Medical self-service machine intelligent guidance method, storage medium and device

Abstract

The application discloses a medical self-service machine intelligent triage method, a storage medium and equipment, first, a differential diagnosis set is generated based on symptom description information, personal health record information and a real-time epidemiology prior probability model, and a clinical path time-prognosis model is used to perform Monte Carlo simulation on each candidate path to form a risk-reward quantitative portrait; combined with personal preference information and real-time resource state data of the hospital, a multi-objective optimization algorithm is used to calculate a personalized recommendation score, and an optimal candidate path is screened; if the department capacity of the hospital is over the limit, a regional medical resource collaborative query based on a consortium chain smart contract is triggered to generate a shunt triage scheme, and finally, a structured hierarchical triage path graph is output, and a closed-loop service of online review and offline appointment is started. The application realizes accurate triage from quantitative diagnosis, prognosis evaluation to cross-institution resource collaboration, and significantly improves the scientificity, individualization level of triage decision and overall utilization efficiency of regional medical resources.

Description

Technical Field

[0001] This invention relates to the field of intelligent medical technology, specifically to a smart triage method, storage medium, and device for a medical self-service machine. Background Technology

[0002] In the field of healthcare IT, self-service triage devices have become an important tool for hospitals to optimize service processes, triage patients, and improve efficiency. Current intelligent triage systems typically recommend appropriate departments based on simple matching of symptom keywords input by the patient and a pre-set departmental knowledge base. Some more advanced systems also consider real-time registration and queuing information for various hospital departments to estimate waiting times and provide route planning suggestions. These technologies alleviate, to some extent, the problem of patients wandering aimlessly due to unfamiliarity with hospital department layouts, and reduce unnecessary travel time within the hospital.

[0003] However, with the deepening of precision medicine and personalized health management concepts, the limitations of existing triage technologies are becoming increasingly apparent. First, their diagnostic recommendation process largely relies on static, general medical knowledge graphs or rule bases, lacking deep integration of individual health records and real-time responses to regional and seasonal epidemiological dynamics. This results in insufficient accuracy and timeliness of recommendations, making it difficult to provide probability-based differential diagnostic support. Second, in route planning, existing systems primarily focus on single or a few efficiency indicators such as time and distance, failing to quantitatively correlate different treatment route choices with the patient's expected health outcome (i.e., prognosis) and conduct risk assessments, thus failing to provide decision-making basis from the perspective of long-term health benefits for patients. More importantly, when resources within a single hospital become saturated, existing systems can usually only passively suggest waiting times or alternative appointment times, lacking the ability to proactively coordinate and allocate medical resources across a wider scope (such as medical consortia or alliances), making it difficult to fundamentally resolve the contradiction between the uneven distribution of high-quality medical resources and the overload of instantaneous demand.

[0004] Therefore, how to build a system that can deeply integrate personal health information and dynamic medical prior knowledge, and can conduct intelligent triage from the perspective of quantitative assessment of health gains and losses and collaborative optimization of regional resources, has become a key technical challenge to improve the quality and efficiency of medical services. Summary of the Invention

[0005] In view of the above problems, the present invention provides a smart triage method, storage medium and device for medical self-service machines. By integrating personal health records and real-time epidemiological data to generate a quantitative diagnostic set, and based on health loss simulation and regional resource collaborative optimization, it realizes personalized, prognosis-oriented and cross-hospital scheduling capabilities for precise triage.

[0006] To achieve the above objectives, in a first aspect, this application provides a smart triage method for medical self-service machines, comprising:

[0007] Obtain symptom descriptions input by users, personal medical treatment preferences, and authorized personal health records;

[0008] Based on symptom description information and personal health record information, combined with a real-time access to a localized spatiotemporal epidemiological prior probability model, a differential diagnosis set containing at least one suspected disease diagnosis and its corresponding probability is generated.

[0009] Based on the differential diagnosis set, the preset clinical pathway time-prognosis model is invoked to perform Monte Carlo simulation of the expected health gains and losses for each suspected disease diagnosis under different initial treatment pathways, generating a quantitative risk-benefit profile for each candidate pathway.

[0010] Based on individual medical preference information, a multi-objective function is constructed with path matching degree, doctor quality score, expected waiting time and medical cost as optimization objectives. Combined with real-time resource status data of the hospital, a multi-objective optimization algorithm is used to calculate a personalized recommendation score for each candidate path, and the optimal candidate path and the optimal personalized recommendation score are selected.

[0011] If the real-time capacity prediction value of the department corresponding to the optimal candidate path exceeds the preset diversion threshold, a regional medical resource collaborative query will be triggered. Based on the consortium blockchain smart contract, the available resource commitments of the collaborating hospitals will be obtained, and a diversion and guidance plan containing cross-hospital referral options will be generated.

[0012] Based on the optimal candidate path and triage plan, a structured hierarchical triage path map is generated and output to the user;

[0013] In addition, in response to the user's confirmation instruction for the online review service, the symptom description information, differential diagnosis set and hierarchical triage path map are packaged into a structured consultation form and pushed to the Internet hospital platform to start the closed-loop service of online doctor review and offline appointment.

[0014] Furthermore, based on symptom description information and personal health record information, combined with a real-time access to a localized spatiotemporal epidemiological prior probability model, a differential diagnosis set is generated, containing at least one suspected disease diagnosis and its corresponding probability, including:

[0015] Natural language processing is performed on symptom descriptions to extract a set of key symptom entities.

[0016] By performing correlation analysis between the set of key symptom entities and historical diagnostic records and drug allergy history in personal health records, a preliminary symptom-medical history correlation map is generated.

[0017] Based on the preliminary symptom-medical history association graph, query the medical knowledge graph to obtain a set of potential diseases that are directly or indirectly related to the set of key symptom entities;

[0018] Based on current time and geographic location information, obtain the spatiotemporal adjustment factors corresponding to each disease in the potential disease set from the localized spatiotemporal epidemiology prior probability model;

[0019] Using the Bayesian inference algorithm, the spatiotemporal adjustment factor is used as the prior probability correction term. Combined with the matching degree between the key symptom entity set and the symptom manifestation of each disease, the posterior probability of each potential disease is calculated.

[0020] The potential disease set is sorted according to the posterior probability, and diseases with probabilities higher than a preset threshold are selected to form a differential diagnosis set. Each disease in the set is labeled with its corresponding posterior probability value.

[0021] Furthermore, based on the differential diagnosis set, a pre-defined clinical pathway time-prognosis model is invoked to perform Monte Carlo simulations of the expected health gains and losses for each suspected disease diagnosis under different initial treatment pathways, generating a quantitative risk-benefit profile for each candidate pathway, including:

[0022] The sampling weights for the Monte Carlo simulation are determined based on the posterior probability of each disease in the differential diagnosis set.

[0023] For each suspected disease diagnosis, a corresponding clinical pathway time-prognostic model is loaded. The clinical pathway time-prognostic model defines the expected health status transition function and time delay cost function after the disease receives different levels of medical intervention at different time points.

[0024] Based on the real-time resource status data of the current hospital and regional collaborative hospitals, multiple initial medical paths, including different first-visit departments, different medical institutions, and different appointment times, are constructed as a set of candidate paths.

[0025] The differential diagnosis set is randomly sampled using sampling weights to simulate and determine the actual disease diagnosis of this sampling, and a medical treatment path from the candidate path set is randomly assigned.

[0026] Based on the clinical pathway time-prognosis model corresponding to the actual disease diagnosis, and combined with the time delay of each link in the assigned medical treatment path, the simulated health loss value along the medical treatment path to the preset assessment time point is calculated.

[0027] Repeat the random sampling, path allocation, and profit and loss calculation process to reach the preset number of simulations, and statistically analyze the health profit and loss value distribution of all simulation results for each candidate path;

[0028] Based on the distribution of health profit and loss values, the expected mean health profit and loss and risk variance of each candidate path are calculated, and the resource consumption estimate of the path is integrated to generate a risk-return quantitative profile to characterize the overall benefits and uncertainties of the path.

[0029] Furthermore, a corresponding clinical pathway time-prognostic model is loaded for each suspected disease diagnosis. This model defines the expected health status transition function and time delay cost function after receiving different levels of medical intervention at different time points for the disease, including:

[0030] Based on the disease code of the suspected disease diagnosis, the corresponding standard diagnosis and treatment protocol framework is retrieved from the pre-built clinical pathway knowledge base;

[0031] Based on the standard diagnosis and treatment protocol framework, multiple key decision-making time points are defined and a set of optional medical intervention measures are available at each key decision-making time point. The medical intervention measures include at least the selection of the first-visit department, examination and testing items, consultation mechanism and treatment plan.

[0032] For each medical intervention, a health status transfer function is associated. The health status transfer function takes the health status score before the intervention and the key physiological parameters in the patient's personal health record as inputs, and outputs the expected change in the health status score after the intervention.

[0033] For each pair of adjacent critical decision time nodes, a time delay cost function is defined. The time delay cost function takes the length of the time interval and the urgency level of the suspected disease diagnosis as input, and outputs the cumulative loss of expected health status score due to delay.

[0034] By integrating the health status transfer function and the time delay cost function with temporal logic, a clinical pathway time-prognosis model is constructed. This model can simulate and calculate the overall health gain or loss up to a preset assessment time point based on any given initial health status, medical visit time series, and intervention sequence.

[0035] Furthermore, based on individual patient preference information, a multi-objective function is constructed with path matching degree, doctor quality score, estimated waiting time, and treatment cost as optimization objectives. Combined with real-time hospital resource status data, a multi-objective optimization algorithm is used to calculate a personalized recommendation score for each candidate path, ultimately selecting the optimal candidate path and the optimal personalized recommendation score, including:

[0036] Analyze individual medical treatment preferences and extract the user's preference weight coefficients for waiting time, doctor level, cost, and distance to medical treatment;

[0037] For each candidate path, a path matching sub-score is calculated based on the target department and doctor information. The path matching sub-score is determined based on the correlation between the differential diagnosis set and the target doctor's area of ​​expertise.

[0038] For each candidate path, the doctor's comprehensive quality score is obtained based on the corresponding doctor information and used as a sub-score of the doctor's quality score. The comprehensive quality score integrates historical efficacy data and patient evaluations.

[0039] For each candidate path, the estimated waiting time is calculated based on the current queue length in the hospital's real-time resource status data and the prediction model, and then converted into a waiting time sub-score.

[0040] For each candidate path, the cost of seeking medical treatment is estimated based on the corresponding medical institution, department level, and doctor level information, and then converted into a cost of seeking medical treatment sub-score.

[0041] Using path matching degree sub-score, doctor quality score sub-score, waiting time sub-score, and medical cost sub-score as basic variables, and the corresponding preference weight coefficients extracted from personal medical preference information as weighting coefficients, a multi-objective function in the form of linear weighted sum is constructed.

[0042] By using a multi-objective optimization algorithm, the multi-objective function is solved, and a comprehensive personalized recommendation score is calculated for each candidate path;

[0043] All candidate paths are sorted according to the personalized recommendation score, and the candidate path with the highest score is determined as the optimal candidate path, and its score is recorded as the optimal personalized recommendation score.

[0044] Furthermore, if the real-time capacity prediction of the department corresponding to the optimal candidate path exceeds the preset triage threshold, a regional medical resource collaborative query is triggered. Based on the consortium blockchain smart contract, the available resource commitments of collaborating hospitals are obtained, and a triage and guidance plan including cross-hospital referral options is generated, including:

[0045] The real-time operational data of the target department in the hospital corresponding to the optimal candidate path is obtained and input into the department capacity prediction model to calculate the predicted saturation within a future preset time period.

[0046] Determine whether the predicted saturation exceeds the preset diversion threshold;

[0047] If the number of cases exceeds the limit, a resource query request containing the core information of the optimal candidate path will be generated. The core information will include at least the target department type, the required doctor's specialty, and the expected consultation time window.

[0048] The resource query request is used as a trigger condition to invoke the preset smart contract deployed on the regional medical consortium blockchain;

[0049] When the smart contract is executed, it automatically queries each collaborating hospital node on the chain for a summary of available resources that match the core information. The resource summary includes the availability of appointment slots and the expected capacity to receive patients.

[0050] The pre-defined smart contract matches and filters available resources based on pre-defined collaboration rules, and initiates a resource reservation request to the collaborative hospital node with the highest matching degree.

[0051] After receiving a successful resource reservation response from the collaborating hospital node, the smart contract generates an encrypted referral commitment token and binds the token to the collaborating hospital's detailed information.

[0052] Based on the referral commitment token and detailed information of collaborating hospitals, a triage and guidance plan is constructed that includes cross-hospital referral options. The triage and guidance plan clearly indicates the name, address, estimated waiting time, transportation guidance, and explanation of cost differences of the referral hospital.

[0053] Furthermore, the pre-defined smart contract, based on pre-set collaboration rules, matches and filters the available resources found, and initiates a resource reservation request to the collaborative hospital node with the highest matching degree, including:

[0054] The core information in the smart contract parsing resource query request is extracted as matching constraints, including the target department type, the required doctor's expertise, and the expected consultation time window.

[0055] Obtain the available resource summary returned from each collaborating hospital node. The available resource summary should include at least the department matching identifier, the set of doctor specialty tags, the sequence of appointment slots to be released in the future time period, and the basic patient reception capacity score.

[0056] The first-level screening is performed based on the target department type and the department matching identifier, retaining collaborative hospital nodes that match the department type or have equivalent reception capabilities;

[0057] Based on the required physician specialty direction and the set of physician specialty tags of the retained collaborative hospital nodes, the specialty direction matching degree is calculated, and combined with the basic patient reception capacity score, the comprehensive capacity matching score of each collaborative hospital node is calculated.

[0058] Based on the expected appointment time window, an alignment analysis is performed on the future time slot release sequence returned by each collaborating hospital node to calculate the matching degree of appointment time that each collaborating hospital node can provide within the expected time window;

[0059] The overall ability matching score and the appointment time matching degree are weighted and integrated to generate a final matching degree score for each reserved node;

[0060] All retained collaborative hospital nodes are sorted according to the final matching score, and the collaborative hospital node with the highest score is selected as the target collaborative hospital node.

[0061] Send a resource reservation request to the target collaborative hospital node, which includes a specific appointment time option. The resource reservation request includes a unique on-chain identifier for this collaborative transaction generated by the smart contract.

[0062] Furthermore, in response to the user's confirmation instruction for the online review service, the symptom description information, differential diagnosis set, and hierarchical triage path diagram are encapsulated into a structured consultation form, including:

[0063] Receive confirmation instructions for online review service triggered by users through the interactive interface;

[0064] Extract and integrate key symptom entities, symptom duration, and self-assessment information on severity from symptom description information;

[0065] Extract the name of each suspected disease diagnosis and its corresponding posterior probability value from the differential diagnosis set;

[0066] Extract the recommended preferred route details, alternative route details, and corresponding summary of the recommendation reasons from the tiered triage route map;

[0067] The key symptom entities, symptom duration, self-assessment information on severity, suspected disease diagnosis name and its posterior probability value, details of preferred and alternative pathways and summary of recommendation reasons are mapped and filled according to a preset structured data pattern;

[0068] In structured data models, allergy history and key past medical history are linked together from personal health record information;

[0069] The filled structured data is standardized, encoded, and serialized to generate a machine-readable, structured consultation form that conforms to medical information exchange standards.

[0070] In a second aspect, the present invention also provides a computer-readable storage medium having stored thereon computer program instructions that, when executed by a processor, implement the method described in the first aspect.

[0071] In a third aspect, the present invention also provides an electronic device including a memory and a processor, the memory being used to store one or more computer program instructions, wherein the one or more computer program instructions are executed by the processor to implement the method described in the first aspect.

[0072] Unlike existing technologies, the above-mentioned technical solution provides a smart triage method, storage medium, and device for medical self-service machines. First, it generates a differential diagnosis set based on symptom description information, personal health record information, and a real-time epidemiological prior probability model. Then, it uses a clinical pathway time-prognosis model to perform Monte Carlo simulations on each candidate path, forming a quantitative risk-reward profile. Combining personal medical preference information with real-time hospital resource status data, a multi-objective optimization algorithm calculates a personalized recommendation score to select the optimal candidate path. If the hospital's department capacity exceeds its limit, it triggers a regional medical resource collaborative query based on a consortium blockchain smart contract to generate a triage plan. Finally, it outputs a structured hierarchical triage path map and initiates a closed-loop service of online review and offline appointment. This invention achieves precise triage from quantitative diagnosis and prognostic assessment to cross-hospital resource collaboration, significantly improving the scientific nature and personalization of triage decisions and the overall utilization efficiency of regional medical resources.

[0073] The above description of the invention is merely an overview of the technical solution of this application. In order to enable those skilled in the art to better understand the technical solution of this application and to implement it based on the description and drawings, and to make the above-mentioned objectives and other objectives, features and advantages of this application easier to understand, the following description is provided in conjunction with the specific embodiments and drawings of this application. Attached Figure Description

[0074] The accompanying drawings are only used to illustrate the principles, implementation methods, applications, features, and effects of specific embodiments of the present invention and other related contents, and should not be considered as limitations on this application.

[0075] In the accompanying drawings of the instruction manual:

[0076] Figure 1 This is a schematic diagram illustrating steps S101 to S106 of the method described in the specific implementation embodiment;

[0077] Figure 2 This is a schematic diagram illustrating steps S201 to S206 of the method described in a specific implementation.

[0078] Figure 3 This is a schematic diagram illustrating steps S301 to S307 of the method described in the specific implementation embodiment;

[0079] Figure 4 This is a schematic diagram illustrating steps S401 to S405 of the method described in a specific embodiment;

[0080] Figure 5 This is a schematic diagram of the structure of the electronic device described in a specific embodiment.

[0081] The reference numerals used in the above figures are explained as follows:

[0082] 1. Electronic equipment;

[0083] 11. Memory;

[0084] 12. Processor. Detailed Implementation

[0085] To illustrate the possible application scenarios, technical principles, implementable specific solutions, and achievable objectives and effects of this application in detail, the following description, in conjunction with the listed specific embodiments and accompanying drawings, provides a detailed explanation. The embodiments described herein are merely illustrative of the technical solutions of this application and are therefore intended to limit the scope of protection of this application.

[0086] In this document, the term "embodiment" means that a specific feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The term "embodiment" appearing in various places throughout the specification does not necessarily refer to the same embodiment, nor does it specifically limit its independence or connection with other embodiments. In principle, in this application, as long as there are no technical contradictions or conflicts, the technical features mentioned in each embodiment can be combined in any way to form corresponding implementable technical solutions.

[0087] Unless otherwise defined, the technical terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains; the use of related terms herein is merely for the purpose of describing particular embodiments and is not intended to limit this application.

[0088] In the description of this application, the term "and / or" is used to describe the logical relationship between objects, indicating that three relationships can exist. For example, A and / or B means: A exists, B exists, and A and B exist simultaneously. Additionally, the character " / " in this document generally indicates that the preceding and following objects have an "or" logical relationship.

[0089] In this application, terms such as “first” and “second” are used only to distinguish one entity or operation from another, and do not necessarily require or imply any actual quantity, hierarchy or order relationship between these entities or operations.

[0090] Without further limitations, the use of terms such as “comprising,” “including,” “having,” or other similar open-ended expressions in this application is intended to cover non-exclusive inclusion, which does not exclude the presence of additional elements in a process, method, or product that includes the stated elements, such that a process, method, or product that includes a list of elements may include not only those defined elements but also other elements not expressly listed, or elements inherent to such a process, method, or product.

[0091] As understood in the Examination Guidelines, in this application, expressions such as "greater than," "less than," and "exceeding" are understood to exclude the stated number; expressions such as "above," "below," and "within" are understood to include the stated number. Furthermore, in the description of the embodiments in this application, "multiple" means two or more (including two), and similar expressions related to "multiple" are also understood in this way, such as "multiple groups" and "multiple times," unless otherwise explicitly specified.

[0092] In the description of the embodiments of this application, the space-related expressions used, such as "center," "longitudinal," "lateral," "length," "width," "thickness," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "vertical," "top," "bottom," "inner," "outer," "clockwise," "counterclockwise," "axial," "radial," and "circumferential," indicate the orientation or positional relationship based on the orientation or positional relationship shown in the specific embodiments or drawings. They are only for the purpose of describing the specific embodiments of this application or for the reader's understanding, and do not indicate or imply that the device or component referred to must have a specific position, a specific orientation, or be constructed or operated in a specific orientation. Therefore, they should not be construed as limitations on the embodiments of this application.

[0093] Please see Figure 1 In a first aspect, this embodiment provides a smart triage method for medical self-service machines, including:

[0094] S101. Obtain symptom description information, personal medical treatment preference information, and authorized personal health record information input by the user;

[0095] S102. Based on symptom description information and personal health record information, combined with a real-time access to a localized spatiotemporal epidemiological prior probability model, generate a differential diagnosis set containing at least one suspected disease diagnosis and its corresponding probability.

[0096] S103. Based on the differential diagnosis set, call the preset clinical pathway time-prognosis model to perform Monte Carlo simulation on the expected health gains and losses of each suspected disease diagnosis under different initial medical treatment paths, and generate a risk-benefit quantitative profile for each candidate path.

[0097] S104. Based on personal medical preference information, construct a multi-objective function with path matching degree, doctor quality score, expected waiting time and medical cost as optimization objectives. Combined with real-time hospital resource status data, calculate personalized recommendation scores for each candidate path through a multi-objective optimization algorithm, and select the optimal candidate path and the optimal personalized recommendation score.

[0098] S105. If the real-time capacity prediction value of the department in this hospital corresponding to the optimal candidate path exceeds the preset diversion threshold, a regional medical resource collaborative query will be triggered. Based on the consortium blockchain smart contract, the available resource commitments of the collaborating hospitals will be obtained, and a diversion and guidance plan containing cross-hospital referral options will be generated.

[0099] S106. Based on the optimal candidate path and triage plan, generate a structured hierarchical triage path map and output it to the user;

[0100] In addition, in response to the user's confirmation instruction for the online review service, the symptom description information, differential diagnosis set and hierarchical triage path map are packaged into a structured consultation form and pushed to the Internet hospital platform to start the closed-loop service of online doctor review and offline appointment.

[0101] In step S101, the symptom description information is the text or voice description of the user's discomfort entered through the self-service machine interface; the personal medical preference information is the user's pre-set or temporarily selected preference parameters regarding the medical process, such as the weight given to waiting time, doctor level, cost, and distance; and the authorized personal health record information is the user's historical medical data securely retrieved from the regional health information platform or hospital information system. This step, by integrating the information actively entered by the user with the authorized historical health data, provides a comprehensive and personalized data foundation for subsequent intelligent analysis.

[0102] In step S102, the localized spatiotemporal epidemiological prior probability model is a dynamically updated statistical model. Based on historical medical visits and public health data for a specific geographical region and the current time period, it reflects the prevalence trends of different diseases. This model connects to external monitoring data in real time via a data interface. When generating the differential diagnosis set, the system performs correlation analysis between symptom description information and individual health record information to initially screen possible disease directions. It then uses the disease prevalence probability provided by the localized model as prior knowledge for correction, calculating the posterior probability of each suspected disease in the current context, ultimately forming a disease list sorted by probability. This step combines individual health information with dynamic population epidemiological trends, improving the rationality and timeliness of disease prediction.

[0103] In step S103, the clinical pathway time-prognosis model is a pre-constructed simulation model of the diagnosis and treatment process for a specific disease, used to quantify the expected health outcomes and time delay costs under different treatment pathways. Monte Carlo simulation estimates the behavior of an uncertain system through extensive random sampling. Preferably, the system uses the probability of each disease in the differential diagnosis set as weights to randomly simulate a user actually having a certain disease, and randomly assigns an initial treatment path to this simulated scenario; subsequently, the corresponding clinical pathway time-prognosis model is called to calculate the expected health gains and losses along the path to the preset assessment time point; through statistical analysis of a large number of simulation results, a probability distribution of health gains and losses is generated for each candidate path, i.e., a risk-benefit quantitative profile. This step transforms the selection of treatment paths into the prediction and risk assessment of health outcomes, providing a quantitative basis for subsequent decision-making.

[0104] In step S104, path matching degree measures the degree of fit between the candidate path and the diseases in the differential diagnosis set; doctor quality score is a comprehensive evaluation index constructed based on multi-dimensional information; estimated waiting time is the waiting time predicted by analyzing real-time hospital resource status data; and medical cost covers related direct economic expenditures. These four sub-objectives are combined into a comprehensive scoring formula based on the weights assigned by the user in their personal medical preference information, which is the multi-objective function. A multi-objective optimization algorithm is used to handle these potentially conflicting objectives, calculating a personalized recommendation score that balances various factors for each candidate path, and selecting the optimal candidate path accordingly. This step, based on quantifying health risks, integrates user subjective preferences and the objective resource status of the hospital to complete personalized path optimization.

[0105] In step S105, the real-time capacity prediction value of the hospital's department is estimated by a prediction model based on the department's current operating data, representing the service load saturation over a future period. The preset triage threshold is a critical value used to determine whether a department is overloaded. When the real-time capacity prediction value of the hospital's department exceeds the preset triage threshold, the system triggers a collaborative query of regional medical resources. The consortium blockchain smart contract is an automatically executing program code deployed on the regional medical consortium blockchain, defining the rules for cross-hospital resource queries and appointments. By calling this smart contract to initiate a query to collaborating hospitals and obtain commitments for available resources, a cross-hospital referral option containing detailed information such as recommended referral hospitals and estimated times is generated, forming a triage and guidance plan. This step uses blockchain technology to achieve cross-institutional collaboration and dynamically schedule regional medical resources to achieve load balancing.

[0106] In step S106, the structured hierarchical triage path map is a visual guidance solution that integrates and presents the optimal candidate paths and triage plans in a hierarchical manner, outputting them to users through self-service machine screens or related applications. When a user confirms the need for online review services, the system encapsulates the symptom description information, differential diagnosis set, and hierarchical triage path map into a structured consultation form conforming to medical information exchange standards and pushes it to the internet hospital platform. After online doctor review, offline hospital appointments can be made directly for the user within the platform, thus forming a closed-loop service connecting online review and offline appointments. This step provides clear action guidelines and enhances the reliability of the triage process and user experience by connecting internet medical resources.

[0107] This embodiment achieves intelligent triage by constructing a multi-layered decision-making framework. First, it integrates personal health records and real-time epidemiological data to generate probabilistic differential diagnoses. Then, it utilizes clinical pathway time-prognosis models and Monte Carlo simulations to quantify health benefits and risks associated with patient pathways. Based on this, it integrates user preferences and real-time resource data to generate personalized recommendations through multi-objective optimization. When hospital resources are overloaded, cross-hospital resource collaboration and triage are achieved through consortium blockchain smart contracts. Finally, structured output and a closed-loop internet hospital system ensure the availability and continuity of the solution. This embodiment systematically improves the efficiency of triage in terms of medical rationality, personalization, resource allocation scope, and service closed-loop.

[0108] Please see Figure 2 In some embodiments, based on symptom description information and personal health record information, combined with a real-time access to a localized spatiotemporal epidemiological prior probability model, a differential diagnosis set containing at least one suspected disease diagnosis and its corresponding probability is generated, including:

[0109] S201. Perform natural language processing on the symptom description information to extract the set of key symptom entities;

[0110] S202. Perform correlation analysis between the key symptom entity set and the historical diagnosis records and drug allergy history in the personal health record information to generate a preliminary symptom-medical history correlation map.

[0111] S203. Based on the preliminary symptom-medical history association graph, query the medical knowledge graph to obtain a set of potential diseases that are directly or indirectly related to the set of key symptom entities.

[0112] S204. Based on the current time and geographic location information, obtain the spatiotemporal adjustment factors corresponding to each disease in the potential disease set from the localized spatiotemporal epidemiology prior probability model.

[0113] S205. Using the Bayesian inference algorithm, the spatiotemporal adjustment factor is used as the prior probability correction term. Combined with the matching degree between the key symptom entity set and the symptom manifestation of each disease, the posterior probability of each potential disease is calculated.

[0114] S206. Sort the potential disease set according to the posterior probability, select diseases with a probability higher than a preset threshold to form a differential diagnosis set, and label the corresponding posterior probability value for each disease in the set.

[0115] In step S201, natural language processing technology is used to parse the symptom description information input by the user. Specifically, a named entity recognition model can be used to automatically identify and extract symptom terms with clear clinical significance from the text, such as "fever," "cough," and "chest pain." These identified terms constitute a set of key symptom entities.

[0116] In step S202, association analysis is achieved by calculating the association strength between key symptom entities and structured data items in a user's health record. For example, matching current symptoms with historical diagnostic records establishes a strong association if the current symptom of "joint pain" appears in a user's record where they were previously diagnosed with "gout." Simultaneously, cross-validation with drug allergy history is performed; if a user is allergic to "penicillin" and currently has a "rash," a possible allergic reaction is indicated. Based on these matching and validation results, a network graph is constructed with symptoms, historical diagnoses, and allergic drugs as nodes and association relationships as edges—a preliminary symptom-medical history association graph. This graph quantifies the correlation between current symptoms and personal medical history.

[0117] In step S203, the medical knowledge graph stores the semantic relationships between medical concepts such as diseases, symptoms, and examinations. During a query, the symptom entity in the preliminary symptom-medical history association graph serves as the starting point, and the graph is traversed and reasoned along relationship paths such as "symptom-disease." For example, inputting "fever" and "cough" retrieves the directly associated "upper respiratory tract infection," and the indirectly associated "pleurisy" through disease nodes such as "pneumonia." Through this process, all diseases with direct or indirect pathological associations to the input symptoms are obtained, forming a potential disease set. This step utilizes an authoritative medical knowledge base to expand the scope of disease screening.

[0118] In step S204, the spatiotemporal adjustment factor is a correction coefficient reflecting the spatiotemporal prevalence intensity of a disease. The localized spatiotemporal epidemiological prior probability model establishes a functional relationship between disease incidence and time and geographical location based on historical monitoring data. According to the current time and geographical location provided by the user, the model outputs the adjustment factor for each disease in the potential disease set under the current spatiotemporal context. For example, in peak influenza seasons and regions, the adjustment factor for influenza will be significantly greater than 1. This factor is used to adjust the baseline prior probability of the disease in subsequent probability calculations.

[0119] In step S205, a Bayesian inference algorithm is used to calculate the posterior probability. The prior probability of the algorithm is based on the baseline incidence rate of the disease and corrected by the spatiotemporal adjustment factor in step S204. The likelihood function is defined by calculating the matching degree between the user's key symptom entity set and the typical symptom sets of each disease. The matching degree can be quantified using methods such as the Jaccard similarity coefficient. The Bayesian formula combines the corrected prior probability with the symptom matching degree (likelihood) to calculate the posterior probability of each potential disease given the current individual symptoms and spatiotemporal context. This probability integrates individual clinical manifestations and population epidemiological evidence.

[0120] In step S206, a preset threshold is used to filter high-probability diseases. All potential diseases are sorted in descending order of their calculated posterior probability values. Diseases with probability values ​​higher than the preset threshold are included in the final differential diagnosis set, and each disease in the set is labeled with its specific posterior probability value. The preset threshold can be adjusted according to clinical needs, for example, set to 0.1 to filter out options with extremely low probabilities. This step outputs a sorted and filtered, probability-weighted, structured list of diagnostic hypotheses.

[0121] This embodiment extracts symptom entities through natural language processing and constructs an association graph by combining personal medical history, enhancing the individualized dimension of the analysis. It ensures the medical completeness of disease screening by querying a medical knowledge graph; introduces a spatiotemporal adjustment factor to dynamically correct prior probabilities; and utilizes Bayesian inference to fuse individual symptom evidence, ultimately outputting a probabilistic diagnostic ranking list. This embodiment dynamically combines symptomatology, personal medical history, and population epidemiology, realizing the transformation from symptom description to probabilistic diagnosis, laying a reliable medical reasoning foundation for subsequent precision triage.

[0122] Please see Figure 3 In some embodiments, based on the differential diagnosis set, a preset clinical pathway time-prognosis model is invoked to perform Monte Carlo simulations on the expected health gains and losses for each suspected disease diagnosis under different initial treatment pathways, generating a quantitative risk-benefit profile for each candidate pathway, including:

[0123] S301. Determine the sampling weights for the Monte Carlo simulation based on the posterior probability of each disease in the differential diagnosis set.

[0124] S302. Load the corresponding clinical pathway time-prognostic model for each suspected disease diagnosis. The clinical pathway time-prognostic model defines the expected health status transfer function and time delay cost function after the disease receives different levels of medical intervention at different time points.

[0125] S303. Based on the real-time resource status data of the current hospital and regional collaborative hospitals, construct multiple initial medical treatment paths as a candidate path set, including different first-visit departments, different medical institutions, and different appointment times.

[0126] S304. Randomly sample the differential diagnosis set with sampling weights, simulate and determine the true disease diagnosis of this sampling, and randomly assign a medical treatment path from the candidate path set.

[0127] S305. Based on the clinical pathway time-prognosis model corresponding to the actual disease diagnosis, and combined with the time delay of each link in the assigned medical treatment path, calculate the simulated health loss value along the medical treatment path to the preset assessment time point.

[0128] S306. Repeat the random sampling, path allocation and profit and loss calculation process to reach the preset number of simulations, and statistically analyze the health profit and loss value distribution of all simulation results corresponding to each candidate path;

[0129] S307. Based on the distribution of health profit and loss values, calculate the expected health profit and loss mean and risk variance of each candidate path, and integrate the resource consumption estimate of the path to generate a risk-return quantitative profile to characterize the comprehensive benefits and uncertainties of the path.

[0130] In step S301, the sampling weights of the Monte Carlo simulation are directly determined by the posterior probability value of each disease in the differential diagnosis set. For example, if the posterior probability of disease A is 0.6, that of disease B is 0.3, and that of disease C is 0.1, then the probability weight of disease A being selected during the simulation sampling is 0.6, ensuring that the frequency of each disease in the simulation is proportional to its actual probability, making the simulation results closer to the real diagnostic uncertainty.

[0131] In step S302, the clinical pathway time-prognosis model consists of two parts: first, a health status transition function, used to quantify the expected change in a patient's health status after implementing a certain medical intervention (such as using a specific drug or undergoing surgery) at a specific time point; and second, a time delay cost function, used to quantify the cumulative loss of health status caused by time delays such as waiting for examinations and queuing for treatment. Through function definitions, the model transforms the abstract diagnosis and treatment process into a computable health status trajectory.

[0132] In step S303, the real-time resource status data includes the doctor's schedule for each department, the current waiting queue, the availability of examination equipment, and appointment information from regional collaborating hospitals. The system uses this data to generate multiple specific initial treatment paths, each path clearly specifying the first-visit department, the treatment institution (this hospital or a collaborating hospital), and a feasible appointment time. All possible paths constitute a candidate path set, providing a selection space for simulation.

[0133] In step S304, a specific simulation sampling is performed. Specifically, based on the weights determined in step S301, a disease is randomly selected from the differential diagnosis set as the assumed real disease of the user in this simulation. Subsequently, a medical treatment path is randomly selected from the candidate path set constructed in step S303 and assigned to this simulation. This process simulates a single random event under the combined effects of diagnostic uncertainty and path selectivity.

[0134] In step S305, the health gain / loss value for a single simulation is calculated. Based on the real disease diagnosis extracted in step S304, the corresponding clinical pathway time-prognosis model is invoked. The specific time delays of each link in the assigned medical pathway (such as registration waiting, examination queuing, and treatment implementation) are input into the model's time delay cost function to calculate the cumulative health loss. Simultaneously, according to the planned medical intervention sequence in the pathway, the health state transition function is invoked sequentially to calculate the cumulative health gain. Finally, the gain is subtracted from the loss to obtain the simulated health gain / loss value at a preset future assessment time point (such as one week later) along the pathway. This value can be positive or negative.

[0135] In step S306, a statistical distribution is obtained through numerous repeated simulations. The preset number of simulations is typically set to a large number, such as 10,000. Steps S304 and S305 are repeated, with diseases and pathways randomly selected and profit / loss values ​​calculated each time. After completion, for each pathway in the candidate pathway set, all simulation results assigned to that pathway are selected, and their health profit / loss values ​​are summarized to form a data distribution. This distribution describes the various health outcomes that may result from choosing that pathway and their frequency of occurrence.

[0136] In step S307, for the health benefit distribution corresponding to each candidate path, its mean (expected health benefit) and variance (risk uncertainty) are calculated; simultaneously, the estimated resource consumption of the path (such as doctor's time and examination costs) can be integrated. These indicators (expected benefit, risk variance, and resource consumption) are combined to form a multi-dimensional quantitative profile for comprehensively comparing the advantages and disadvantages of different paths. For example, a path may have high expected benefits but also be accompanied by high risks and high costs.

[0137] This embodiment determines sampling weights through diagnostic probabilities to ensure the statistical rationality of the simulation; it transforms time delays and medical interventions into calculable health gains and losses by loading a clinical pathway time-prognosis model; it ensures the feasibility of the simulation by constructing a candidate pathway set based on real-time resources; and finally, through extensive repeated simulations and statistical analysis, it transforms the complex and uncertain diagnosis and treatment decision-making problem into a quantitative assessment of the clear expected benefits, risks, and resource consumption of each pathway. This embodiment provides a scientific basis for health outcome-oriented pathway selection under diagnostic uncertainty and resource constraints.

[0138] Please see Figure 4 In some embodiments, a corresponding clinical pathway time-prognostic model is loaded for each suspected disease diagnosis. The clinical pathway time-prognostic model defines the expected health status transition function and time delay cost function after the disease receives different levels of medical intervention at different time points, including:

[0139] S401. Based on the disease code of the suspected disease diagnosis, retrieve the corresponding standard diagnosis and treatment protocol framework from the pre-built clinical pathway knowledge base.

[0140] S402. Based on the standard diagnosis and treatment protocol framework, define multiple key decision-making time nodes and a set of optional medical intervention measures at each key decision-making time node. The medical intervention measures shall at least include the selection of the first-visit department, examination and testing items, consultation mechanism and treatment plan.

[0141] S403. Associate a health status transfer function with each medical intervention. The health status transfer function takes the health status score before the intervention and the key physiological parameters in the patient's personal health record as inputs and outputs the expected change in the health status score after the intervention.

[0142] S404. Define a time delay cost function for the time interval between each pair of adjacent critical decision time nodes. The time delay cost function takes the length of the time interval and the urgency level of the suspected disease diagnosis as input, and outputs the cumulative loss of expected health status score due to delay.

[0143] S405. Integrate the health status transfer function and the time delay cost function with time-series logic to construct a clinical pathway time-prognosis model. The clinical pathway time-prognosis model can simulate and calculate the overall health gain or loss value up to the preset assessment time point based on any given initial health status, medical visit time series and intervention measure series.

[0144] In step S401, the disease code is a standardized code used to uniquely identify a specific disease, such as the International Classification of Diseases (ICD). A pre-built clinical pathway knowledge base stores standardized diagnostic and treatment process frameworks for various diseases. Based on the disease code corresponding to a suspected disease diagnosis, a matching standard diagnostic and treatment protocol framework is retrieved from this knowledge base. This framework describes the recognized and typical diagnostic and treatment steps and sequence for that disease.

[0145] In step S402, based on the retrieved standard treatment protocol framework, several key decision-making time points in the treatment process are identified and defined. These time points may include "initial consultation time," "after completion of key examinations," and "when determining the treatment plan." At each key decision-making time point, according to the treatment protocol, all available and reasonable medical interventions are listed to form a set of medical intervention measures. These measures may cover everything from determining the initial department, arranging specific examinations and tests, whether to initiate a multidisciplinary consultation, and selecting the appropriate treatment plan.

[0146] In step S403, a health status transition function is associated with each measure in the set of medical interventions. The input parameters of this function include at least the patient's health status score before receiving the intervention (e.g., using a quality of life scale) and key physiological parameters extracted from the individual's health record (such as age and underlying disease indicators). The function's output is a numerical value representing the expected change in the patient's health status score after implementing the intervention; a positive value indicates improvement, and a negative value indicates potential risks or side effects. This function can be trained by analyzing historical clinical data using methods such as regression analysis.

[0147] In step S404, a time delay cost function is defined for the time interval between each pair of adjacent critical decision-making time nodes. This function quantifies the health loss caused by waiting between treatment steps. The input parameters of the function include the actual length of the time interval and the inherent urgency level of the disease progression (e.g., classified as urgent, sub-urgent, or routine). The output of the function is a numerical value representing the cumulative loss of health status score due to delayed treatment within that time interval. The higher the urgency level of disease progression, the greater the loss per unit time delay is typically.

[0148] In step S405, all health status transition functions defined in step S403 and all time delay cost functions defined in step S404 are integrated according to the sequence of key decision-making time nodes specified in the standard treatment protocol framework. Through this temporal logical integration, a complete clinical pathway time-prognostic model is constructed. Given an initial health state, a time series of a specific treatment path (i.e., the actual occurrence time of each step), and a sequence of interventions taken along that path, the model can progressively calculate the changes in health status at each time point, and ultimately simulate and extrapolate the overall health gain or loss value at a preset assessment time point (e.g., one week after the end of treatment).

[0149] This embodiment establishes the model framework by retrieving standard treatment protocols, clarifies decision variables by defining key decision nodes and a set of interventions, and quantifies medical interventions and time factors into changes in health values ​​by associating health state transition functions and time-delay cost functions. Finally, through temporal logic integration, a computational model capable of dynamically simulating health outcomes based on specific treatment pathways is formed. The construction of this model transforms the assessment of health benefits and losses along treatment pathways from qualitative, empirical judgments to quantifiable and computable scientific analysis, providing the core computational engine for the aforementioned Monte Carlo simulation.

[0150] In some embodiments, based on individual patient preference information, a multi-objective function is constructed with path matching degree, doctor quality score, estimated waiting time, and patient cost as optimization objectives. Combined with real-time hospital resource status data, a multi-objective optimization algorithm is used to calculate a personalized recommendation score for each candidate path, and the optimal candidate path and optimal personalized recommendation score are selected, including:

[0151] Analyze individual medical treatment preferences and extract the user's preference weight coefficients for waiting time, doctor level, cost, and distance to medical treatment;

[0152] For each candidate path, a path matching sub-score is calculated based on the target department and doctor information. The path matching sub-score is determined based on the correlation between the differential diagnosis set and the target doctor's area of ​​expertise.

[0153] For each candidate path, the doctor's comprehensive quality score is obtained based on the corresponding doctor information and used as a sub-score of the doctor's quality score. The comprehensive quality score integrates historical efficacy data and patient evaluations.

[0154] For each candidate path, the estimated waiting time is calculated based on the current queue length in the hospital's real-time resource status data and the prediction model, and then converted into a waiting time sub-score.

[0155] For each candidate path, the cost of seeking medical treatment is estimated based on the corresponding medical institution, department level, and doctor level information, and then converted into a cost of seeking medical treatment sub-score.

[0156] Using path matching degree sub-score, doctor quality score sub-score, waiting time sub-score, and medical cost sub-score as basic variables, and using the corresponding preference weight coefficients extracted from personal medical preference information as weighting coefficients, a multi-objective function in the form of a linear weighted sum is constructed.

[0157] By using a multi-objective optimization algorithm, the multi-objective function is solved, and a comprehensive personalized recommendation score is calculated for each candidate path;

[0158] All candidate paths are sorted according to the personalized recommendation score, and the candidate path with the highest score is determined as the optimal candidate path, and its score is recorded as the optimal personalized recommendation score.

[0159] In this embodiment, personal medical preference information is parsed, and the structured preference data set by the user is processed. For example, users can set their emphasis on four dimensions—"waiting time," "doctor level," "cost," and "distance to the clinic"—using interface sliders or options. After normalizing these settings, the system obtains corresponding preference weight coefficients, with the sum of each coefficient being 1. This set of coefficients reflects the user's subjective value orientation when facing multiple objective trade-offs.

[0160] The calculation of the path matching score depends on the relevance between the differential diagnosis set and the target physician's area of ​​expertise. Specifically, it calculates the sum of the matching scores between all diseases (or their probability weights) in the differential diagnosis set and the target physician's department and subspecialty. For example, if the target physician's specialty is cardiology, and "coronary artery disease" has a high probability in the differential diagnosis set, the matching score will be high; if the set mainly contains diagnoses related to "dermatology," the score will be low. This score quantifies, from a medical professional perspective, the degree to which the path aligns with the user's potential medical condition.

[0161] The physician quality score sub-scores are obtained based on a pre-built comprehensive physician evaluation system. This system integrates physicians' historical efficacy data (such as treatment effectiveness rate for a certain disease and complication rate) with patient evaluation data (such as satisfaction scores and complaint rates), and calculates a comprehensive quality score through weighted averages. For a specific physician in the candidate pathway, their current score is directly retrieved from this system as the sub-score.

[0162] The calculation of estimated waiting time requires consideration of real-time hospital resource status data. The system obtains the current queue length for the target department or doctor and uses a predictive model (e.g., a time series prediction model based on historical average consultation speed) to estimate the waiting time after a new member joins the queue. This estimated physical time value is then mapped to a standardized waiting time sub-score through a transformation function; generally, the longer the waiting time, the lower the waiting time sub-score.

[0163] The cost of a medical visit is estimated based on information such as the pricing standards of the medical institutions corresponding to the path, departmental level markups, and doctor level markups. The system calculates an estimated total cost based on this information. This cost value is also mapped to a standardized cost sub-score through a transformation function; the higher the cost, the lower the cost sub-score. The design of the transformation function can take into account the user's affordability or generally acceptable range.

[0164] The multi-objective function is constructed using a linear weighted sum. The calculated path matching degree sub-score, doctor quality score sub-score, waiting time sub-score, and consultation cost sub-score are used as four basic variables. The corresponding preference weight coefficients obtained from analysis are used as the weighting coefficients of each variable. These are linearly combined to form a comprehensive scoring formula, which unifies sub-scores of different dimensions and directions into a comparable scalar value.

[0165] The above function is solved using a multi-objective optimization algorithm. Since the sub-scores and weighting coefficients for each candidate path are fixed, the "solution" here essentially involves calculating the value of the linear weighted sum formula for each candidate path to obtain its personalized recommendation score. This score comprehensively reflects the degree to which the path meets the user's personal preferences across four dimensions: medical matching degree, doctor quality, time efficiency, and economic benefits.

[0166] All candidate paths are sorted in descending order of their calculated personalized recommendation scores. The path ranked first is determined as the optimal candidate path, and its score is recorded as the optimal personalized recommendation score. This result represents the single recommendation scheme that best meets the user's personalized needs given all current constraints.

[0167] This embodiment obtains subjective weights by analyzing user preferences, ensures medical rationality by calculating path matching degree, introduces objective efficacy evaluation by querying doctor quality scores, reflects realistic constraints and burdens by estimating waiting time and treatment costs, and finally transforms multi-objective decision-making into a computable optimization problem using a linear weighted model. This method respects users' subjective wishes while incorporating medical professionalism and resource realities, achieving a scientific transformation from multi-objective trade-offs to personalized single recommendations, providing users with a clear, interpretable, and value-aligned optimal path selection.

[0168] In some embodiments, if the real-time capacity prediction value of the department corresponding to the optimal candidate path exceeds a preset triage threshold, a regional medical resource collaborative query is triggered. Based on the consortium blockchain smart contract, the available resource commitments of collaborating hospitals are obtained, and a triage and guidance plan including cross-hospital referral options is generated, including:

[0169] The real-time operational data of the target department in the hospital corresponding to the optimal candidate path is obtained and input into the department capacity prediction model to calculate the predicted saturation within a future preset time period.

[0170] Determine whether the predicted saturation exceeds the preset diversion threshold;

[0171] If the number of cases exceeds the limit, a resource query request containing the core information of the optimal candidate path will be generated. The core information will include at least the target department type, the required doctor's specialty, and the expected consultation time window.

[0172] The resource query request is used as a trigger condition to invoke the preset smart contract deployed on the regional medical consortium blockchain;

[0173] When the smart contract is executed, it automatically queries each collaborating hospital node on the chain for a summary of available resources that match the core information. The resource summary includes the availability of appointment slots and the expected capacity to receive patients.

[0174] The pre-defined smart contract matches and filters available resources based on pre-defined collaboration rules, and initiates a resource reservation request to the collaborative hospital node with the highest matching degree.

[0175] After receiving a successful resource reservation response from the collaborating hospital node, the smart contract generates an encrypted referral commitment token and binds the token to the collaborating hospital's detailed information.

[0176] Based on the referral commitment token and detailed information of collaborating hospitals, a triage and guidance plan is constructed that includes cross-hospital referral options. The triage and guidance plan clearly indicates the name, address, estimated waiting time, transportation guidance, and explanation of cost differences of the referral hospital.

[0177] In this embodiment, real-time operational data of the target department of the hospital is acquired, including the current number of patients in consultation, the number of patients with appointments but no visits, the doctors' outpatient status, and the utilization rate of examination equipment. The department capacity prediction model is a time series prediction model based on historical and real-time data, such as an autoregressive integral moving average model or a long short-term memory network model. By inputting the real-time operational data into this model, the ratio of the department's service demand to its maximum capacity can be calculated for a future preset time period (such as the afternoon), i.e., the predicted saturation level.

[0178] The preset triage threshold is a pre-defined value used to determine whether a department is overloaded. This threshold can be set according to the hospital's management strategy, for example, to 80%. The system checks whether the predicted saturation exceeds this threshold; if it does, it determines that the hospital's resources are strained and cross-hospital collaborative triage needs to be initiated.

[0179] The generation of a resource query request requires encapsulating the core information of the optimal candidate path. This information is extracted from the determined optimal candidate path, including the target department type (such as cardiology), the required doctor's specialty (such as coronary intervention), and the user's desired consultation time window (such as tomorrow morning). Together, these constitute the basic constraints for querying resources from other hospitals.

[0180] A regional healthcare consortium blockchain is a blockchain network jointly maintained by multiple hospitals as nodes. A pre-defined smart contract is a piece of automatically executable computer program code deployed on this chain, which encodes the collaborative rules for cross-hospital resource queries and appointments. Calling a smart contract means submitting a transaction containing a resource query request to the blockchain network.

[0181] Once the smart contract is triggered, it automatically broadcasts a query request to all registered collaborating hospital nodes on the blockchain. Each collaborating hospital node queries its local information system to check the availability of appointment slots in the target department and the doctors' expected capacity to see patients within the time window specified in the request, and returns a summary of available resources to the smart contract.

[0182] The smart contract has pre-defined coordination rules for matching and filtering multiple available resource summaries. The matching process mainly relies on the degree of consistency between the resources and the core information in the query request (department type, area of ​​expertise, and time window). After filtering, the smart contract sends a resource reservation request to the cooperating hospital node with the highest matching degree, aiming to lock in a portion of its appointment slots.

[0183] After receiving a reservation request, the collaborating hospital node performs the reservation operation in its local system. If successful, it returns a success response to the smart contract. Upon receiving the success response, the smart contract automatically generates a unique and tamper-proof encrypted digital credential, namely a referral commitment token. This token is bound on the blockchain to the collaborating hospital's detailed information (such as hospital code and appointment number) and serves as the basis for subsequent referral verification.

[0184] Finally, based on the successfully obtained referral commitment token and detailed information of collaborating hospitals, the system constructs a structured triage and guidance plan. This plan not only includes basic information and appointment details of the referral hospital, but also integrates the estimated waiting time at other hospitals, transportation route guidance, and explanations of potential cost differences compared to the in-hospital plan, providing users with a complete and feasible alternative treatment plan.

[0185] This embodiment dynamically identifies resource bottlenecks within the hospital through a departmental capacity prediction model, automatically triggering a regional resource collaboration mechanism based on blockchain smart contracts when overloaded. This mechanism leverages the automatic execution and immutability of smart contracts to achieve secure and efficient collaboration in cross-hospital resource querying, matching, reservation, and commitment, ultimately generating a concrete triage and guidance plan. This not only alleviates the instantaneous pressure on individual hospitals but also enhances the service resilience and accessibility of the overall healthcare system through regional resource pooling.

[0186] In some embodiments, the pre-defined smart contract matches and filters the available resources found based on pre-defined collaboration rules, and initiates a resource reservation request to the collaborative hospital node with the highest matching degree, including:

[0187] The core information in the smart contract parsing resource query request is extracted as matching constraints, including the target department type, the required doctor's expertise, and the expected consultation time window.

[0188] Obtain the available resource summary returned from each collaborating hospital node. The available resource summary should include at least the department matching identifier, the set of doctor specialty tags, the sequence of appointment slots to be released in the future time period, and the basic patient reception capacity score.

[0189] The first-level screening is performed based on the target department type and the department matching identifier, retaining collaborative hospital nodes that match the department type or have equivalent reception capabilities;

[0190] Based on the required physician specialty direction and the set of physician specialty tags of the retained collaborative hospital nodes, the specialty direction matching degree is calculated, and combined with the basic patient reception capacity score, the comprehensive capacity matching score of each collaborative hospital node is calculated.

[0191] Based on the expected appointment time window, an alignment analysis is performed on the future time slot release sequence returned by each collaborating hospital node to calculate the matching degree of appointment time that each collaborating hospital node can provide within the expected time window;

[0192] The overall ability matching score and the appointment time matching degree are weighted and integrated to generate a final matching degree score for each reserved node;

[0193] All retained collaborative hospital nodes are sorted according to the final matching score, and the collaborative hospital node with the highest score is selected as the target collaborative hospital node.

[0194] Send a resource reservation request to the target collaborative hospital node, which includes a specific appointment time option. The resource reservation request includes a unique on-chain identifier for this collaborative transaction generated by the smart contract.

[0195] In this embodiment, the available resource summary returned by each collaborating hospital node adopts a standardized format. The department matching identifier is used to determine whether the node has the same or equivalent reception capacity as the target department. The doctor specialty tag set lists the diseases or technical fields that the relevant doctors are good at. The future time slot appointment release sequence displays the available appointment slots in the form of a time list. The basic reception capacity score is a comprehensive capacity value calculated based on indicators such as historical service data, equipment configuration, and doctor-nurse ratio.

[0196] The smart contract first performs a preliminary screening based on the department matching identifier, retaining only nodes that fully match the department type or are identified by preset rules as having equivalent reception capabilities, while excluding irrelevant institutions.

[0197] For nodes that pass the initial screening, the smart contract calculates their comprehensive capability matching score. This score is derived from a fusion of two dimensions: first, the degree of matching in terms of expertise, calculated by comparing the overlap or semantic similarity between the required expertise and the node's set of doctor expertise tags; and second, the node's basic patient reception capability score. The two are combined according to preset weights to form a quantitative indicator reflecting the node's professional service capabilities.

[0198] Simultaneously, the smart contract calculates the appointment time matching degree for each node. By comparing the future time slot release sequence of a node with the expected appointment time window, it assesses the likelihood of providing services within the user's expected time. For example, it can calculate the proportion of available appointments within the window, or find the appointment closest to the expected time and calculate the matching degree based on the time difference.

[0199] The smart contract calculates a final matching score by weighting the overall capability matching score of each node with the appointment time matching score according to preset weights. The weight configuration can reflect the emphasis on different factors, such as giving more attention to professional capabilities or emphasizing time convenience.

[0200] All retained nodes are sorted according to their final matching score, and the one with the highest score is selected as the target collaborative hospital node.

[0201] Finally, the smart contract sends a resource reservation request to the target node. This request includes a specific reservation time option selected from the target node's source sequence, as well as an on-chain identifier generated by the smart contract to uniquely identify this collaborative transaction. This on-chain identifier will be used throughout all subsequent collaborative steps to ensure the integrity and traceability of the transaction.

[0202] This embodiment clarifies the core matching logic of smart contracts in cross-hospital resource collaboration. Through standardized resource descriptions, multi-level screening mechanisms, and multi-dimensional quantitative scoring models, it automatically and objectively selects the optimal resource provider from numerous collaborative nodes. This embodiment not only improves the accuracy and efficiency of resource matching but also reduces human intervention through algorithmic rules, ensuring fairness and transparency in the regional medical resource collaboration process.

[0203] In some embodiments, in response to a user's confirmation instruction for the online review service, symptom description information, differential diagnosis set, and hierarchical triage path diagram are encapsulated into a structured consultation form, including:

[0204] Receive confirmation instructions for online review service triggered by users through the interactive interface;

[0205] Extract and integrate key symptom entities, symptom duration, and self-assessment information on severity from symptom description information;

[0206] Extract the name of each suspected disease diagnosis and its corresponding posterior probability value from the differential diagnosis set;

[0207] Extract the recommended preferred route details, alternative route details, and corresponding summary of the recommendation reasons from the tiered triage route map;

[0208] The key symptom entities, symptom duration, self-assessment information on severity, suspected disease diagnosis name and its posterior probability value, details of preferred and alternative pathways and summary of recommendation reasons are mapped and filled according to a preset structured data pattern;

[0209] In structured data models, allergy history and key past medical history are linked together from personal health record information;

[0210] The filled structured data is standardized, encoded, and serialized to generate a machine-readable, structured consultation form that conforms to medical information exchange standards.

[0211] In this embodiment, the user can trigger an online review service confirmation instruction by clicking the confirmation button on the self-service machine or associated application. The system receives the instruction and starts the consultation form packaging process.

[0212] When extracting and integrating symptom description information, a list of key symptom entities obtained through natural language processing is retrieved from the stored data. At the same time, the duration (e.g., "coughing for three days") and severity self-assessment information (e.g., selection of "mild", "moderate" or "severe") that the user may have entered when describing the symptoms are extracted. This information is then correlated and integrated to form a concise summary of the current symptoms.

[0213] Information is extracted from the differential diagnosis set, which involves reading the list of diseases sorted by probability in the set, obtaining the standardized name of each disease and its calculated posterior probability value, and forming the diagnostic hypothesis part.

[0214] Extracting information from the tiered patient guidance route map involves parsing the structured data within the map. This includes extracting detailed information about the preferred route, including the target department, hospital, appointment time, and estimated waiting time; and extracting similar information for alternative routes. Additionally, it's necessary to extract a summary of the recommendation reasons generated by the system for each route, such as "highest matching degree," "shortest waiting time," or "optimal overall score."

[0215] A predefined structured data schema is a predefined data template or pattern, such as using JSONSchema or a data structure defined according to medical information standards like HL7 FHIR. The data extracted in the preceding steps are mapped and populated one by one according to the predefined fields in this schema. For example, a list of key symptom entities is entered into the "Chief Complaint" field, the disease name and probability are entered into the "Preliminary Diagnosis" field, and the path details are entered into the "Recommended Medical Arrangement" field.

[0216] During the population process, the system also extracts the user's allergy history (such as drug allergies) and key past medical history (such as history of major surgery and chronic diseases) from the personal health record information, and associates this information with the corresponding "allergy history" and "past medical history" fields in the structured data schema, making the consultation form more complete.

[0217] The structured data objects that have been filled in are standardized and serialized. For example, the data objects are converted into strings that conform to JSON or XML format and their encoding is ensured to conform to common standards such as UTF-8. The final structured consultation form is generated. This consultation form is machine readable and its format conforms to the established medical information exchange standards, so that it can be directly pushed to the Internet hospital platform and parsed and processed by its system.

[0218] This embodiment extracts, integrates, maps, associates, and standardizes coding to transform unstructured symptom descriptions, probabilistic diagnostic lists, graphical path recommendations, and personalized health records into a standardized electronic document with a unified format, complete information, and direct online doctor review. This achieves seamless data connection from intelligent triage to manual review, improves the efficiency and accuracy of online review, and provides a key data bridge for forming a closed loop of online and offline services.

[0219] In a second aspect, this embodiment also provides a computer-readable storage medium storing computer program instructions thereon, which, when executed by a processor, implement the method described in the first aspect.

[0220] The computer program involved in this embodiment can be stored in a computer device readable storage medium, which includes, but is not limited to, disks, magnetic tapes, magnetic cards, floppy disks, flash memory, optical disks, optical cards, read-only memory (ROM), random access memory (RAM), erasable programmable ROM (EPROM), and electrically erasable programmable ROM (EEPROM), etc. It also includes other biological, physical, or chemical structures capable of performing similar or equivalent functions to the storage media listed above, such as DNA, RNA, proteins, and other units with information storage capabilities. In specific embodiments, the storage medium involved can be one of the above-mentioned media types or a combination of the above media types. In different embodiments, the computer program involved in the embodiment can be centrally stored in a single medium or distributed across multiple media. The memory containing the computer device readable storage medium can be non-volatile memory or random access memory. These computer device readable storage media can be built into the device or connected to the device involved in the embodiment as an external device or part of an external device. In some embodiments, the memory having a computer device readable storage medium is deployed locally; in other embodiments, the memory may be deployed remotely from the processor, for example, as a network-attached memory accessed via RF circuitry or an external port and a communication network, wherein the communication network may be the Internet, one or more intranets, a local area network (LAN), a wide area network (WLAN), a storage area network (SAN), or a suitable combination thereof, as long as computer device access to the memory is enabled. Furthermore, the computer program involved in the embodiments may be stored in plaintext / ciphertext form, or it may be designed as training data, integrated and recombined through model training and implicitly stored in the parameter states of a deep neural network or other machine learning model.

[0221] Please see Figure 5 In a third aspect, this embodiment also provides an electronic device 1, including a memory 11 and a processor 12, wherein the memory 11 is used to store one or more computer program instructions, wherein the one or more computer program instructions are executed by the processor 12 to implement the method described in the first aspect.

[0222] The processor described in this embodiment can be implemented by hardware, firmware, software, or a combination thereof. It can be a circuit, one or more of an application-specific integrated circuit (ASIC), a digital signal processor (DSP), a digital signal processing device (DSPD), a programmable logic device (PLD), a field-programmable gate array (FPGA), a central processing unit (CPU), a controller, a microcontroller, or a microprocessor. It also includes other physical, biological, or chemical structures that can implement the same or equivalent functions as the processors listed above, such as biological neurons, quantum computing units, DNA computing units, etc., so that the processor can execute some or all of the steps in the computer program or method involved in the various embodiments of this application, or any combination of the steps mentioned therein.

[0223] By adopting the above technical solutions, this invention differs from existing technologies and possesses the following beneficial effects: By integrating personal health records with real-time localized spatiotemporal epidemiological data to generate a probabilistic differential diagnosis set, the medical rationality and timeliness of preliminary diagnostic recommendations are significantly improved. Furthermore, by invoking a clinical pathway time-prognosis model and combining it with Monte Carlo simulation, the selection of different treatment paths is transformed into a quantitative assessment of expected health gains and losses and risks, realizing a shift from experience-based judgment to scientific decision-making guided by health outcomes. Based on this, the method constructs a multi-objective optimization function that integrates path matching degree, doctor quality, waiting time, and treatment costs. Combining user preferences with real-time hospital resource data, it generates highly personalized treatment path recommendations, effectively balancing medical professionalism, user subjective desires, and real-world resource constraints. When the hospital's resources are overloaded, the system can automatically trigger regional medical resource collaborative query and matching based on consortium blockchain smart contracts, generating feasible inter-hospital referral and triage schemes, thereby breaking through the resource bottleneck of single-point hospitals and improving the overall service elasticity and efficiency of the regional medical system. Ultimately, by generating a structured hierarchical triage path map and a structured consultation form that can be seamlessly pushed to the internet hospital platform, a complete service loop from intelligent triage and online doctor review to offline appointment is achieved. This invention systematically solves the problems of static diagnostic criteria, decoupling of path selection from health prognosis, limited resource allocation scope, and broken service chains in existing triage technologies, achieving a comprehensive improvement in the accuracy, personalization, collaboration, and closed-loop nature of the triage process.

[0224] Finally, it should be noted that although the above embodiments have been described in the text and drawings of this application, this should not limit the scope of patent protection of this application. Any technical solutions that are based on the essential concept of this application and utilize the content described in the text and drawings of this application, resulting in equivalent structural or procedural substitutions or modifications, as well as the direct or indirect application of the technical solutions of the above embodiments to other related technical fields, are all included within the scope of patent protection of this application.

Claims

1. A smart triage method for medical self-service machines, characterized in that, include: Obtain symptom descriptions input by users, personal medical treatment preferences, and authorized personal health records; Based on the symptom description information and personal health record information, combined with the real-time access to the localized spatiotemporal epidemiological prior probability model, a differential diagnosis set containing at least one suspected disease diagnosis and its corresponding probability is generated. Based on the differential diagnosis set, a preset clinical pathway time-prognosis model is invoked to perform Monte Carlo simulation on the expected health gains and losses of each suspected disease diagnosis under different initial medical treatment pathways, generating a risk-benefit quantitative profile for each candidate pathway. Based on the personal medical preference information, a multi-objective function is constructed with path matching degree, doctor quality score, expected waiting time and medical cost as optimization objectives. Combined with real-time hospital resource status data, a multi-objective optimization algorithm is used to calculate a personalized recommendation score for each candidate path, and the optimal candidate path and the optimal personalized recommendation score are selected. If the real-time capacity prediction value of the department corresponding to the optimal candidate path exceeds the preset diversion threshold, a regional medical resource collaborative query will be triggered. Based on the consortium blockchain smart contract, the available resource commitments of the collaborating hospitals will be obtained, and a diversion and guidance plan containing cross-hospital referral options will be generated. Based on the optimal candidate path and triage plan, a structured hierarchical triage path map is generated and output to the user; In addition, in response to the user's confirmation instruction for the online review service, the symptom description information, differential diagnosis set and hierarchical triage path map are packaged into a structured consultation form and pushed to the Internet hospital platform to start the closed-loop service of online doctor review and offline appointment. Specifically, based on the differential diagnosis set, a preset clinical pathway time-prognosis model is invoked to perform Monte Carlo simulations on the expected health gains and losses for each suspected disease diagnosis under different initial medical pathways, generating a quantitative risk-benefit profile for each candidate pathway, including: The sampling weights for the Monte Carlo simulation are determined based on the posterior probability of each disease in the differential diagnosis set. For each suspected disease diagnosis, a corresponding clinical pathway time-prognostic model is loaded. The clinical pathway time-prognostic model defines the expected health status transition function and time delay cost function after the disease receives different levels of medical intervention at different time points. Based on the real-time resource status data of the current hospital and regional collaborative hospitals, multiple initial medical paths, including different first-visit departments, different medical institutions, and different appointment times, are constructed as a set of candidate paths. The differential diagnosis set is randomly sampled using the sampling weights to simulate and determine the actual disease diagnosis of this sampling, and a medical treatment path from the candidate path set is randomly assigned. Based on the clinical pathway time-prognosis model corresponding to the real disease diagnosis, and combined with the time delay of each link in the assigned medical treatment path, the simulated health loss value along the medical treatment path to the preset assessment time point is calculated. Repeat the random sampling, path allocation, and profit and loss calculation process to reach a preset number of simulations, and statistically analyze the health profit and loss value distribution of all simulation results corresponding to each candidate path; Based on the health profit and loss value distribution, the expected health profit and loss mean and risk variance of each candidate path are calculated, and the resource consumption estimate of the path is integrated to generate a risk-return quantitative profile to characterize the overall benefits and uncertainties of the path.

2. The intelligent triage method for medical self-service machines according to claim 1, characterized in that, Based on the symptom description information and personal health record information, combined with a real-time access to a localized spatiotemporal epidemiological prior probability model, a differential diagnosis set is generated, containing at least one suspected disease diagnosis and its corresponding probability, including: Natural language processing is performed on the symptom description information to extract a set of key symptom entities; The key symptom entity set is correlated with the historical diagnosis records and drug allergy history in the personal health record information to generate a preliminary symptom-medical history association map. Based on the preliminary symptom-medical history association graph, query the medical knowledge graph to obtain a set of potential diseases that are directly or indirectly associated with the set of key symptom entities. Based on the current time and geographic location information, obtain the spatiotemporal adjustment factors corresponding to each disease in the potential disease set from the localized spatiotemporal epidemiology prior probability model; Using the Bayesian inference algorithm, the spatiotemporal adjustment factor is used as a prior probability correction term. Combined with the matching degree between the key symptom entity set and the symptom manifestation of each disease, the posterior probability of each potential disease is calculated. The potential disease set is sorted according to the posterior probability, and diseases with probabilities higher than a preset threshold are selected to form the differential diagnosis set. Each disease in the set is labeled with its corresponding posterior probability value.

3. The intelligent triage method for medical self-service machines according to claim 1, characterized in that, For each suspected disease diagnosis, a corresponding clinical pathway time-prognostic model is loaded. This model defines the expected health status transition function and time delay cost function after receiving different levels of medical intervention at different time points for the disease, including: Based on the disease code of the suspected disease diagnosis, the corresponding standard diagnosis and treatment protocol framework is retrieved from the pre-built clinical pathway knowledge base; Based on the aforementioned standard diagnosis and treatment protocol framework, multiple key decision-making time nodes and a set of optional medical intervention measures are defined at each key decision-making time node. The medical intervention measures include at least the selection of the first-visit department, examination and testing items, consultation mechanism, and treatment plan. For each medical intervention, a health status transfer function is associated. The health status transfer function takes the health status score before the intervention and key physiological parameters in the patient's personal health record as inputs, and outputs the expected change in the health status score after the intervention. For each pair of adjacent critical decision time nodes, a time delay cost function is defined. The time delay cost function takes the length of the time interval and the urgency level of the suspected disease diagnosis as input, and outputs the cumulative loss of expected health status score due to delay. The health status transfer function and the time delay cost function are integrated by time-series logic to construct the clinical pathway time-prognosis model. The clinical pathway time-prognosis model can simulate and calculate the overall health gain or loss value up to a preset assessment time point based on any given initial health status, medical visit time series and intervention measure series.

4. The intelligent triage method for medical self-service machines according to claim 1, characterized in that, Based on the individual's medical treatment preference information, a multi-objective function is constructed with path matching degree, doctor quality score, estimated waiting time, and medical treatment cost as optimization objectives. Combined with real-time hospital resource status data, a multi-objective optimization algorithm is used to calculate a personalized recommendation score for each candidate path, and the optimal candidate path and optimal personalized recommendation score are selected, including: The personal medical preference information is analyzed to extract the user's preference weight coefficients for waiting time, doctor level, cost and distance to medical treatment; For each candidate path, a path matching sub-score is calculated based on the corresponding target department and doctor information. The path matching sub-score is determined based on the correlation between the differential diagnosis set and the target doctor's area of ​​expertise. For each candidate path, the doctor's comprehensive quality score is obtained based on the corresponding doctor information and used as a sub-score of the doctor's quality score. The comprehensive quality score integrates historical efficacy data and patient evaluations. For each candidate path, the estimated waiting time is calculated based on the current queue length in the hospital's real-time resource status data and the prediction model, and then converted into a waiting time sub-score. For each candidate path, the cost of seeking medical treatment is estimated based on the corresponding medical institution, department level, and doctor level information, and then converted into a cost of seeking medical treatment sub-score. Using the path matching degree sub-score, doctor quality score sub-score, waiting time sub-score, and consultation cost sub-score as basic variables, and the corresponding preference weight coefficients extracted from the individual consultation preference information as weighting coefficients, a multi-objective function in the form of a linear weighted sum is constructed. A multi-objective optimization algorithm is used to solve the multi-objective function, and a comprehensive personalized recommendation score is calculated for each candidate path. All candidate paths are sorted according to the personalized recommendation score, and the candidate path with the highest score is determined as the optimal candidate path, and its score is recorded as the optimal personalized recommendation score.

5. The intelligent triage method for medical self-service machines according to claim 1, characterized in that, If the real-time capacity prediction of the department corresponding to the optimal candidate path exceeds the preset triage threshold, a regional medical resource collaborative query is triggered. Based on the consortium blockchain smart contract, the available resource commitments of collaborating hospitals are obtained, and a triage and guidance plan including cross-hospital referral options is generated, including: The real-time operation data of the target department of the hospital corresponding to the optimal candidate path is obtained and input into the department capacity prediction model to calculate the predicted saturation within a future preset time period. Determine whether the predicted saturation exceeds a preset diversion threshold; If the number of cases exceeds the limit, a resource query request containing the core information of the optimal candidate path will be generated. The core information includes at least the target department type, the required doctor's specialty, and the expected consultation time window. The resource query request is used as a trigger condition to invoke a preset smart contract deployed on the regional medical consortium blockchain; When the preset smart contract is executed, it automatically queries each collaborating hospital node on the chain for available resource summaries that match the core information. The resource summaries include the availability of appointment slots and the expected capacity to receive patients. The preset smart contract, based on pre-defined collaboration rules, matches and filters the available resources found in the query, and initiates a resource reservation request to the collaborative hospital node with the highest matching degree. After receiving a successful resource reservation response from the collaborating hospital node, the smart contract generates an encrypted referral commitment token and binds the token to the detailed information of the collaborating hospital. Based on the referral commitment token and detailed information of the collaborating hospitals, a triage and guidance plan is constructed that includes cross-hospital referral options. The triage and guidance plan clearly indicates the name, address, estimated waiting time, transportation guidance, and explanation of cost differences of the referral hospital.

6. The intelligent triage method for medical self-service machines according to claim 5, characterized in that, The pre-defined smart contract, based on pre-set collaboration rules, matches and filters the available resources found in the query, and initiates a resource reservation request to the collaborative hospital node with the highest matching degree, including: The preset smart contract parses the core information in the resource query request and extracts the target department type, the required doctor's specialty, and the expected consultation time window as matching constraints. Obtain the available resource summary returned from each collaborating hospital node. The available resource summary shall include at least the department matching identifier, the doctor's specialty tag set, the future time period appointment release sequence, and the basic patient reception capacity score. The first-level screening is performed based on the target department type and the department matching identifier, retaining collaborative hospital nodes that match the department type or have equivalent reception capabilities; Based on the required doctor's specialty direction and the set of doctor's specialty tags of the retained collaborative hospital nodes, the specialty direction matching degree is calculated, and combined with the basic patient reception capacity score, the comprehensive capacity matching score of each collaborative hospital node is calculated. Based on the expected appointment time window, an alignment analysis is performed on the future time slot appointment release sequence returned by each collaborating hospital node to calculate the matching degree of appointment time that each collaborating hospital node can provide within the expected time window; The comprehensive capability matching score and the appointment time matching degree are weighted and fused to generate a final matching degree score for each reserved node; Based on the final matching score, all retained collaborative hospital nodes are sorted, and the collaborative hospital node with the highest score is selected as the target collaborative hospital node. A resource reservation request containing specific appointment time options is initiated to the target collaborative hospital node. The resource reservation request includes a unique on-chain identifier for this collaborative transaction generated by the smart contract.

7. The intelligent triage method for medical self-service machines according to claim 1, characterized in that, In response to the user's confirmation instruction for the online review service, the symptom description information, differential diagnosis set, and hierarchical triage path diagram are encapsulated into a structured consultation form, including: Receive confirmation instructions for online review service triggered by users through the interactive interface; Extract and integrate key symptom entities, symptom duration, and self-assessment information of severity from the symptom description information; Extract the name of each suspected disease diagnosis and its corresponding posterior probability value from the differential diagnosis set; Extract the recommended preferred route details, alternative route details, and corresponding summary of the recommendation reasons from the tiered triage route map; The key symptom entities, symptom duration, severity self-assessment information, suspected disease diagnosis name and its posterior probability value, preferred and alternative path details and recommendation reason summary are mapped and filled according to a preset structured data pattern; In the structured data model, allergy history and key past medical history extracted from the personal health record information are linked; The filled structured data is standardized, encoded, and serialized to generate a machine-readable, structured consultation form that conforms to medical information exchange standards.

8. A computer-readable storage medium storing computer program instructions thereon, characterized in that, The computer program instructions, when executed by a processor, implement the method as described in any one of claims 1 to 7.

9. An electronic device comprising a memory and a processor, characterized in that, The memory is used to store one or more computer program instructions, wherein the one or more computer program instructions are executed by the processor to implement the method as described in any one of claims 1 to 7.

Images