Three-level medical network intelligent referral method and device

Abstract

The invention relates to the technical field of intelligent medical treatment, and discloses a three-level medical network intelligent referral method and equipment, and the method comprises the steps: collecting patient diagnosis and treatment information and medical institution resource information, and converting the information into structured data in a unified format; inputting the structured data into a preset referral decision model, carrying out correlation analysis on the illness condition severity feature of the patient and the medical institution reception capability feature, and generating a referral necessity evaluation result; when the referral necessity evaluation result indicates that referral is needed, performing comprehensive scoring on candidate receiving mechanisms in the structured data based on a multi-dimensional matching rule, screening out a target receiving mechanism according to a scoring result, and generating a referral suggestion scheme; and analyzing the referral suggestion scheme to construct a referral execution instruction, planning a transfer path in response to the referral execution instruction, and locking medical resources of the target receiving mechanism by using an intelligent contract mechanism to obtain a referral result. The referral efficiency of the medical network is improved.

Description

Technical Field

[0001] This invention relates to the field of intelligent medical technology, and more specifically, to a three-tiered medical network intelligent referral method and equipment. Background Technology

[0002] The tiered medical system, a core measure in my country's deepening reform of the medical and health system, aims to achieve the rational allocation of medical resources and the scientific stratification of medical treatment order by clarifying the division of responsibilities among primary, secondary, and tertiary medical institutions and prioritizing patients based on the severity of their conditions. In recent years, with the continuous and in-depth application of digital, networked, and intelligent technologies in the medical and health field, the standardization of information and the construction of public service information platforms have gradually accelerated, providing a technological foundation for collaborative development among medical institutions at different levels. However, in actual cross-institutional business collaboration, how to implement efficient and safe referrals remains a crucial issue.

[0003] Current medical referral technologies largely rely on human experience for judgment and communication, or are based solely on simple information systems for point-to-point communication, lacking systematic and intelligent assistance. Specifically, data silos still exist between medical institutions, hindering the effective flow of medical records and examination results, leading to inaccurate information transmission. Referral route planning primarily depends on the subjective experience of medical staff, lacking overall coordination and intelligent matching of resources such as ambulance dispatch, bed reservation, and doctor availability within the region. Furthermore, the existing model lacks a full-process tracking, feedback, and closed-loop optimization mechanism, making it impossible to effectively evaluate and analyze the referral process. These problems result in poor patient referral matching, delayed resource allocation, and cumbersome procedures, thus leading to low referral efficiency in the existing three-tiered medical network. Summary of the Invention

[0004] To overcome the shortcomings of low referral efficiency in existing technologies, the present invention proposes the following technical solution: Firstly, this invention proposes an intelligent referral method for a three-tiered medical network, comprising: Collect multi-source heterogeneous patient diagnosis and treatment information and medical institution resource information within the regional medical network, and convert the multi-source heterogeneous patient diagnosis and treatment information and medical institution resource information into structured data in a unified format through a standardized interface protocol; The structured data is input into a pre-set referral decision model to perform a correlation analysis between the severity characteristics of the patient's condition and the characteristics of the medical institution's capacity to receive patients, and to generate a referral necessity assessment result. When the referral necessity assessment result indicates that a referral is required, the candidate receiving institutions in the structured data are comprehensively scored based on multidimensional matching rules, and the target receiving institution is selected and a referral recommendation plan is generated based on the scoring results. The referral suggestion scheme is parsed to construct a referral execution instruction. In response to the referral execution instruction, a transfer route is planned, and the medical resources of the target receiving institution are locked using a smart contract mechanism to obtain the referral result.

[0005] As a preferred technical solution, the step of converting the multi-source heterogeneous patient diagnosis and treatment information and medical institution resource information into structured data in a unified format includes: Configure a standardized interface protocol to access raw business data from medical institutions at different levels, and use a data cleaning engine to remove redundant fields from the raw business data; The cleaned data is converted into intermediate data that conforms to the general health information exchange standard through mapping rules; The intermediate data is parsed and reconstructed into a key-value pair-based object model, and the object model is written into the encrypted storage space as the structured data.

[0006] As a preferred technical solution, before inputting the structured data into a pre-set referral decision model, the method further includes constructing a referral decision model based on a graph network architecture, including: Construct a regional medical network topology, mapping each medical institution to a network node, and assigning the network node attributes including institution level, department configuration, and personnel load; Establish edges connecting the network nodes to represent referral paths, and assign the edges attribute features including geographical distance and historical referral success rate; The attribute features are input into the graph network architecture for feature learning, and a quantitative indicator representing the real-time admission capability of each medical institution is output.

[0007] As a preferred technical solution, the correlation analysis between the severity characteristics of the patient's condition and the characteristics of the medical institution's capacity to receive patients includes: Obtain parameters such as the percentage of available beds, the workload ratio of medical staff, and the availability of key specialized equipment from the target institution. Each parameter is assigned a corresponding weighting coefficient to the vacancy rate parameter, the medical staff load ratio parameter, and the availability parameter of the key specialized equipment. A weighted summation operation is performed on each parameter after weighting to obtain a comprehensive score that characterizes the medical institution's capacity to receive patients, and the comprehensive score is used as a feature of the medical institution's capacity to receive patients.

[0008] As a preferred technical solution, the generation of referral necessity assessment results includes: Extract patient vital sign data to generate a disease severity score, and obtain a comprehensive score for the treatment acceptance capacity and a bed resource score; Based on the preset decision-making logic, influencing factors are set for the severity score of the illness, the comprehensive score, and the bed resource score; The referral necessity score is obtained by linearly weighting the above scores based on the influencing factors, and then compared with a preset multi-level threshold range. The referral level is determined based on the range in which the comparison falls.

[0009] As a preferred technical solution, the step of comprehensively scoring the candidate receiving institutions in the structured data based on multi-dimensional matching rules includes: Construct a multi-objective optimization function that includes variables such as expected referral waiting time, medical capacity matching degree, navigation distance, and medical insurance reimbursement compatibility. Assign corresponding priority weights to each variable in the multi-objective optimization function; The data of the candidate receiving institutions are fed into the multi-objective optimization function for calculation. Based on the calculation results, all candidate receiving institutions are sorted, and the institution with the highest ranking is selected as the recommended object.

[0010] As a preferred technical solution, the step of planning a transport route in response to the referral execution instruction includes: Acquire real-time travel time data and traffic congestion index of the road network, and analyze the patient's vital sign data to generate a disease tolerance coefficient; A weighted path graph is constructed using the real-time passage time data and the disease tolerance coefficient, wherein the weight of the path edge is determined by the product of the passage time and the risk coefficient; The shortest path search algorithm is executed in the weighted path graph, and the route with the minimum overall cost is output as the transfer path.

[0011] As a preferred technical solution, the method of using a smart contract mechanism to lock the medical resources of the target receiving institution includes: We use natural language processing models to parse the clinical text in the referral suggestion plan and extract key disease features and corresponding resource requirements. Based on the resource requirements, a smart contract on the distributed ledger is triggered, and the smart contract queries the time period status of the corresponding resources of the target receiving institution. When it is confirmed that the time period is idle, the status marker of the resource corresponding to the time period on the distributed ledger is changed to occupied.

[0012] As a preferred technical solution, updating the parameters of the referral decision model and the multidimensional matching rules using the actual execution status data and the usage status data includes: Record the actual timestamps of key nodes in the entire referral process, and calculate the actual time consumption and resource load of each step; An anomaly detection algorithm is invoked to compare the actual time consumption with a preset benchmark model, and anomalies where the deviation exceeds a threshold are identified. Extract the deviation feature data of the abnormal link, and use the deviation feature data to reverse correct the weight coefficients in the referral decision model.

[0013] In a second aspect, the present invention also proposes an electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to perform the operations performed by the three-tier medical network intelligent referral method as described in any of the embodiments of the first aspect.

[0014] The beneficial effects of the present invention include at least the following: This invention first breaks down information barriers between different levels of medical institutions by collecting multi-source heterogeneous data within a region and converting it into structured data in a unified format using a standardized interface, thus laying a data foundation for intelligent processing. Second, it uses a referral decision-making model to correlate patient conditions with reception capacity, objectively assessing the necessity of referrals and accurately selecting target institutions using multi-dimensional matching rules, effectively avoiding the blind referrals and supply-demand mismatches caused by traditional manual experience-based judgments. Finally, it automatically plans transport routes and uses smart contract mechanisms to lock in medical resources, ensuring the timeliness of the referral process and resource ownership confirmation, eliminating the lag and conflicts caused by manual resource coordination. In summary, this method achieves fully automated collaboration from information sharing and precise decision-making to resource locking, thereby significantly improving the referral efficiency within a three-tiered medical network. Attached Figure Description

[0015] Figure 1 This is a flowchart illustrating the intelligent referral method for a three-tiered medical network provided in an embodiment of the present invention.

[0016] Figure 2 This is an architectural diagram of the three-tiered medical network intelligent referral device provided in an embodiment of the present invention.

[0017] Figure 3 This is a schematic diagram of the data interaction system provided in an embodiment of the present invention.

[0018] Figure 4 This is a schematic diagram of the referral decision system provided in an embodiment of the present invention.

[0019] Figure 5 This is a schematic diagram of the referral management system provided in an embodiment of the present invention.

[0020] Figure 6 This is a schematic diagram of the structure of the electronic device provided in an embodiment of the present invention. Detailed Implementation

[0021] The embodiments of the present invention will be described below with reference to the accompanying drawings and preferred technical solutions. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. It should be understood that the preferred technical solutions are only for illustrating the present invention and not for limiting the scope of protection of the present invention.

[0022] It should be noted that the illustrations provided in the following embodiments are only schematic representations of the basic concept of the present invention. Therefore, the drawings only show the components related to the present invention and are not drawn according to the actual number, shape and size of the components in the actual implementation. In the actual implementation, the form, quantity and proportion of each component can be arbitrarily changed, and the layout of the components may also be more complex.

[0023] In the following description, numerous details are explored to provide a more thorough explanation of embodiments of the invention. However, it will be apparent to those skilled in the art that embodiments of the invention may be practiced without these specific details. In other embodiments, well-known structures and devices are shown in block diagram form rather than in detail to avoid obscuring embodiments of the invention.

[0024] Example 1 This embodiment proposes a three-tiered medical network intelligent referral method, such as... Figure 1 As shown, Figure 1 This is a flowchart illustrating a three-tiered medical network intelligent referral method provided in this embodiment. The method includes the following steps: S1: Collect multi-source heterogeneous patient diagnosis and treatment information and medical institution resource information within the regional medical network, and convert the multi-source heterogeneous patient diagnosis and treatment information and medical institution resource information into structured data in a unified format through a standardized interface protocol; S2: Input the structured data into a pre-set referral decision model, perform correlation analysis between the characteristics of the severity of the patient's condition and the characteristics of the medical institution's capacity to receive patients, and generate a referral necessity assessment result; S3: When the referral necessity assessment result indicates that a referral is required, the candidate receiving institutions in the structured data are comprehensively scored based on the multidimensional matching rules, the target receiving institution is selected according to the scoring results, and a referral suggestion plan is generated. S4: Parse the referral suggestion to construct a referral execution instruction, plan the transfer route in response to the referral execution instruction, and use the smart contract mechanism to lock the medical resources of the target receiving institution to obtain the referral result.

[0025] Understandably, by collecting heterogeneous data from multiple sources within the region and converting it into structured data in a unified format using standardized interfaces, information barriers between medical institutions at different levels are broken down, laying a data foundation for intelligent processing. Secondly, by using a referral decision-making model to correlate patient conditions with reception capacity, the necessity of referrals is objectively assessed, and multi-dimensional matching rules are used to accurately screen target institutions, effectively avoiding the blind referrals and supply-demand mismatches caused by traditional manual experience-based judgments. Finally, by automatically planning transfer routes and using smart contract mechanisms to lock in medical resources, the timeliness of the referral process and the confirmation of resource ownership are ensured, eliminating the lag and conflicts caused by manual resource coordination. In summary, this method achieves fully automated collaboration from information sharing and accurate decision-making to resource locking, thereby significantly improving the referral efficiency within the three-tiered medical network.

[0026] Example 2 This embodiment is an improvement on the three-tiered medical network intelligent referral method proposed in Embodiment 1.

[0027] In this embodiment, converting the multi-source heterogeneous patient diagnosis and treatment information and medical institution resource information into structured data in a unified format includes: Configure a standardized interface protocol to access raw business data from medical institutions at different levels, and use a data cleaning engine to remove redundant fields from the raw business data; The cleaned data is converted into intermediate data that conforms to the general health information exchange standard through mapping rules; The intermediate data is parsed and reconstructed into a key-value pair-based object model, and the object model is written into the encrypted storage space as the structured data.

[0028] It should be noted that in the basic data construction process of this embodiment, a standardized interface protocol is used to connect with the HIS systems of medical institutions at all levels. The built-in standardization engine uniformly converts the raw data into an intermediate format conforming to the HealthLevelSeven (HL7) standard, and finally stores it in structured JSON format, thereby eliminating data barriers between different institutions. At the data transmission level, AES symmetric encryption is used to process sensitive data involving patient privacy. In addition, to support subsequent intelligent decision-making, a graph neural network (GNN) is used to perform topological modeling of the resource distribution among medical institutions: static attributes such as institution level, department configuration, and reception capacity are mapped to node features, and dynamic attributes such as geographical distance and historical referral success rate are mapped to edge weights. This graph structure construction provides a digital foundation for subsequent algorithms to accurately assess the collaborative capabilities between institutions.

[0029] Understandably, the above implementation methods effectively solve the long-standing problem of data silos in regional medical networks. The dual standardization of HL7 and JSON ensures data readability and consistency across institutions and systems, enabling higher-level hospitals to read medical records from primary care hospitals without loss of data. The application of AES encryption technology ensures high-speed flow of medical data while meeting stringent privacy protection compliance requirements. Furthermore, topology modeling based on graph neural networks transforms the complex distribution of medical resources in the physical world into a computer-understandable mathematical model. This transforms the system from a simple information transporter into one capable of perceiving the overall situation of the regional medical network, laying a solid foundation for optimal global resource scheduling.

[0030] In this embodiment, before inputting the structured data into a preset referral decision model, the method further includes constructing a referral decision model based on a graph network architecture, including: Construct a regional medical network topology, mapping each medical institution to a network node, and assigning the network node attributes including institution level, department configuration, and personnel load; Establish edges connecting the network nodes to represent referral paths, and assign the edges attribute features including geographical distance and historical referral success rate; The attribute features are input into the graph network architecture for feature learning, and a quantitative indicator representing the real-time admission capability of each medical institution is output.

[0031] It should be noted that this embodiment digitizes physical-world medical resources through a graph neural network architecture, mapping medical institutions as nodes and assigning them static and dynamic attributes such as institution level, department configuration, and staff load; and mapping referral paths as edges and assigning them interactive attributes such as geographical distance and historical referral success rate. Through feature learning of these nodes and edges, the model outputs quantitative indicators that are not merely single-dimensional scores, but rather a comprehensive representation of the overall acceptance capacity that integrates the entire network topology.

[0032] Understandably, by transforming heterogeneous data into a unified structured object model, it ensures that higher-level hospitals can interpret the diagnostic and treatment information of primary care hospitals without loss. At the same time, by using graph neural networks to replace traditional linear rule matching, the system can capture complex implicit relationships between medical institutions (for example, two hospitals that are far apart but have specific specialty green channels), thus providing a computational foundation with a global perspective and deep correlation analysis capabilities for subsequent accurate referrals.

[0033] In this embodiment, the correlation analysis between the severity characteristics of the patient's condition and the characteristics of the medical institution's capacity to receive patients includes: Obtain parameters such as the percentage of available beds, the workload ratio of medical staff, and the availability of key specialized equipment from the target institution. Each parameter is assigned a corresponding weighting coefficient to the vacancy rate parameter, the medical staff load ratio parameter, and the availability parameter of the key specialized equipment. A weighted summation operation is performed on each parameter after weighting to obtain a comprehensive score that characterizes the medical institution's capacity to receive patients, and the comprehensive score is used as a feature of the medical institution's capacity to receive patients.

[0034] It should be noted that this embodiment employs a multi-level parameter weighting approach to quantify complex medical decisions. First, the system acquires three core indicators from the target institution: the percentage of available beds, the staff workload ratio, and the availability of key specialized equipment. These are then linearly weighted using specific weighting coefficients to calculate a comprehensive score characterizing the institution's capacity to receive patients. Subsequently, when generating the referral necessity assessment result, the system converts the patient's vital signs into a severity score. This score, combined with the aforementioned comprehensive score and bed resource score, is then weighted again using influencing factors to obtain the final referral necessity score. The system incorporates multi-level threshold intervals (e.g., no referral required, referral recommended, mandatory referral). By determining whether the calculated score falls within the corresponding interval, an objective referral level is output.

[0035] In this embodiment, generating the referral necessity assessment result includes: Extract patient vital sign data to generate a disease severity score, and obtain a comprehensive score for the treatment acceptance capacity and a bed resource score; Based on the preset decision-making logic, influencing factors are set for the severity score of the illness, the comprehensive score, and the bed resource score; The referral necessity score is obtained by linearly weighting the above scores based on the influencing factors, and then compared with a preset multi-level threshold range. The referral level is determined based on the range in which the comparison falls.

[0036] It should be noted that the XGBoost model is first used to process structured data such as the patient's vital signs and past medical history to predict the severity of the disease.

[0037] To quantify the capacity of receiving institutions, a comprehensive score for their medical admission capabilities is calculated. : , in , , These are preset weight parameters, each corresponding to the importance of different resources; The percentage of available beds in the target institution; An inverse indicator representing the workload ratio of medical staff (i.e., the lower the workload, the higher the value). This represents the availability score of key equipment for the specialty (threshold is 0-10).

[0038] Based on this, a referral necessity score is further calculated. R：

[0039] in The severity score of the disease generated by XGBoost (threshold 0-10). The above-calculated diagnostic and treatment capability score (threshold 0-10) is used. Assess bed resources by rating them (threshold 0-10). R The values ​​are divided into three ranges: 0-3 points indicate no need for referral, 4-6 points indicate a recommendation for referral, and 7-10 points indicate a strong recommendation for immediate referral. These values ​​serve as the core basis for decision support.

[0040] Understandably, this multi-parameter weighted quantitative assessment mechanism changes the traditional subjective model of referrals that relies on doctors' personal experience. By calculating a comprehensive score of treatment capacity, it can capture the true capacity of target hospitals in real time, avoiding the transfer of patients to hospitals with full beds or insufficient equipment. The calculation of the referral necessity score, through a scientific weighting (0.5 / 0.3 / 0.2), prioritizes the urgency of the patient's condition while also considering the accessibility of medical resources. This tiered scoring mechanism (not necessary / recommended / strongly recommended) not only provides doctors with clear action guidelines but also effectively filters unnecessary referral requests, thereby alleviating the treatment pressure on higher-level hospitals and improving the efficiency of medical resource utilization.

[0041] In this embodiment, the comprehensive scoring of candidate receiving institutions in the structured data based on multidimensional matching rules includes: Construct a multi-objective optimization function that includes variables such as expected referral waiting time, medical capacity matching degree, navigation distance, and medical insurance reimbursement compatibility. Assign corresponding priority weights to each variable in the multi-objective optimization function; The data of the candidate receiving institutions are fed into the multi-objective optimization function for calculation. Based on the calculation results, all candidate receiving institutions are sorted, and the institution with the highest ranking is selected as the recommended object.

[0042] In this embodiment, the step of planning a transport route in response to the referral execution command includes: Acquire real-time travel time data and traffic congestion index of the road network, and analyze the patient's vital sign data to generate a disease tolerance coefficient; A weighted path graph is constructed using the real-time passage time data and the disease tolerance coefficient, wherein the weight of the path edge is determined by the product of the passage time and the risk coefficient; The shortest path search algorithm is executed in the weighted path graph, and the route with the minimum overall cost is output as the transfer path.

[0043] It should be noted that once a referral is determined, the system needs to address the questions of where and how to transfer the patient. When selecting the optimal receiving institution, a multi-objective optimization algorithm is used to construct a function. Taking all factors into consideration, including Estimated referral waiting time (hours). Score the medical capability matching degree (0-10). Navigation distance (kilometers) Score the suitability for medical insurance reimbursement (0-10). to These are the weighting coefficients for each element.

[0044] At the same time, calculate the final recommendation index. :

[0045] in For the workload score of the specialty, Historical reception records are scored. During the transfer execution phase, a modified Dijkstra algorithm is used for route planning, incorporating high-risk disease features extracted from the BERT model. This algorithm models route optimization as a weighted graph search, where the weight of each edge is determined by the traffic time score. (Based on congestion index) and disease tolerance coefficient (Based on the criticality index) jointly determined that the objective function is set as the shortest time. Risk factor is used to plan a route that balances timeliness and safety.

[0046] Understandably, this embodiment achieves personalized customization of referral plans by introducing a multi-objective optimization function and a recommendation index. The algorithm not only considers the matching degree at the medical level but also humanely incorporates the patient's economic and time costs, thereby greatly improving patient compliance and satisfaction with referral arrangements. Furthermore, incorporating the disease tolerance coefficient into the right-of-way calculation during the route planning stage is a highly innovative design. This means that for critically ill patients, the system will sacrifice distance costs to take more accessible routes to gain valuable time for rescue, while for routine referrals, the focus is on cost-effectiveness, achieving a dynamic balance between life safety and transport efficiency.

[0047] In this embodiment, the step of using a smart contract mechanism to lock the medical resources of the target receiving institution includes: We use natural language processing models to parse the clinical text in the referral suggestion plan and extract key disease features and corresponding resource requirements. Based on the resource requirements, a smart contract on the distributed ledger is triggered, and the smart contract queries the time period status of the corresponding resources of the target receiving institution. When it is confirmed that the time period is idle, the status marker of the resource corresponding to the time period on the distributed ledger is changed to occupied.

[0048] It should be noted that, to ensure efficient execution of referrals and continuous system evolution, a blockchain-based smart contract and monitoring / early warning module has been integrated. After a referral application generates a handover form, the smart contract is automatically triggered to attempt to lock the target institution's time-slot resources (such as beds and operating rooms), leveraging the immutability of blockchain to prevent resource duplication. Simultaneously, a time-series prediction model is built using a Long Short-Term Memory (LSTM) network to predict the time consumption of each step; an isolated forest algorithm is used for anomaly detection. Once anomalies such as response delays or information flow interruptions are identified, an early warning is immediately triggered and on-chain evidence is stored. The analysis and feedback system periodically collects full-process operational data (such as application time and transfer path), identifies bottlenecks, and uses anomaly characteristics to correct the weight parameters in the decision-making model, achieving strategy iteration.

[0049] The aforementioned intelligent referral method for a three-tiered medical network can be understood as follows: the application of smart contracts resolves common issues in traditional referrals such as bed conflicts and buck-passing, establishing a contractual spirit for resource allocation through technological means. Furthermore, the monitoring system combining LSTM and Isolation Forest allows managers to gain real-time insights into process bottlenecks from a holistic perspective, shifting from passive post-event accountability to proactive in-process intervention. More importantly, data-driven feedback and automatic parameter correction enable this intelligent referral method to possess self-learning capabilities. With the passage of time and the accumulation of data, its referral recommendations will become increasingly accurate, and resource allocation will become increasingly efficient, ultimately driving the regional hierarchical medical system towards intelligence and self-adaptability.

[0050] In this embodiment, updating the parameters of the referral decision model and the multidimensional matching rules using the actual execution status data and the usage status data includes: Record the actual timestamps of key nodes in the entire referral process, and calculate the actual time consumption and resource load of each step; An anomaly detection algorithm is invoked to compare the actual time consumption with a preset benchmark model, and anomalies where the deviation exceeds a threshold are identified. Extract the deviation feature data of the abnormal link, and use the deviation feature data to reverse correct the weight coefficients in the referral decision model.

[0051] It's important to note that the system utilizes Natural Language Processing (NLP) models to parse the clinical text in referral recommendations, extracting key resource requirements (such as ICU beds and ambulances) and triggering smart contracts on a distributed ledger. These contracts can query the target institution's resource status over real-time, and once an idle resource is confirmed, its status is immediately changed to occupied, creating an immutable on-chain credential. Furthermore, the system records the actual timestamps of each key node throughout the process, calculating actual time consumption and resource load. By comparing the system with a pre-defined benchmark model using anomaly detection algorithms (such as the Isolation Forest algorithm), once an anomaly exceeding a threshold is identified (e.g., a hospital experiencing prolonged response timeouts), the system automatically extracts deviation feature data and reverses the weighting coefficients in the referral decision model (e.g., reducing the hospital's recommendation weight).

[0052] Example 3 like Figures 2-5 As shown, this embodiment proposes a three-tiered medical network intelligent referral device, which is applied to a regional medical service network consisting of primary medical institutions 100, secondary medical institutions 200 and tertiary medical institutions 300. The device includes a data interaction system 400, a referral decision system 500, a referral management system 600 and an analysis and feedback system 700.

[0053] The data interaction system 400 serves as the underlying data support unit, used to achieve cross-organizational information exchange and standardized processing, such as... Figure 3 As shown, the system specifically includes an information acquisition module 401, an algorithm analysis module 402, an information storage module 403, and an information transmission module 404. The information acquisition module 401 is configured to connect to hospital information systems of different levels of medical institutions via a standard interface protocol, capturing data including basic patient characteristics, clinical records, and real-time resource status data of each institution. The algorithm analysis module 402 primarily undertakes data cleaning and standardization tasks. It integrates a data cleaning engine and format conversion rules, capable of removing redundant fields from the original business data and mapping heterogeneous data into intermediate data conforming to general health information exchange standards, thereby reconstructing a structured object model. The information storage module 403 uses encryption technology to persistently store the processed structured data and provides fast retrieval services. The information transmission module 404 is responsible for establishing secure data transmission channels between various system components and medical institution nodes, utilizing intelligent routing technology to ensure efficient data packet flow in the network topology.

[0054] The referral decision system 500, as the core computing unit, performs intelligent analysis and matching based on the aggregated data, such as... Figure 4As shown, the system specifically includes a data analysis module 501, a communication and coordination module 502, and a monitoring and early warning module 503. The data analysis module 501 internally deploys a pre-trained graph neural network model and a multi-objective optimization algorithm. It acquires parameters such as the percentage of available beds, the workload ratio of medical staff, and the availability of key specialized equipment at the target institution, assigns corresponding weight coefficients to these parameters, and performs a weighted summation operation to obtain a quantitative indicator representing the treatment acceptance capacity. Simultaneously, this module extracts patient vital sign data to generate a disease severity score, and performs correlation analysis between the disease score and the acceptance capacity indicator according to preset logic, thereby outputting a referral necessity assessment result and a recommendation for the optimal matching institution. The communication and coordination module 502, based on natural language processing technology, transforms complex referral suggestions into standardized clinical communication text and automatically pushes it to the terminals of relevant medical staff. The monitoring and early warning module 503 tracks the status of the referral process in real time. Once an abnormal situation such as a response timeout or data stream interruption is detected, an early warning mechanism is immediately triggered and abnormal node data is recorded.

[0055] The referral management system 600, as a business execution unit, is responsible for implementing the specific operations of the referral process, such as... Figure 5 As shown, the system specifically includes a referral application module 601, a referral approval module 602, and a referral execution module 603. The referral application module 601 automatically parses the patient's electronic medical record information, fills in key fields in the referral application form, and supports multimodal input completion. The referral approval module 602 intelligently matches the approval level based on the severity of the patient's condition, automatically triggering the green channel logic for applications meeting the criticality criteria. The referral execution module 603 is the control center for physical scheduling. It obtains real-time travel time and traffic congestion index of the road network, analyzes the patient's vital signs to generate a disease tolerance coefficient, constructs a weighted path graph using real-time travel time and the disease tolerance coefficient, and executes a shortest path search algorithm in the path graph to plan the transfer route. Furthermore, this module also uses a smart contract mechanism to interact with the resource management system of the target receiving institution, changing the status flag on the distributed ledger when resources are confirmed to be idle, thereby locking in bed or equipment resources.

[0056] The analysis and feedback system 700, acting as the system's self-evolving unit, is bidirectionally connected to the three systems mentioned above. This system continuously collects operational data throughout the entire referral process, including actual time consumption, resource utilization, and referral success rate at each stage. It uses statistical analysis and anomaly detection algorithms to identify bottlenecks in the process. The system extracts deviation characteristic data from abnormal stages and feeds this data back to the data analysis module 501 of the referral decision system 500. This data is used to reverse-correct the weight coefficients and parameters of the multi-objective optimization function in the decision model, thereby achieving self-optimization and performance iteration of the entire device during operation.

[0057] It should be noted that the foregoing explanation of the embodiment of the intelligent referral method for a three-tiered medical network also applies to the intelligent referral device for a three-tiered medical network in this embodiment, and will not be repeated here.

[0058] Example 4 Figure 6 This is a schematic diagram of the structure of the electronic device 800 provided in this embodiment. The electronic device 800 includes: a memory 801, a processor 802, and a computer program stored in the memory 801 and executable on the processor 802.

[0059] When the processor 802 executes the program, it implements the three-level medical network intelligent referral method provided in the above embodiments.

[0060] Furthermore, the electronic device 800 also includes a communication interface 803 for communication between the memory 801 and the processor 802.

[0061] The memory 801 may include high-speed RAM (Random Access Memory) memory, and may also include non-volatile memory, such as at least one disk storage.

[0062] If the memory 801, processor 802, and communication interface 803 are implemented independently, then the communication interface 803, memory 801, and processor 802 can be interconnected via a bus to complete communication between them. The bus can be an ISA (Industry Standard Architecture) bus, a PCI (Peripheral Component Interconnect) bus, or an EISA (Extended Industry Standard Architecture) bus, etc. The bus can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 6 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0063] Optionally, in a specific implementation, if the memory 801, processor 802, and communication interface 803 are integrated on a single chip, then the memory 801, processor 802, and communication interface 803 can communicate with each other through an internal interface.

[0064] The processor 802 may be a CPU (Central Processing Unit), an ASIC (Application Specific Integrated Circuit), or one or more integrated circuits configured to implement embodiments of the present invention.

[0065] This invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described intelligent referral method for a three-tiered medical network.

[0066] In the description of this specification, the references to the terms "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0067] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this invention, "N" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0068] Any process or method description in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or more N executable instructions for implementing custom logic functions or processes, and the scope of preferred embodiments of the invention includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functions involved, as should be understood by those skilled in the art to which embodiments of the invention pertain.

[0069] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, the N steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware as in another embodiment, it can be implemented using any of the following techniques known in the art, or a combination thereof: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (FPGAs), field-programmable gate arrays (FPGAs), etc.

[0070] Those skilled in the art will understand that all or part of the steps of the methods described in the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, the program includes one or a combination of the steps of the method embodiments.

[0071] Obviously, the above embodiments of the present invention are merely examples for clearly illustrating the present invention, and are not intended to limit the implementation of the present invention. Those skilled in the art will recognize that other variations or modifications can be made based on the above description. It is neither necessary nor possible to exhaustively describe all embodiments here. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the claims of the present invention.

Claims

1. A three-tiered medical network intelligent referral method, characterized in that, include: Collect multi-source heterogeneous patient diagnosis and treatment information and medical institution resource information within the regional medical network, and convert the multi-source heterogeneous patient diagnosis and treatment information and medical institution resource information into structured data in a unified format through a standardized interface protocol; The structured data is input into a pre-set referral decision model to perform a correlation analysis between the severity characteristics of the patient's condition and the characteristics of the medical institution's capacity to receive patients, and to generate a referral necessity assessment result. When the referral necessity assessment result indicates that a referral is required, the candidate receiving institutions in the structured data are comprehensively scored based on multidimensional matching rules, and the target receiving institution is selected and a referral recommendation plan is generated based on the scoring results. The referral suggestion scheme is parsed to construct a referral execution instruction. In response to the referral execution instruction, a transfer route is planned, and the medical resources of the target receiving institution are locked using a smart contract mechanism to obtain the referral result.

2. The intelligent referral method for a three-tiered medical network according to claim 1, characterized in that, The process of converting the multi-source, heterogeneous patient diagnosis and treatment information and medical institution resource information into structured data in a unified format includes: Configure a standardized interface protocol to access raw business data from medical institutions at different levels, and use a data cleaning engine to remove redundant fields from the raw business data; The cleaned data is converted into intermediate data that conforms to the general health information exchange standard through mapping rules; The intermediate data is parsed and reconstructed into a key-value pair-based object model, and the object model is written into the encrypted storage space as the structured data.

3. The intelligent referral method for a three-tiered medical network according to claim 1, characterized in that, Before inputting the structured data into a pre-defined referral decision model, the method further includes constructing a referral decision model based on a graph network architecture, including: Construct a regional medical network topology, mapping each medical institution to a network node, and assigning the network node attributes including institution level, department configuration, and personnel load; Establish edges connecting the network nodes to represent referral paths, and assign the edges attribute features including geographical distance and historical referral success rate; The attribute features are input into the graph network architecture for feature learning, and a quantitative indicator representing the real-time admission capability of each medical institution is output.

4. The intelligent referral method for a three-tiered medical network according to claim 3, characterized in that, The correlation analysis between the severity of patients' conditions and the capacity of medical institutions to receive patients includes: Obtain parameters such as the percentage of available beds, the workload ratio of medical staff, and the availability of key specialized equipment from the target institution. Each parameter is assigned a corresponding weighting coefficient to the vacancy rate parameter, the medical staff load ratio parameter, and the availability parameter of the key specialized equipment. A weighted summation operation is performed on each parameter after weighting to obtain a comprehensive score that characterizes the medical institution's capacity to receive patients, and the comprehensive score is used as a feature of the medical institution's capacity to receive patients.

5. The intelligent referral method for a three-tiered medical network according to claim 4, characterized in that, The generation of the referral necessity assessment results includes: Extract patient vital sign data to generate a disease severity score, and obtain a comprehensive score for the treatment acceptance capacity and a bed resource score; Based on the preset decision-making logic, influencing factors are set for the severity score of the illness, the comprehensive score, and the bed resource score; The referral necessity score is obtained by linearly weighting the above scores based on the influencing factors, and then compared with a preset multi-level threshold range. The referral level is determined based on the range in which the comparison falls.

6. The intelligent referral method for a three-tiered medical network according to claim 1, characterized in that, The comprehensive scoring of candidate receiving institutions in the structured data based on multidimensional matching rules includes: Construct a multi-objective optimization function that includes variables such as expected referral waiting time, medical capacity matching degree, navigation distance, and medical insurance reimbursement compatibility. Assign corresponding priority weights to each variable in the multi-objective optimization function; The data of the candidate receiving institutions are fed into the multi-objective optimization function for calculation. Based on the calculation results, all candidate receiving institutions are sorted, and the institution with the highest ranking is selected as the recommended object.

7. The intelligent referral method for a three-tiered medical network according to claim 1, characterized in that, The method of planning a transport route in response to the referral execution instruction includes: Acquire real-time travel time data and traffic congestion index of the road network, and analyze the patient's vital sign data to generate a disease tolerance coefficient; A weighted path graph is constructed using the real-time passage time data and the disease tolerance coefficient, wherein the weight of the path edge is determined by the product of the passage time and the risk coefficient; The shortest path search algorithm is executed in the weighted path graph, and the route with the minimum overall cost is output as the transfer path.

8. The intelligent referral method for a three-tiered medical network according to claim 1, characterized in that, The method of using smart contract mechanisms to lock the medical resources of the target receiving institution includes: We use natural language processing models to parse the clinical text in the referral suggestion plan and extract key disease features and corresponding resource requirements. Based on the resource requirements, a smart contract on the distributed ledger is triggered, and the smart contract queries the time period status of the corresponding resources of the target receiving institution. When it is confirmed that the time period is idle, the status marker of the resource corresponding to the time period on the distributed ledger is changed to occupied.

9. The intelligent referral method for a three-tiered medical network according to claim 1, characterized in that, The step of updating the parameters of the referral decision model and the multidimensional matching rules using the actual execution status data and the usage status data includes: Record the actual timestamps of key nodes in the entire referral process, and calculate the actual time consumption and resource load of each step; An anomaly detection algorithm is invoked to compare the actual time consumption with a preset benchmark model, and anomalies where the deviation exceeds a threshold are identified. Extract the deviation feature data of the abnormal link, and use the deviation feature data to reverse correct the weight coefficients in the referral decision model.

10. An electronic device, characterized in that, The electronic device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to perform the operations performed by the three-tier medical network intelligent referral method as described in any one of claims 1 to 9.

Images