintaining the sustainable operation of the primary medical system. The weight coefficient is determined through a four-stage closed-loop process of "business constraint definition, AHP expert modeling, historical data backtesting, and review by the military medical expert group", and has been solidified into the physician intelligent recommendation strategy configuration of the server platform.

[0128] The server platform selects the doctor with the highest score as the recommended specialist physician and integrates health risk level, risk score, personalized health prescription and recommended physician information into a structured assessment result.

[0129] S4. Load the response data, health risk level, and personalized health prescriptions into the doctor's end of the server platform, and automatically trigger the green channel when an emergency sign is identified, send admission arrangement instructions to offline medical institutions, and build and maintain personal health records.

[0130] S4.1: After the server generates the structured assessment results, it notifies the recommending physician through an asynchronous task. The physician initiates a request, and after identity and qualification verification, the server aggregates the patient questionnaire, risk assessment, health prescription, and health record to form an assessment package. The package is encrypted using SM4 and returned through a secure channel. The physician decrypts the package and loads it onto the workbench interface.

[0131] Specifically, after generating the evaluation result object, the server platform pushes the task to the matched recommended specialist physician through an asynchronous task notification mechanism;

[0132] After receiving the task, the doctor's application initiates a data retrieval request based on the doctor's identity and the validity of the session;

[0133] After verifying that the requesting doctor's identity matches the recommended physician and that the doctor has the qualifications to diagnose and treat hepatobiliary diseases, the server platform aggregates the patient's questionnaire responses, risk assessment results, health prescriptions, and historical health records to form a doctor's view assessment package. This package is then encrypted using the SM4 algorithm with the doctor's exclusive session key and returned to the doctor's end through a secure communication channel. The doctor decrypts the package and loads the structured data into the doctor's workbench interface.

[0134] S4.2: After the doctor loads the assessment data, the emergency rule matching is immediately performed: the free text is segmented and the terminology is standardized. If two keywords for hepatobiliary emergency are matched or "confusion" is matched alone, it is judged as a suspected emergency.

[0135] Specifically, after the doctor's application completes the loading of the assessment package, it immediately calls the locally embedded emergency recognition engine to perform rule matching on the data in parallel: for the free text answers in the response data, it performs Chinese word segmentation and medical terminology standardization, and performs multi-modal fuzzy matching with the pre-built military hepatobiliary emergency keyword library:

[0136] The emergency keywords include jaundice, high fever greater than 38.5°C, persistent severe pain in the right upper quadrant, chills, and confusion;

[0137] If any text contains two or more keywords or a single hit of "confusion", it is marked as a suspected acute condition.

[0138] S4.3: For clinical features from structured question items, if the risk level is high and all three features are present—jaundice, high fever ≥38.5℃, and abdominal pain lasting for 48 hours—then it is considered a highly suspected emergency.

[0139] Specifically, regarding structured risk characteristics, if the health risk level is High and the risk feature vector... If a patient presents with all three positive characteristics—jaundice, high fever ≥38.5℃, and abdominal pain lasting ≥48 hours—then the high-risk combination rule is triggered, and the patient is marked as highly suspected of having an emergency.

[0140] S4.4: Once any rule is triggered, a high-priority warning immediately pops up on the doctor's workbench interface. After the doctor clicks to activate the green channel, a structured admission instruction object is generated. After the server verifies the signature and patient status, it matches the designated military tertiary hospital and sends the admission instruction to the target hospital through the military network using a standard medical interface. The hospital automatically reserves a bed and notifies the emergency department to prepare for admission. After the instruction is successfully sent, the server returns a receipt to the doctor's end, pushes an emergency notification to the user's end, and writes all key operations in the blockchain audit log.

[0141] Specifically, once any emergency rule is triggered, the doctor's application will immediately display a high-priority warning bar at the top of the workbench interface and provide options for one-click activation of the green channel and manual review;

[0142] When a doctor clicks to activate the green channel, the doctor's application generates a structured admission instruction object, which includes the patient's identity, military unit, emergency clinical summary, recommended department and urgency level.

[0143] The green channel dispatch service uses the doctor's private key to perform SM2 digital signature and uploads it to the server platform through the established secure communication channel;

[0144] After receiving the admission instruction, the server verifies the validity of the SM2 signature, checks the patient's current status, matches the target medical institution, calls the dedicated receiving endpoint interface of the green channel set up by the target hospital, sends the admission arrangement instruction in standard format through the military medical network, automatically reserves a bed and notifies the emergency department to prepare for admission;

[0145] After successfully sending the instruction, the server platform returns an execution receipt to the doctor, pushes an emergency notification to the user, and writes all key events of the process into the blockchain audit log.

[0146] S4.5: Upon receiving feedback from the green channel and new health events, an update is triggered. Based on historical risk scores, risk change trends and dynamic health indices are calculated. After the update is completed, a synchronization data package is pushed to the user's terminal, and full lifecycle management of health records is implemented.

[0147] Specifically, after the server platform confirms that the target hospital has successfully received the green channel admission instruction, a structured health event record object is generated based on this pre-diagnosis assessment task:

[0148] ;

[0149] in, For identifying military personnel receiving healthcare, This is a unique number for this risk assessment record. Code the target hospital. The admission record number assigned to the hospital. The timestamp of the event;

[0150] The server platform records health events into the health event table of the health record service and maintains a unique master index for each healthcare recipient's health record.

[0151] ;

[0152] in, For basic information of the healthcare recipients, This is the current health record version number. The most recent update time;

[0153] When a new health event record is detected, the server platform automatically triggers the health record update process, which integrates and updates the pre-consultation data, risk assessment results, personalized health prescriptions, green channel medical records and hospital feedback information.

[0154] The server-side platform calculates the trend of health risk changes based on historical risk score sequences. :

[0155] ;

[0156] in, The most recent assessment time The calculated health risk score, The time of the last assessment Health risk score, The time interval between the two assessments;

[0157] when When this occurs, it indicates an upward trend in health risks, and a risk warning marker is generated in the health record.

[0158] It should be noted that, To pre-determine the risk escalation threshold, the value range is 0.4 ≤ θ ≤ 0.6 points / day. The specific process is as follows: Based on the pilot data of military hepatobiliary disease screening, the initial selection is optimized by combining receiver operating characteristic (ROC) curve analysis with Youden index, and further simulation of the carrying capacity of grassroots medical resources is carried out. The upper limit is set at an average monthly risk warning volume of ≤30 cases / 10,000 people. After review and calibration by the military medical expert group based on the tolerance for missed diagnoses, requiring a missed diagnose rate of ≤5% and the grassroots treatment capacity, the optimal value θ = 0.5 points / day is determined. In the hepatobiliary disease screening for troops undergoing high-intensity training, the pilot verification of 4,218 people shows that, under the premise of ensuring a missed diagnose rate of ≤5%, the average monthly risk warning volume can be stably controlled at 28.7 cases / 10,000 people, effectively matching the actual carrying capacity of grassroots medical resources. The threshold is determined through a three-stage closed-loop process of "ROC analysis initial selection - resource pressure simulation - military medical expert group review" and has been solidified into the health record update strategy configuration of the server platform.

[0159] The server-side platform introduces dynamic health index calculation. :

[0160] ;

[0161] in, Assess the current risk level. For chronic disease risk scoring, , , These are the weighting coefficients;

[0162] It should be noted that the weighting coefficients , , The value ranges are 0.4≤ ≤0.6, 0.2≤ ≤0.4, 0.1≤ ≤0.3, and satisfy The specific values ​​can be determined based on historical health record data, using existing medical statistical methods such as multiple regression analysis, analytic hierarchy process, or machine learning parameter training methods. The specific process involves collecting longitudinal health records of 4000 officers and soldiers, including... , , Based on pre-consultation response records, physical examination and laboratory data, "deterioration of actual health status" is used as a positive sample label. Multiple linear regression or logistic regression modeling is employed to predict clinical outcomes using HI (hypothesis-based prediction), and optimization is achieved through cross-validation. , , To maximize the AUC or Youden index, the monthly average warning volume under different weight combinations was simulated on the validation set. Feasible solutions that meet the carrying capacity of primary care and have a missed diagnosis rate of ≤5% were selected. The candidate weights were submitted to the military medical expert group, and their clinical rationality was evaluated in conjunction with the retrospective assessment of typical cases. Finally, the optimal value was determined and solidified into the dynamic health index calculation module on the server side.

[0163] After the health record is updated, the server platform generates a health record synchronization object and pushes it to the user platform through the established secure communication channel.

[0164] Meanwhile, the server platform establishes a lifecycle management mechanism for personal health records, performing creation, incremental updates, cold data archiving, and audit traceability management of health records.

[0165] S5. Based on personal health records and preset follow-up plans, the user-side platform automatically pushes follow-up reminders and appropriate science popularization content, while the server-side platform generates individualized risk assessment results based on historical data in the personal health records.

[0166] S5.1: After the health record is updated, the server triggers a follow-up assessment, determines the follow-up level based on the current risk level, historical follow-ups, prescription execution and green channel records, and generates the next follow-up time based on the most recent health event or the last follow-up time.

[0167] Specifically, after the server platform completes the health record update and synchronization, the health record service sends a record update event to the follow-up management service, which includes the health care recipient's identity, current health risk level, risk score, risk change trend, dynamic health index and the timestamp of the most recent health event;

[0168] Upon receiving a record update event, the follow-up management service retrieves the individual's master health record index, historical follow-up records, past prescription execution records, and green channel processing records to determine the follow-up level. It is divided into three levels: high frequency, medium frequency, and low frequency. If the current health risk level is High, it is determined to be high frequency follow-up; if the risk is at a medium level... If there are no signs of urgent deterioration, then it is determined to be a medium-frequency follow-up; if the risk is low If the patient's health status is stable, then it is considered a low-frequency follow-up.

[0169] Follow-up period Determined using a piecewise mapping method:

[0170] High-frequency follow-up: 7 days; medium-frequency follow-up: 30 days; low-frequency follow-up: 90 days;

[0171] Recent health event time Or the time of completion of the most recent follow-up visit As the reference time for the next follow-up Generate the next follow-up time :

[0172] ;

[0173] S5.2: Upon reaching the set time, a follow-up task is automatically created and a push notification is triggered. Based on the disease type, risk characteristics, task environment, and behavioral compliance tags in the health record, appropriate popular science content is matched, merged with the reminder into a push package, encrypted with SM4, and sent to the user terminal through a secure channel. The user terminal then automatically pushes the content to the health care recipient.

[0174] Specifically, a follow-up task object is constructed and written to the task pool, when it arrives... The push process is automatically triggered. The server platform extracts disease type, risk characteristics, task environment and behavioral compliance tags from health records, matches appropriate health science content, combines it with follow-up reminders into a push package, and sends it to the user platform through a secure communication channel after being encrypted with SM4. The user platform then automatically pushes the package to the health care recipients.

[0175] The task environment refers to the current task environment of the health care subject, including regular camps, field training, and maritime duty, and is used to adapt health intervention strategies in different scenarios;

[0176] The behavioral compliance tags are automatically generated based on the user's completion of historical health prescriptions and follow-up tasks. These tags are used to evaluate the effectiveness of interventions and optimize subsequent push strategies.

[0177] S5.3: After the file is updated, the server retrieves the health management records of the health care recipient, standardizes the key information and weights it according to the time period, and calculates the individualized risk score based on the preset liver and gallbladder disease risk scoring rules.

[0178] Specifically, after the personal health record is updated, the server platform automatically retrieves the health management records of the health care recipient within the last 180 days, including risk scores, chronic disease indicators, symptom descriptions and follow-up feedback. The key information in the records is standardized and assigned different weights according to the time period.

[0179] Using pre-defined hepatobiliary disease risk scoring rules, standardized health data are evaluated to calculate an individualized hepatobiliary disease risk score:

[0180] ;

[0181] Among them, threshold

[0182] , These correspond to the low-risk threshold of the risk score. With high risk threshold ;

[0183] S5.4: Write the assessment results, including risk level, risk value and key risk factors, into the health record, update the dynamic health index in sync, push the full report to the doctor and the summary to the user for follow-up and popular science, and store the key assessment information in the blockchain audit log.

[0184] Specifically, the server-side platform calls upon health record services, filters relevant historical records based on the predicted target disease type, and constructs a personalized historical sequence:

[0185] ;

[0186] in, Each For the first Structured record of each health management event;

[0187] Define the attenuation factor:

[0188] ;

[0189] in, Refers to the first Timestamp of the occurrence of the health event The time interval between events and the present. Refers to the current timestamp;

[0190] It should be noted that the attenuation coefficient The typical value range is 0.01–0.05 / day, with 0.02 / day being preferred. This value can be determined using existing machine learning hyperparameter tuning techniques such as cross-validation on historical health data, grid search, or Bayesian optimization based on clinical expert feedback, combined with metrics like AUC, sensitivity, or calibration in risk prediction tasks. The preferred value of 0.02 / day was achieved through a pilot program of hepatobiliary disease screening among 4218 soldiers in a combined arms brigade, employing rolling backtesting and multi-objective constraint optimization methods to optimize the time decay coefficient. Multiple candidate values ​​were evaluated offline to screen out feasible solutions that simultaneously met the requirements of sensitivity ≥85% and monthly average warning volume ≤30 cases / 10,000 people. These solutions were then reviewed by the medical expert group in conjunction with clinical cases, and the final solution was determined. To achieve the optimal value, it can effectively highlight the predictive weight of key symptoms in the past two weeks to capture rapidly progressing hepatobiliary emergencies, while avoiding over-warning that exceeds the capacity of primary healthcare, thus achieving the best balance between risk identification accuracy and the sustainability of medical resources in military scenarios.

[0191] The processed output is a standardized historical feature sequence:

[0192] ;

[0193] in, , For the first Numericalized feature vector at time step;

[0194] server platform from Extracting specific risk dimensions for hepatobiliary diseases, we construct an individualized historical risk feature vector:

[0195] ;

[0196] Will Input a pre-set hepatobiliary disease risk prediction model to generate individualized risk probability values. :

[0197] ;

[0198] in, For bias terms , which is the model parameter vector;

[0199] The It includes an input layer, a fusion layer, and an output layer;

[0200] The input layer receives multi-source heterogeneous health features after standardization and time-series alignment, and constructs a dual-channel input vector, including a structured historical feature sub-vector and a time-series evolution feature sub-vector.

[0201] The fusion layer is as follows: a fully connected neural network or gradient boosting decision tree is used as the basic fusion device, a service context gating mechanism is introduced, the feature sensitivity weights are dynamically adjusted according to the combat unit to which the health care object belongs, and the fused high-order risk representation vector is output, while retaining the discriminative information of key risk patterns of liver and gallbladder diseases.

[0202] The output layer is: mapping the fused representation to an individualized risk probability of hepatobiliary diseases;

[0203] Model training steps: Standardized health records of officers and soldiers are extracted from historical data of military hepatobiliary disease screening. Input samples containing structured indicators and temporal evolution characteristics are constructed, and whether hepatobiliary-related clinical events will occur within the next 7 days are marked. L1 regularized logistic regression or gradient boosting tree is used with binary cross-entropy as the loss function. Service branch adaptive fine-tuning is introduced, and risk classification thresholds are initially selected based on ROC curves and Youden index. Combined with simulation of grassroots medical support resource carrying capacity and review by the military medical expert group, low-risk and high-risk thresholds are calibrated and solidified. After verifying the model sensitivity ≥85% and specificity ≥78% on an independent test set, the model is deployed as a server-side microservice component, and an incremental update mechanism based on new data is established every six months.

[0204] The server platform builds personalized historical sequences The system automatically extracts key risk drivers, generates explanatory summaries, integrates them into structured assessment results, writes the assessment results into the risk assessment history table of the individual's health record, updates the risk trend sequence and dynamic health index, pushes the complete assessment package to the doctor's platform, and pushes the summary version of the results to the user's platform for the next round of follow-up reminders and science popularization content adaptation. The assessment metadata is written into the blockchain audit log.

[0205] In summary, this invention achieves closed-loop management of the entire process of hepatobiliary disease screening, risk assessment, emergency early warning, physician collaboration, and follow-up intervention by constructing a dual-platform collaborative health service architecture for military scenarios. On the one hand, relying on a domestically developed security platform and national cryptographic encryption communication, it ensures terminal trustworthiness, data controllability, and operational auditability, meeting the high-level network security requirements of the military. On the other hand, through dynamic questionnaire paths, multi-source feature fusion scoring, parallel identification of emergency rules, and automatic triggering mechanisms for green channels, it significantly improves the early detection rate of high-risk cases and the efficiency of emergency response. Simultaneously, based on the continuous updating and risk reassessment of individual health records, it supports precise and personalized health interventions, effectively enhancing the grassroots health support capabilities.

[0206] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A health service and management system based on a "dual-platform collaborative" architecture, characterized in that: include, The intelligent consultation module is used to configure two interconnected platforms, one for the user end and one for the service end, respectively targeting the healthcare recipients. The risk assessment module is used by the user-end platform to push intelligent pre-diagnosis questionnaires for hepatobiliary diseases to health care subjects, and dynamically adjust the question path according to the screening and analysis strategies preset in the server-end platform. The doctor collaboration module is used to automatically calculate the health risk level, generate personalized health prescriptions, and recommend matching specialist physicians based on the response data submitted by the health care recipients on the user-end platform. The emergency response module is used to load response data, health risk levels, and personalized health prescriptions to the doctor's end of the server platform, and automatically trigger the green channel when emergency signs are identified, sending admission arrangement instructions to offline medical institutions, and building and maintaining personal health records. The health record follow-up module is used to automatically push follow-up reminders and appropriate science popularization content to the user platform based on personal health records and preset follow-up plans, while the server platform generates individualized risk assessment results based on historical data in personal health records.

2. The health service and management system based on a "dual-platform collaboration" architecture as described in claim 1, characterized in that: The configuration of two interconnected platforms, a user-end platform and a service-end platform respectively targeting the healthcare recipients, includes: A microservice cluster based on domestic Linux is deployed in a data center that has passed the Level 3 certification of the military's network security level protection. User-end applications can only be installed through the military's unified application distribution platform. Upon first launch, two-factor authentication is performed. After successful authentication, the user-end generates an encrypted device fingerprint and completes the binding with the user's identity. At the same time, the functions of screenshot, screen recording, and data export are forcibly disabled. Communication adopts the HTTPS protocol. The server calls the military's PKI system to authenticate the user's identity and verifies the legitimacy of the terminal through the device whitelist. Both parties negotiate the session key based on the ECDH protocol and use the national cryptographic SM4 algorithm to encrypt and transmit subsequent communication data. After the secure channel is established, the server issues collaborative configuration parameters, and the user-end automatically keeps the configuration synchronized through a heartbeat mechanism.

3. The health service and management system based on a "dual-platform collaboration" architecture as described in claim 2, characterized in that: The user-side platform pushes intelligent pre-diagnosis questionnaires for hepatobiliary diseases to healthcare recipients, and dynamically adjusts the question path according to the preset screening and analysis strategies in the server-side platform, including: The user client requests and receives SM4-encrypted questionnaire data returned by the server through a secure channel. After decryption, a local questionnaire flowchart is constructed. The interaction begins with the initial question, and the next question is dynamically determined based on the healthcare recipient's answers and preset jump rules to achieve path adaptation. During the answering process, emergency keywords are detected in the free text in real time. Once a keyword is detected, the questionnaire is terminated and an emergency warning is reported. After the questionnaire is completed, the response data is uploaded to the server via SM4 encryption.

4. The health service and management system based on a "dual-platform collaboration" architecture as described in claim 3, characterized in that: Based on the response data submitted by the healthcare recipients on the user-side platform, the server-side platform automatically calculates the health risk level, generates personalized health prescriptions, and recommends matching specialist physicians, including: The server receives and decrypts the response data uploaded by the user, extracts features through structured and unstructured dual-channel processing, and calculates a comprehensive risk score based on a preset screening strategy, classifying it into three risk levels: low, medium, and high. Based on the risk level and suspected disease type, it calls the health intervention rule base to generate a personalized health prescription, and matches and recommends the best physician from the specialist physician pool, integrating the risk level, health prescription, and recommended physician information into the evaluation result.

5. The health service and management system based on a "dual-platform collaboration" architecture as described in claim 4, characterized in that: The process of loading response data, health risk levels, and personalized health prescriptions into the doctor's end of the server platform includes: After the server generates the assessment results, it notifies the recommending physician through an asynchronous task. The physician initiates a request, and after identity and qualification verification, the server aggregates the patient questionnaire, risk assessment, health prescription, and health record to form an assessment package. The package is encrypted using SM4 and returned through a secure channel. The physician decrypts the package and loads it onto the workbench interface.

6. The health service and management system based on a "dual-platform collaboration" architecture as described in claim 5, characterized in that: The automatic activation of the green channel upon recognizing emergency signs, and the sending of admission arrangement instructions to offline medical institutions, includes: After the doctor loads the assessment data, the emergency rule matching is immediately executed: the free text is segmented and the terminology is standardized. If two keywords for hepatobiliary emergencies are matched or "confusion" is matched alone, it is judged as a suspected emergency. For clinical features from structured question items, if the risk level is high and three features are present at the same time: jaundice, high fever ≥38.5℃, and abdominal pain lasting for 48 hours, it is judged as a highly suspected emergency. After any rule is triggered, a high-priority warning immediately pops up on the doctor's workbench interface. After the doctor clicks to activate the green channel, a structured admission instruction object is generated. After the server verifies the signature and patient status, it matches the designated military tertiary hospital and sends the admission instruction to the target hospital through the military network using a standard medical interface. The bed is automatically reserved and the emergency department is notified to prepare for admission. After the instruction is sent successfully, the server returns a receipt to the doctor's end, pushes an emergency notification to the user's end, and writes the key operations of the whole process into the blockchain audit log.

7. The health service and management system based on a "dual-platform collaboration" architecture as described in claim 6, characterized in that: The construction and maintenance of personal health records includes: Upon receiving feedback from the green channel and new health events, an update is triggered. Based on historical risk scores, risk change trends and dynamic health indices are calculated. After the update is completed, a synchronization data package is pushed to the user's device, and full lifecycle management of health records is implemented.

8. The health service and management system based on a "dual-platform collaboration" architecture as described in claim 7, characterized in that: Based on personal health records and preset follow-up plans, the user-end platform automatically pushes follow-up reminders and appropriate science popularization content, including: After the health record is updated, the server triggers a follow-up assessment. The follow-up level is determined based on the current risk level, historical follow-ups, prescription execution, and green channel records. The next follow-up time is generated based on the most recent health event or the last follow-up time. When the set time is reached, a follow-up task is automatically created and a push notification is triggered. Based on the disease type, risk characteristics, task environment, and behavioral compliance tags in the health record, appropriate popular science content is matched, and combined with the reminder into a push package. After being encrypted with SM4, it is sent to the user terminal through a secure channel, and the user terminal automatically pushes it to the health care recipient.

9. The health service and management system based on a "dual-platform collaboration" architecture as described in claim 8, characterized in that: The server-side platform generates individualized risk assessment results based on historical data in personal health records, including: After the health records are updated, the server retrieves the health management records of the health care recipients, standardizes key information and weights it according to time, calculates individualized risk scores based on pre-set hepatobiliary disease risk scoring rules, writes the assessment results including risk level, risk value and key risk factors into the health records, updates the dynamic health index simultaneously, pushes a complete report to the doctor and a summary to the user for follow-up and popular science, and stores the key assessment information in the blockchain audit log.

Images