Hospital information integration platform interconnection system for medical whole process

Abstract

The application relates to the technical field of medical information management, and discloses a hospital information integrated platform interconnection system for a medical whole process, which performs multi-dimensional effectiveness verification of mutual recognition project directory matching, institution qualification verification and effective mutual recognition period determination through a built-in examination result mutual recognition rule engine, meanwhile, in the data access stage, whole process diagnosis and treatment data scattered in various business systems are integrated through a patient main index association algorithm, thereby replacing the traditional manual mutual recognition condition checking work link, greatly improving the mutual recognition determination efficiency and the uniformity of the determination standard, and reducing the diagnosis and treatment burden of clinical doctors; when cross-hospital reading is performed, a structured complete patient diagnosis and treatment archive can be output, doctors on duty can comprehensively master patient medical history information, and the accuracy and continuity of cross-hospital continuous diagnosis and treatment are improved.

Description

Technical Field

[0001] This invention relates to the technical field of medical information management, specifically to an interconnected hospital information integration platform system for the entire medical process. Background Technology

[0002] With the deepening of the hierarchical medical system and the construction of medical consortia, the mutual recognition of examination results across medical institutions and the cross-hospital access to electronic medical records have become core needs for improving diagnostic and treatment efficiency and reducing the burden of medical treatment for patients. Traditional hospital information integration platforms mostly focus on data integration and business collaboration within heterogeneous systems within the hospital. They face significant technical barriers when facing cross-hospital interconnection scenarios, making it difficult to adapt to the performance and functional requirements of high-frequency business scenarios and failing to fully support medical business collaboration under the hierarchical medical system. Existing technologies have several shortcomings in the implementation of cross-hospital interconnection: the cross-hospital data interaction mode is rigid and singular, and the batch push mode... There are issues with data redundancy and latency. The pure on-demand retrieval mode is prone to high concurrency congestion during peak hours, and a single mode cannot balance transmission efficiency, storage cost, and response time. The mutual recognition of inspection results lacks automated verification capabilities, and mutual recognition judgment relies entirely on manual verification, which is inefficient and has inconsistent judgment standards. The entire medical process data is fragmented, with data from different stages of diagnosis and treatment scattered across various heterogeneous business systems, making it impossible to form a complete view of patient diagnosis and treatment when accessing data across hospitals. The granularity of cross-hospital data security control is insufficient, only achieving institutional-level access control, lacking field-level data anonymization and full-link traceability capabilities, making it difficult to meet medical data security management standards.

[0003] Chinese invention patent application CN118711768A, published on September 27, 2024, discloses a medical device hospital end-to-end traceability management system, method, equipment, and storage medium. It obtains product information through a supplier self-service management platform, verifies and organizes the product information before storing it, and sends the product information to a PDA scanning terminal system. The PDA scanning terminal system obtains product information by scanning the UDI code on the product and sends the updated product information back to the UDI management system. A relay service receives UDI data from medical institutions or regulatory departments and sends the UDI data to the UDI management system, which periodically synchronizes data with the relay service. The UDI management system also connects to third-party interface services for information sharing and integration. This can help hospitals quickly establish an in-hospital end-to-end traceability system for medical devices, achieving refined on-site management of the entire lifecycle of medical devices from procurement, warehousing, delivery, use to disposal, meeting the requirements for end-to-end traceability supervision of medical devices. However, the above technical solution cannot achieve efficient and compliant cross-hospital interactive management of patient medical information. Summary of the Invention

[0004] (a) Technical problems to be solved To address the issues of rigid and singular cross-hospital data interaction models, data redundancy and latency problems in batch push models, and high concurrency congestion during peak hours in pure on-demand pull models, a single model cannot balance transmission efficiency, storage costs, and response time. Furthermore, the lack of automated verification capabilities for mutual recognition of examination results, relying entirely on manual verification, leads to low efficiency and inconsistent judgment standards. The fragmented nature of medical data across the entire process, with data from different stages of diagnosis and treatment scattered across various heterogeneous business systems, prevents cross-hospital access from forming a complete view of patient care. Insufficient granularity in cross-hospital data security control, achieving only institutional-level access control and lacking field-level data anonymization and end-to-end traceability capabilities, fails to meet medical data security management standards. This paper proposes a push-pull hybrid scheduling model to balance the timeliness of cross-hospital data transmission and storage costs. A configurable examination result mutual recognition rule engine enables multi-dimensional automated verification, unifying judgment standards and improving mutual recognition efficiency. By integrating the entire process of diagnosis and treatment data based on the patient master index, coupled with field-level anonymization to ensure privacy and security, this approach provides technical support for the implementation of mutual recognition of examination results and hierarchical medical treatment, ultimately improving the efficiency of regional medical services.

[0005] (II) Technical Solution This invention is achieved through the following technical solution: an interconnected hospital information integration platform system for the entire medical process, the system comprising: The data access and resource pool module connects to heterogeneous business systems within the hospital. It synchronously collects full-process medical business data and system operation status data through standardized business interfaces and message middleware, performs unified metadata mapping and transformation and millisecond-level timestamp alignment processing, and builds a multi-source medical data resource pool. The standardization governance and indexing module, based on medical industry data standards and medical terminology specifications, sequentially performs structured field encoding mapping, unstructured text semantic entity extraction and normalization, and standardized cleaning processing of the entire data format on the multi-source medical data resource pool to generate a standardized medical dataset, and simultaneously builds a hierarchical index directory and a data mutual recognition tag library. The hybrid scheduling and service orchestration module receives cross-hospital access requests, parses the request type, data range and permission level, locates the target data resource according to the hierarchical index directory and data mutual recognition tag library, and completes cross-hospital data interaction using a push-pull hybrid scheduling mode. The mutual recognition verification module calls the preset mutual recognition rule engine for inspection results to perform multi-dimensional validity verification on the inspection data retrieved from other institutions, including matching the mutual recognition project catalog, issuing institutional qualification verification, and determining the effective mutual recognition period. It then outputs the mutual recognition judgment result and the structured access data package.

[0006] Preferably, the process of connecting to heterogeneous business systems within the hospital, synchronously collecting full-process medical business data and system operation status data through standardized business interfaces and message middleware, performing unified metadata mapping and transformation and millisecond-level timestamp alignment processing, and constructing a multi-source medical data resource pool specifically includes: The heterogeneous business systems within the hospital that are being integrated include the electronic medical record system, the laboratory information system, the image archiving and communication system, the outpatient treatment system, and the inpatient management system. By using a patient master index association algorithm, the collected full-process medical business data and system operation status data are uniquely identified and associated. The outpatient records, examination reports, test results, inpatient medical records and prescription data of the same patient are associated and mounted in all dimensions to build a multi-source medical data resource pool with the patient as the core.

[0007] Preferably, the hospital information integration platform interconnection system for the entire medical process according to claim 1 is characterized in that: the step of performing structured field encoding mapping, unstructured text semantic entity extraction and normalization, and standardized cleaning processing of the multi-source medical data resource pool based on medical industry data standards and medical terminology specifications to generate a standardized medical dataset, and simultaneously constructing a hierarchical index directory and a data mutual recognition tag library includes the following steps: Based on national medical and health data standards and internationally accepted coding specifications, the field names, data formats, and coding systems of heterogeneous sources in the multi-source medical data resource pool are mapped and converted one by one through the field mapping rule base, so as to complete the standardized cleaning of structured data and obtain structured medical data with unified format and compliant coding. The unstructured medical record text and image report text in the multi-source medical data resource pool are extracted and normalized using a medical natural language processing model. Diagnostic terms, examination items and drug names are extracted and mapped to a standard medical terminology database to obtain semantically unified unstructured medical text data. A multi-level index directory is constructed based on three dimensions: data type, business domain, and access frequency. At the same time, mutual recognition tags are marked for examination data according to the mutual recognition directory of examination items, the qualification level of medical institutions, and the data validity period, generating a hierarchical index directory and a data mutual recognition tag library.

[0008] Preferably, the process of receiving cross-department access requests, parsing the request type, data range, and permission level, locating the target data resource based on the hierarchical index directory and data mutual recognition tag library, and completing cross-department data interaction using a push-pull hybrid scheduling mode, specifically includes: For the patient basic records and recent diagnosis and treatment summaries in the standardized medical dataset, which are low-frequency and highly reusable data, an incremental timed push strategy is adopted to synchronize the data to the regional interconnection front-end cache node, so as to realize low-latency access to high-frequency data. For the large volume of low-reuse data such as original images, complete medical records, and historical test details in the standardized medical dataset, an on-demand real-time retrieval strategy is adopted. After receiving the retrieval request, the data is extracted from the multi-source medical data resource pool and transmitted through a fragmented transmission protocol to avoid redundant storage of large volumes of data. Based on the current request concurrency and target data volume, a dynamic threshold scheduling algorithm is used to switch the trigger threshold between push and pull modes, balancing cross-department transmission latency and storage redundancy.

[0009] Preferably, the specific steps for calling the preset inspection result mutual recognition rule engine to perform multi-dimensional validity checks on the cross-institutional inspection data, including mutual recognition item catalog matching, issuing institution qualification verification, and valid mutual recognition cycle determination, and outputting the mutual recognition determination result and structured access data package are as follows: Extract the inspection item code, issuing institution level, inspection time and inspection equipment information from the data corresponding to the cross-institute access request, and use them as input parameters for mutual recognition verification; The extracted input parameters are matched and verified against the mutual recognition rule library of the inspection results built into the mutual recognition rule engine in multiple dimensions, and it is determined in turn whether the inspection item belongs to the mutual recognition catalog, whether the issuing institution has mutual recognition qualifications, and whether the inspection result is within the valid mutual recognition period. If all dimensions pass the verification, it is marked as mutually recognized. If any dimension fails the verification, it is marked as recommended for review. A structured access data package containing mutual recognition identifiers and verification basis is generated by combining the structured inspection data.

[0010] Preferably, the trigger threshold for switching push and pull modes based on the current request concurrency and target data volume using a dynamic threshold scheduling algorithm balances cross-departmental transmission latency and storage redundancy. The calculation basis for the dynamic switching trigger threshold of push and pull modes includes: When the request concurrency is lower than the set concurrency threshold and the target data volume is smaller than the set target data volume threshold, a real-time pull mode is adopted to directly retrieve data from the multi-source medical data resource pool; when the request concurrency is higher than the set concurrency threshold or the target data volume is greater than the set target data volume threshold, a pre-push caching mechanism is automatically triggered to push hot data to the regional interconnection front-end cache node.

[0011] Preferably, the step of calling a preset mutual recognition rule engine for examination results to perform multi-dimensional validity checks on the examination data retrieved across institutions, including matching the mutual recognition item catalog, verifying the issuing institution's qualifications, and determining the effective mutual recognition period, and outputting the mutual recognition determination result and structured access data package, specifically includes: The mutual recognition rule engine for inspection results supports dynamic rule configuration. It can update the mutual recognition project directory, the list of institutional qualifications, and the validity period parameters through a visual configuration interface, and generate a mutual recognition rule library for inspection results that is adapted to the latest policies.

[0012] Preferably, the steps of receiving cross-department access requests, parsing the request type, data range, and permission level, locating the target data resource according to the hierarchical index directory and data mutual recognition tag library, and completing cross-department data access using a push-pull hybrid scheduling mode specifically include: The received inter-hospital access requests include requests for mutual recognition of examination results and requests for inter-hospital access to electronic medical records; In response to cross-hospital requests for access to electronic medical records, the system uses a field-level desensitization algorithm to mask and replace the corresponding data in the multi-source medical data resource pool based on the medical record level and authorized scope before transmitting the data.

[0013] (III) Beneficial Effects This invention provides an interconnected hospital information integration platform system covering the entire medical process. It offers the following advantages: I. By constructing a hybrid data scheduling mode that integrates push and pull, an incremental timed push strategy is adopted for low-frequency, high-reusability data, and an on-demand real-time pull strategy is adopted for large-volume, low-reusability data. In addition, a dynamic threshold scheduling algorithm is used to switch modes and trigger thresholds based on request concurrency and data volume. This not only ensures low-latency response for high-frequency data access, but also avoids redundant storage and invalid transmission of large-volume cold data. It effectively solves the performance bottleneck of the traditional single transmission mode and improves the efficiency and economic efficiency of cross-department data interaction.

[0014] Second, the built-in mutual recognition rule engine for examination results performs multi-dimensional validity checks, including matching mutual recognition item catalogs, verifying the qualifications of issuing institutions, and determining the effective mutual recognition cycle. At the same time, during the data access phase, the patient master index association algorithm integrates the entire process of diagnosis and treatment data scattered across various business systems, replacing the traditional manual verification of mutual recognition conditions. This significantly improves the efficiency of mutual recognition determination and the uniformity of determination standards, and reduces the diagnostic and treatment burden on clinicians. When accessing data across hospitals, the system can output structured and complete patient medical records, supporting attending physicians to fully grasp patient medical history information and improving the accuracy and consistency of continuous diagnosis and treatment across hospitals.

[0015] Third, by using field-level desensitization algorithms to mask and replace privacy fields in accessed data such as electronic medical records, the scope of cross-hospital data access is precisely controlled, ensuring the security of patient privacy information and meeting the compliance requirements for medical data security management. The entire solution forms a complete technical chain from multi-source data access, standardized governance, hybrid scheduling and transmission to automated mutual recognition and verification, extending the capabilities of the hospital information integration platform from intra-hospital data integration to cross-hospital business collaboration. It can provide technical support for the implementation of policies on hierarchical diagnosis and treatment and mutual recognition of test results, and improve the overall efficiency of regional medical services and the patient's medical experience. Attached Figure Description

[0016] Figure 1 A flowchart of the hospital information integration platform interconnection system for the entire medical process provided by this invention; Figure 2 This is a schematic diagram of the modules of the hospital information integration platform interconnection system for the entire medical process provided by the present invention. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] An example of this hospital information integration platform interconnection system, which is designed for the entire medical process, is as follows: Example: Please see Figures 1-2 A hospital information integration platform interconnection system covering the entire medical process, the system includes: The data access and resource pool module connects to heterogeneous business systems within the hospital. It synchronously collects full-process medical business data and system operation status data through standardized business interfaces and message middleware, performs unified metadata mapping and transformation and millisecond-level timestamp alignment processing, and builds a multi-source medical data resource pool. The heterogeneous business systems within the hospital that are being integrated include the electronic medical record system, the laboratory information system, the image archiving and communication system, the outpatient treatment system, and the inpatient management system. By using a patient master index association algorithm, the collected full-process medical business data and system operation status data are uniquely identified and associated. The outpatient records, examination reports, test results, inpatient medical records and prescription data of the same patient are associated and mounted in all dimensions, forming a multi-source medical data resource pool with the patient as the core.

[0019] The hospital's information integration platform is deployed in the hospital's data center. It integrates with the hospital's electronic medical record system, laboratory information system, image archiving and communication system, outpatient treatment system, and inpatient management system via HL7v2.x, FHIRR4 standardized medical business interfaces, and RabbitMQ message middleware. For time-sensitive business data such as test results, examination reports, and outpatient prescriptions, a message queue publish-subscribe model is used for real-time data collection. After each business system generates and approves new data, it automatically pushes the data event to the message queue, and the hospital information integration platform, acting as the consumer, completes data reception within seconds. For less time-sensitive data such as inpatient medical record homepages, historical treatment files, and system operation logs, a timed batch incremental collection model is used, synchronizing only the data added and changed within the previous period, ensuring data timeliness while reducing system resource consumption. The data collection scope comprehensively covers the entire process of medical business data and system operation status data. The medical business data includes patient basic information, outpatient medical records, test results, imaging reports, inpatient medical records, and prescriptions, covering all dimensions of medical treatment content. The system operation status data includes operation indicators such as the response time of each business system interface, data throughput, and abnormal error logs.

[0020] After data collection is completed, the hospital information integration platform sequentially performs three core processes: unified metadata mapping and transformation, millisecond-level timestamp alignment, and association with the patient master index. The unified metadata mapping and transformation relies on a pre-configured metadata mapping table. This table predefines the standard field names, data types, length constraints, and enumeration value ranges for all data, mapping fields with the same meaning but different names, formats, and encodings from different vendors' business systems to a unified standard data model, eliminating field differences between heterogeneous systems. Millisecond-level timestamp alignment is achieved by accessing the National Time Service Center's NTP clock synchronization service, assigning a unified benchmark millisecond-level timestamp to all collected data, correcting local clock deviations in different business systems, and ensuring the timing consistency of all data. Based on this, data association processing is carried out through the patient master index association algorithm. The patient's ID number is used as the core matching primary key and the in-hospital visit ID is used as the auxiliary matching primary key. The data of the entire process of medical business data from different business systems and the system operation status data are uniquely identified and associated. For abnormal matching scenarios such as missing ID numbers and the same name and age, a dual processing mechanism of fuzzy matching and manual review is adopted to ensure the accuracy of matching. The outpatient records, examination reports, test results, inpatient medical records and prescription data of the same patient are associated and mounted in all dimensions.

[0021] After the above collection and processing, a multi-source medical data resource pool centered on patients is finally constructed. This resource pool uses the patient master index as the unique identifier and stores data according to four major business domains: outpatient, laboratory, imaging, and inpatient. This enables centralized and unified management of the diagnosis and treatment data of the same patient throughout the entire life cycle, serving as the sole data foundation for subsequent standardized governance and cross-hospital access.

[0022] The standardization governance and indexing module, based on medical industry data standards and medical terminology specifications, sequentially performs structured field encoding mapping, unstructured text semantic entity extraction and normalization, and standardized cleaning processing of the entire data format on the multi-source medical data resource pool to generate a standardized medical dataset, and simultaneously builds a hierarchical index directory and a data mutual recognition tag library. Based on medical industry data standards and medical terminology specifications, the multi-source medical data resource pool undergoes structured field encoding mapping, unstructured text semantic entity extraction and normalization, and standardized cleaning processing to unify the format of all data, generating a standardized medical dataset. Simultaneously, a hierarchical index directory and a data interoperability tag library are constructed, including the following steps: Based on national medical and health data standards and internationally accepted coding specifications, the field names, data formats, and coding systems from heterogeneous sources in the multi-source medical data resource pool are mapped and converted one by one through the field mapping rule base, so as to complete the standardized cleaning of structured data and obtain structured medical data with unified format and compliant coding. The unstructured medical record text and image report text in the multi-source medical data resource pool are extracted and normalized using a medical natural language processing model. Diagnostic terms, examination items and drug names are extracted and mapped to a standard medical terminology database to obtain semantically unified unstructured medical text data. A multi-level index directory is constructed based on three dimensions: data type, business domain, and access frequency. At the same time, mutual recognition tags are marked for examination data according to the mutual recognition directory of examination items, the qualification level of medical institutions, and the data validity period, generating a hierarchical index directory and a data mutual recognition tag library.

[0023] This step uses a multi-source medical data resource pool as input. Based on the medical and health dataset standards issued by the National Health Commission and internationally recognized coding standards such as LOINC, ICD-10, and ATC, a field mapping rule library is pre-built, which includes three types of rules: field name mapping, data format conversion, and coding system mapping. For all structured data in the multi-source medical data resource pool, the field mapping rule library is called to perform mapping conversion field by field: fields with different names but consistent semantics in heterogeneous systems are unified into standard field names; numeric, date, and enumerated data are uniformly converted into standard formats; and hospital-defined diagnostic codes, test item codes, and drug codes are mapped one by one to national standard codes. This completes the standardized cleaning of structured data, resulting in structured medical data with unified format and compliant coding.

[0024] For unstructured text data such as admission records, medical records, imaging diagnostic reports, and outpatient medical orders in a multi-source medical data resource pool, a medical natural language processing model trained based on the BERT architecture is invoked for processing. First, the boundaries of medical entities are identified through text segmentation and part-of-speech tagging, and five core medical entities are extracted: diagnosis name, symptom description, examination items, drug names, and surgical procedures. Then, the extracted entities are mapped to standard medical terminology databases such as the ICD-10 diagnostic terminology database, the ATC drug classification database, and the LOINC laboratory test database. This unifies the personalized writing expressions of different doctors into standard medical terms, eliminates the semantic differences caused by manual writing, and obtains semantically unified unstructured medical text data.

[0025] After completing the standardized cleaning of the entire dataset, a three-level hierarchical index directory was constructed based on three dimensions: "business domain - data type - access frequency". The first-level index is divided into four categories based on business domain: outpatient, laboratory, imaging, and inpatient. The second-level index is divided into three categories based on data type: reports, raw data, and statistical summaries. The third-level index is divided into three categories based on access frequency: high-frequency data from the past 7 days, medium-frequency data from the past 30 days, and historical low-frequency data. Each index entry is associated with the storage location and volume information of the corresponding data, achieving second-level location of the target data. Simultaneously, a data mutual recognition tag library was constructed. Using the mutual recognition project directory for examination results, the list of mutually recognized designated medical institutions, and the effective mutual recognition period for each project published by the regional health commission as tagging rules, mutual recognition tags were affixed to each piece of examination data. Tags include three attributes: "mutually recognized project / non-mutually recognized project", "mutually recognized institution / non-mutually recognized institution", and "valid / expired". The mutual recognition eligibility of data can be directly determined through tag combinations. After the above processing, three products are generated simultaneously: a standardized medical dataset, a hierarchical index directory, and a data mutual recognition tag library. These products provide a standardized data foundation and a basis for rapid retrieval for subsequent cross-hospital access and mutual recognition verification.

[0026] The hybrid scheduling and service orchestration module receives cross-department access requests, parses the request type, data range and permission level, locates the target data resource according to the hierarchical index directory and data mutual recognition tag library, and completes cross-department data interaction using a push-pull hybrid scheduling mode. This process involves receiving cross-department access requests, parsing the request type, data range, and permission level, locating the target data resource based on the hierarchical index directory and data mutual recognition tag library, and completing cross-department data interaction using a push-pull hybrid scheduling mode. The push-pull hybrid scheduling mode specifically includes: For low-frequency, highly reusable data such as patient basic records and recent diagnosis and treatment summaries in standardized medical datasets, an incremental timed push strategy is adopted to synchronize the data to the regional interconnection front-end cache node, so as to realize low-latency access to high-frequency data; For large volumes of low-reuse data such as original images, complete medical records, and historical test details in standardized medical datasets, an on-demand real-time retrieval strategy is adopted. After receiving a retrieval request, the data is selectively extracted from a multi-source medical data resource pool and transmitted through a fragmented transmission protocol to avoid redundant storage of large volumes of data. Based on the current request concurrency and target data volume, a dynamic threshold scheduling algorithm is used to switch the trigger threshold between push and pull modes, balancing cross-department transmission latency and storage redundancy.

[0027] Based on the current request concurrency and the target data volume, a dynamic threshold scheduling algorithm is used to switch the trigger threshold between push and pull modes, balancing cross-departmental transmission latency and storage redundancy. The calculation basis for the dynamic switching trigger threshold between push and pull modes includes: When the request concurrency is lower than the set concurrency threshold and the target data volume is smaller than the set target data volume threshold, the real-time pull mode is adopted to directly retrieve data from the multi-source medical data resource pool; when the request concurrency is higher than the set concurrency threshold or the target data volume is larger than the set target data volume threshold, the pre-push caching mechanism is automatically triggered to push hot data to the regional interconnection front-end cache node.

[0028] The cross-hospital access requests received include requests for mutual recognition of examination results and requests for cross-hospital access to electronic medical records; In response to cross-hospital requests for access to electronic medical records, the system uses a field-level desensitization algorithm to mask and replace privacy fields in the corresponding data from the multi-source medical data resource pool, based on the medical record level and authorized scope, before transmitting the data.

[0029] This step uses cross-hospital access requests as the trigger entry point, and sequentially completes three processing steps: request parsing, data location, and hybrid scheduling transmission. The first step is request reception and parsing. Cross-hospital access requests initiated by medical institutions within the region are first sent to the intra-hospital interconnection front-end gateway. The hospital information integration platform performs structured parsing of the request message, extracting three core types of information: request type, data range, and permission level. The request type distinguishes between requests for mutual recognition of examination results and cross-hospital access to electronic medical records. The data range includes the target patient identifier, data time range, and data type range. The permission level includes the requesting institution level, the operating physician's role and permissions, and the patient's electronic authorization status.

[0030] After completing the request parsing, the hospital information integration platform uses the patient identifier and data range obtained from the parsing to call the hierarchical index directory and the data mutual recognition tag library to carry out joint retrieval: first, it locates the full data index of the target patient through the patient master index, then filters out data entries that meet the request conditions according to data type and time range, and at the same time reads the mutual recognition tag attributes of each data entry, and finally determines the storage location, data volume, and mutual recognition attributes of the target data, thus completing the precise location of the target data resource.

[0031] Subsequently, a push-pull hybrid scheduling mode is adopted for cross-hospital data transmission. The hospital information integration platform automatically matches the optimal transmission strategy based on the type, volume, and access frequency of the target data. For small-volume, highly reusable data such as patient basic records, medical summaries of the past 7 days, and outpatient prescriptions, an incremental timed push strategy is adopted: the hospital information integration platform performs an incremental synchronization once every morning, pushing newly added and changed data of this type to the regional interconnection front-end cache node. When a request arrives, the data is directly returned from the front-end cache node without needing to retrieve it from the source, keeping the response latency within 1 second. For large-volume, low-reusable data such as original CT / MRI images, complete inpatient medical records of the past 3 years, and historical test details, an on-demand real-time pull strategy is adopted: the hospital information integration platform does not perform advance push. After receiving a request, it selectively extracts the target data from the multi-source medical data resource pool and performs cross-hospital transmission using an HTTP breakpoint resume protocol and a 4MB fixed-segment transmission mechanism. When transmission fails, the corresponding segment is automatically retried to avoid resource waste caused by retransmitting large volumes of data.

[0032] During transmission, the hospital information integration platform monitors the current concurrent access requests and target data volume in real time using a dynamic threshold scheduling algorithm, setting an initial concurrent threshold of 30 requests / second and a target data volume threshold of 50MB. When the concurrent request volume exceeds 50 requests / second during peak weekday visit periods, the algorithm automatically increases the pre-push threshold, including examination reports from the past 30 days in the pre-push scope, caching them in advance to the front-end node and reducing the pressure on the hospital's outbound bandwidth. When the concurrent access volume is below 10 requests / second during off-peak nighttime periods, the pre-push threshold is decreased, retaining only the cached basic patient records, releasing storage resources on the front-end node. This step ultimately achieves stable cross-hospital transmission of the target data, while simultaneously achieving adaptive optimization of the transmission strategy through dynamic threshold adjustment, balancing cross-hospital transmission latency and storage redundancy.

[0033] By constructing a hybrid data scheduling mode that integrates push and pull, an incremental timed push strategy is adopted for low-frequency, high-reusability data, and an on-demand real-time pull strategy is adopted for large-volume, low-reusability data. The mode is switched and the threshold is triggered by a dynamic threshold scheduling algorithm that combines request concurrency and data volume. This not only ensures low-latency response for high-frequency data access, but also avoids redundant storage and invalid transmission of large-volume cold data. It effectively solves the performance bottleneck of the traditional single transmission mode and improves the efficiency and economic efficiency of cross-department data interaction.

[0034] The mutual recognition verification module calls the preset mutual recognition rule engine for inspection results to perform multi-dimensional validity verification on the inspection data retrieved from other institutions, including matching the mutual recognition project catalog, issuing institutional qualification verification, and determining the effective mutual recognition period. It then outputs the mutual recognition judgment result and the structured access data package.

[0035] The specific steps for calling the preset inspection result mutual recognition rule engine to perform multi-dimensional validity checks on the inspection data retrieved from other institutions, including matching the mutual recognition item catalog, verifying the institution's qualifications, and determining the effective mutual recognition period, and outputting the mutual recognition determination result and structured access data package are as follows: Extract the inspection item code, issuing institution level, inspection time and inspection equipment information from the data corresponding to the cross-institute access request, and use them as input parameters for mutual recognition verification; The extracted input parameters are matched and verified against the mutual recognition rule library of the inspection results built into the inspection result mutual recognition rule engine in multiple dimensions. The engine determines whether the inspection item belongs to the mutual recognition catalog, whether the issuing institution has mutual recognition qualifications, and whether the inspection result is within the valid mutual recognition period. If all dimensions pass the verification, it is marked as mutually recognized. If any dimension fails the verification, it is marked as recommended for review. A structured access data package containing mutual recognition identifiers and verification basis is generated by combining the structured inspection data.

[0036] The mutual recognition rule engine for inspection results supports dynamic rule configuration. It can update the mutual recognition project directory, the list of institutional qualifications, and the validity period parameters through a visual configuration interface, and generate a mutual recognition rule library for inspection results that is adapted to the latest policies.

[0037] The built-in mutual recognition rule engine performs multi-dimensional validity checks, including matching mutual recognition item catalogs, verifying the qualifications of issuing institutions, and determining the effective mutual recognition cycle. At the same time, during the data access phase, the patient master index association algorithm integrates the entire process of diagnosis and treatment data scattered across various business systems, replacing the traditional manual verification of mutual recognition conditions. This significantly improves the efficiency of mutual recognition determination and the uniformity of determination standards, and reduces the diagnostic and treatment burden on clinicians. When accessing data across hospitals, the system can output structured and complete patient medical records, supporting attending physicians to fully grasp the patient's medical history information and improving the accuracy and consistency of continuous diagnosis and treatment across hospitals.

[0038] By employing field-level desensitization algorithms to mask and replace privacy fields in accessed data such as electronic medical records, the system precisely controls the scope of cross-hospital data access, ensuring the security of patient privacy information and meeting compliance requirements for medical data security management. The entire solution forms a complete technical chain from multi-source data access, standardized governance, hybrid scheduling and transmission to automated mutual recognition and verification. It extends the capabilities of the hospital information integration platform from intra-hospital data integration to cross-hospital business collaboration, providing technical support for the implementation of policies on hierarchical diagnosis and treatment and mutual recognition of test results, and comprehensively improving the efficiency of regional medical services and the patient's medical experience.

[0039] This step takes the examination data retrieved from other hospitals as the processing object and performs fully automated multi-dimensional validity verification through the examination result mutual recognition rule engine. First, the verification parameters are extracted. The hospital information integration platform automatically extracts four core parameters from the located examination data: the LOINC code of the examination item, the level of the medical institution that issued the report, the examination completion time, and the examination equipment model, as input parameters for mutual recognition verification.

[0040] The system then invokes the built-in mutual recognition rule library of the examination results mutual recognition rule engine, performing matching verification in three dimensions: The first dimension is item matching verification, comparing the extracted examination item code with the mutual recognition item directory one by one to determine if the examination item falls within the scope of mutual recognition stipulated by the region; the second dimension is institution qualification verification, comparing the issuing institution's level with the list of mutually recognized designated medical institutions to determine if the institution possesses mutual recognition qualifications; the third dimension is validity period verification, using the current access time as a benchmark, calculating the time interval between the examination completion time and the current time to determine if the examination result is within the valid mutual recognition period for that type of examination. The mutual recognition period for routine laboratory examinations is set at 7 days, for imaging examinations at 30 days, and for pathology examinations at 90 days. Verification is considered successful if all three dimensions meet the verification conditions; otherwise, verification is considered unsuccessful.

[0041] After verification, the hospital information integration platform marks the examination data with corresponding statuses: all data that passes verification is marked as "mutually recognized," and data that fails verification is marked as "re-examination recommended," with the specific reasons for failure also noted. Finally, the structured examination results, mutual recognition status identifiers, and verification details across the three dimensions are uniformly packaged into a structured access data package, and the mutual recognition judgment result is output synchronously. The examination result mutual recognition rule engine supports backend visual configuration management. When regional mutual recognition policies are updated, administrators can directly modify the mutual recognition project directory, institution qualification list, and validity period parameters in the configuration interface. After saving the configuration, the system automatically generates the latest examination result mutual recognition rule library without modifying the program code, shortening the policy adaptation cycle from weekly to hourly.

[0042] For example, consider the specific scenario where a patient is transferred from a tertiary hospital to a primary healthcare institution for a follow-up visit, requiring cross-hospital access to and mutual recognition of test results: When a patient has a follow-up visit at a primary care hospital, the attending physician initiates an application for cross-hospital examination result retrieval and mutual recognition through the primary care hospital's electronic medical record system, selecting the patient's blood routine and chest CT examination items from the past 30 days. The request is transmitted to the interconnected hospital information integration platform of the tertiary hospital. The data access module collects the patient's full-process diagnosis and treatment data, and after being associated with the patient's master index, a multi-source medical data resource pool is constructed. After field mapping and semantic normalization processing by the standardization governance module, a standardized medical dataset is generated, and a hierarchical index directory and a data mutual recognition tag library are constructed simultaneously. The hospital information integration platform quickly locates the patient's corresponding blood routine and chest CT examination data based on the hierarchical index directory and data mutual recognition tag library, confirming that both examinations are within the regional mutual recognition item directory; it adopts a push-pull hybrid scheduling mode to transmit data: for small data such as blood routine reports, it directly retrieves data from the front-end cache node with a response time of 0.8 seconds; for large data such as raw chest CT images, it triggers an on-demand pull mode, retrieving data from a multi-source medical data resource pool through fragmented transmission. The mutual recognition rule engine automatically verifies the results: if the blood routine test was conducted 12 days ago, exceeding the 7-day mutual recognition period, it is marked as "re-examination recommended"; if the chest CT scan was conducted 15 days ago, within the 30-day mutual recognition period, and the tertiary hospital is a mutual recognition designated institution, it is marked as "mutually recognized"; a structured access data package with mutual recognition identification and verification basis is generated and returned to the primary hospital. The attending physician directly refers to the mutually recognized and valid CT results to formulate a treatment plan, and only arranges for the patient to have a repeat blood routine test, which avoids repeated examinations, reduces the patient's medical costs, and improves the efficiency of diagnosis and treatment.

[0043] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A hospital information integration platform interconnection system for the entire medical process, characterized by: The system includes: The data access and resource pool module connects to heterogeneous business systems within the hospital. It synchronously collects full-process medical business data and system operation status data through standardized business interfaces and message middleware, performs unified metadata mapping and transformation and millisecond-level timestamp alignment processing, and builds a multi-source medical data resource pool. The standardization governance and indexing module, based on medical industry data standards and medical terminology specifications, sequentially performs structured field encoding mapping, unstructured text semantic entity extraction and normalization, and standardized cleaning processing of the entire data format on the multi-source medical data resource pool to generate a standardized medical dataset, and simultaneously builds a hierarchical index directory and a data mutual recognition tag library. The hybrid scheduling and service orchestration module receives cross-hospital access requests, parses the request type, data range and permission level, locates the target data resource according to the hierarchical index directory and data mutual recognition tag library, and completes cross-hospital data interaction using a push-pull hybrid scheduling mode. The mutual recognition verification module calls the preset mutual recognition rule engine for inspection results to perform multi-dimensional validity verification on the inspection data retrieved from other institutions, including matching the mutual recognition project catalog, issuing institutional qualification verification, and determining the effective mutual recognition period. It then outputs the mutual recognition judgment result and the structured access data package.

2. The hospital information integration platform interconnection system for the entire medical process as described in claim 1, characterized in that: The process of connecting to heterogeneous business systems within the hospital involves synchronously collecting end-to-end medical business data and system operation status data through standardized business interfaces and message middleware, performing unified metadata mapping and millisecond-level timestamp alignment, and constructing a multi-source medical data resource pool. Specifically, this includes: The heterogeneous business systems within the hospital that are being integrated include the electronic medical record system, the laboratory information system, the image archiving and communication system, the outpatient treatment system, and the inpatient management system. By using a patient master index association algorithm, the collected full-process medical business data and system operation status data are uniquely identified and associated. The outpatient records, examination reports, test results, inpatient medical records and prescription data of the same patient are associated and mounted in all dimensions to build a multi-source medical data resource pool with the patient as the core.

3. The hospital information integration platform interconnection system for the entire medical process as described in claim 1, characterized in that: Based on medical industry data standards and medical terminology specifications, the process of sequentially performing structured field encoding mapping, unstructured text semantic entity extraction and normalization, and standardized cleaning of the entire data format on the multi-source medical data resource pool to generate a standardized medical dataset, and simultaneously constructing a hierarchical index directory and a data mutual recognition tag library, includes the following steps: Based on national medical and health data standards and internationally accepted coding specifications, the field names, data formats, and coding systems of heterogeneous sources in the multi-source medical data resource pool are mapped and converted one by one through the field mapping rule base, so as to complete the standardized cleaning of structured data and obtain structured medical data with unified format and compliant coding. The unstructured medical record text and image report text in the multi-source medical data resource pool are extracted and normalized using a medical natural language processing model. Diagnostic terms, examination items and drug names are extracted and mapped to a standard medical terminology database to obtain semantically unified unstructured medical text data. A multi-level index directory is constructed based on three dimensions: data type, business domain, and access frequency. At the same time, mutual recognition tags are marked for examination data according to the mutual recognition directory of examination items, the qualification level of medical institutions, and the data validity period, generating a hierarchical index directory and a data mutual recognition tag library.

4. The hospital information integration platform interconnection system for the entire medical process as described in claim 1, characterized in that: The process of receiving cross-department access requests involves parsing the request type, data range, and permission level; locating the target data resource based on the hierarchical index directory and data mutual recognition tag library; and completing cross-department data interaction using a push-pull hybrid scheduling mode. This push-pull hybrid scheduling mode specifically includes: For the patient basic records and recent diagnosis and treatment summaries in the standardized medical dataset, which are low-frequency and highly reusable data, an incremental timed push strategy is adopted to synchronize the data to the regional interconnection front-end cache node, so as to realize low-latency access to high-frequency data. For the large volume of low-reuse data such as original images, complete medical records, and historical test details in the standardized medical dataset, an on-demand real-time retrieval strategy is adopted. After receiving the retrieval request, the data is extracted from the multi-source medical data resource pool and transmitted through a fragmented transmission protocol to avoid redundant storage of large volumes of data. Based on the current request concurrency and target data volume, a dynamic threshold scheduling algorithm is used to switch the trigger threshold between push and pull modes, balancing cross-department transmission latency and storage redundancy.

5. The hospital information integration platform interconnection system for the entire medical process as described in claim 1, characterized in that: The specific steps for calling the preset inspection result mutual recognition rule engine to perform multi-dimensional validity checks on the inspection data retrieved from other institutions, including matching the mutual recognition item catalog, verifying the issuing institution's qualifications, and determining the effective mutual recognition period, and outputting the mutual recognition determination result and structured access data package are as follows: Extract the inspection item code, issuing institution level, inspection time and inspection equipment information from the data corresponding to the cross-institute access request, and use them as input parameters for mutual recognition verification; The extracted input parameters are matched and verified against the mutual recognition rule library of the inspection results built into the mutual recognition rule engine in multiple dimensions, and it is determined in turn whether the inspection item belongs to the mutual recognition catalog, whether the issuing institution has mutual recognition qualifications, and whether the inspection result is within the valid mutual recognition period. If all dimensions pass the verification, it is marked as mutually recognized. If any dimension fails the verification, it is marked as recommended for review. A structured access data package containing mutual recognition identifiers and verification basis is generated by combining the structured inspection data.

6. The hospital information integration platform interconnection system for the entire medical process as described in claim 4, characterized in that: The trigger threshold for switching push and pull modes based on the current request concurrency and target data volume using a dynamic threshold scheduling algorithm balances cross-departmental transmission latency and storage redundancy. The calculation basis for the dynamic switching trigger threshold of push and pull modes includes: When the request concurrency is lower than the set concurrency threshold and the target data volume is smaller than the set target data volume threshold, a real-time pull mode is adopted to directly retrieve data from the multi-source medical data resource pool; when the request concurrency is higher than the set concurrency threshold or the target data volume is greater than the set target data volume threshold, a pre-push caching mechanism is automatically triggered to push hot data to the regional interconnection front-end cache node.

7. The hospital information integration platform interconnection system for the entire medical process as described in claim 5, characterized in that: The aforementioned invocation of the preset inspection result mutual recognition rule engine performs multi-dimensional validity checks on the inspection data retrieved from other institutions, including matching the mutual recognition item catalog, verifying the issuing institution's qualifications, and determining the effective mutual recognition period. The output of the mutual recognition determination result and the structured access data package specifically includes: The mutual recognition rule engine for inspection results supports dynamic rule configuration. It can update the mutual recognition project directory, the list of institutional qualifications, and the validity period parameters through a visual configuration interface, and generate a mutual recognition rule library for inspection results that is adapted to the latest policies.

8. The hospital information integration platform interconnection system for the entire medical process as described in claim 1, characterized in that: The process of receiving cross-department access requests, parsing the request type, data range, and permission level, locating the target data resource based on the hierarchical index directory and data mutual recognition tag library, and completing the cross-department data access process using a push-pull hybrid scheduling mode specifically includes: The received inter-hospital access requests include requests for mutual recognition of examination results and requests for inter-hospital access to electronic medical records; In response to cross-hospital requests for access to electronic medical records, the system uses a field-level desensitization algorithm to mask and replace the corresponding data in the multi-source medical data resource pool based on the medical record level and authorized scope before transmitting the data.

Images