Remote medical service method and system based on past medical records

Abstract

The invention relates to the technical field of remote medical services, in particular to a remote medical service method and system based on past medical records. The system comprises an application layer, a service layer, a resource layer, an exchange layer and a data acquisition layer. The application layer is used for providing remote medical services; the service layer is used for providing registration service, application service and remote video service of the system; the resource layer is a data storage center and a data processing center; the exchange layer exchanges data with the outside; the exchange layer stores externally acquired electronic medical record data in the resource layer, after the service layer receives a remote medical service initiated by a patient, the electronic medical record data of the patient is analyzed and processed, a corresponding expert is matched for consultation of the patient, and through deep analysis and processing of the electronic medical record data, the remote medical service initiated by the patient is obtained. Targeted medical services can be provided for patients, remote consultation, monitoring and other services can be adjusted according to specific conditions of the patients, and medical services better meeting personal requirements are provided.

Description

Technical Field

[0001] The present invention relates to the technical field of telemedicine services, and particularly to a method and system for telemedicine services based on past medical records. Background Art

[0002] The medical service system is a medical data sharing system platform realized based on big data technology and cloud platform storage and sharing technology. The development and application of the medical service system can, to a certain extent, narrow the differences in medical conditions and medical resources between regions. Through cloud platform technology, remote medical services can be realized, and medical information resources can be obtained. It can not only provide better medical service conditions for patients, but also help professional medical staff improve their medical service capabilities.

[0003] The existing Chinese patent with the application number CN202311591000.9 discloses a remote medical consultation system, including: an application operation terminal, an administrator terminal, an expert operation terminal, and a medical institution terminal. The application operation terminal, the administrator terminal, the expert operation terminal, and the medical institution terminal are signal-connected through a network communication server, and the application operation terminal, the administrator terminal, the expert operation terminal, and the medical institution terminal are used for real-time dynamic consultation and non-real-time dynamic consultation. The present invention has the following advantages: After the implementation of the remote medical consultation system, the hardware cost is reduced, the work efficiency of medical staff and information department staff is improved, the time for preparing medical records is reduced, and the problem of unreasonable allocation of traditional medical resources; However, its medical records need to be viewed by experts, resulting in an extended consultation time and an increased workload for experts. For this reason, the present invention proposes a method and system for telemedicine services based on past medical records. Summary of the Invention

[0004] The present invention aims at the technical problems existing in the prior art and provides a method and system for telemedicine services based on past medical records.

[0005] The technical solution of the present invention to solve the above technical problems is as follows: A telemedicine service system based on past medical records includes an application layer, a service layer, a resource layer, an exchange layer, and a data collection layer; Application layer: Used to provide telemedicine services, including but not limited to remote consultation, remote specialist diagnosis, remote outpatient service, remote ward round, and remote monitoring; Service layer: Provides system registration services, application services, and remote video services; Resource layer: Is the storage center and processing center of data; Exchange layer: Exchanges data with the outside; The exchange layer stores the externally obtained electronic medical record data in the resource layer. After the service layer receives the remote medical service initiated by the patient, it analyzes and processes the patient's electronic medical record data and matches the corresponding experts for consultation on the patient.

[0006] Furthermore, in the described remote medical service system based on past medical records, a disease classification knowledge graph is constructed in the resource layer; The knowledge graph includes a concept layer and a data layer. The concept layer is constructed from international and domestic standard medical terminology sets, as well as case annotation and analysis. At the same time, clear normative descriptions are given to the concepts and the relationships between the concepts. After medical experts evaluate the concept system and relationships, the concept layer is formed; The data layer includes, based on the concept layer, descriptions of disease-related knowledge from multiple sources, including structured, semi-structured, and unstructured texts. Entities and their relationships are labeled in a semi-automatic or automatic manner. After fusing the entities and their relationships from multiple sources, the data layer is formed.

[0007] Furthermore, in the described remote medical service system based on past medical records, entity recognition and automatic extraction of their relationships are included in the data layer; During the process of entity recognition, two types of strict S and relaxed R metrics are used to evaluate the results. The automatically recognized result is denoted as and the gold standard result is denoted as ; In the strict metric, when and only when the start position, end position, and category of an entity are exactly the same as the gold standard, that is, , on this basis, the strict intersection , the strict metric precision , the recall rate and the recognition accuracy are defined as: ; ; ; In the relaxed metric, when and only when the start position of an entity with the gold standard ≤ the end position of the entity and the category is the same, on this basis, the strict intersection , the strict metric precision , the recall rate and the recognition accuracy are defined as: ; ; .

[0008] Further, in the described remote medical service system based on past medical records, the electronic medical record data of the patient is read in the resource layer, and the electronic medical record data is processed to obtain recognizable symptoms and diseases, and a link relationship is established with the nodes in the knowledge graph. The matching relationship with the nodes in the knowledge graph is determined by means of similarity calculation, and for a given recognized entity , the n most similar entities in the knowledge graph are calculated according to the symptom and disease categories, forming a candidate entity set for the entity , calculate the similarity between the entity and , and select the entity with the highest similarity as the entity matched by the knowledge graph; Calculate the similarity according to the edit distance, Jaccard coefficient, and longest common substring: ; ; ; where lev represents the entity edit distance, bigram represents the calculation of the Jaccard coefficient, represents the string length, represents the longest common substring between two entities; The average value of the three is used to represent the similarity Sim between two entities: .

[0009] Further, in the described remote medical service system based on past medical records, after entity linking, a symptom set matched by the electronic medical record and the knowledge graph and a disease set including frequency statistics and sorting are obtained; and , the weights of diseases are predicted by comprehensively considering the symptom set and the disease set; In the knowledge graph, match the symptom in the knowledge graph to obtain a set including n candidate diseases. The disease candidate sets corresponding to all symptoms are denoted as D. For a disease in the candidate set, there are m symptom sets associated with it in the knowledge graph. Use to represent the set where the symptoms of the current electronic medical record match the symptoms of a disease in the knowledge graph, and obtain: ; The disease weight of the electronic medical record is calculated through weight calculation, considering the symptoms included in a disease candidate set and also considering the distribution of a symptom in all diseases in the knowledge graph. Based on the joint probability of the two, the disease Calculate the weight: ; Among them and respectively represent and the number of diseases included in D, is the probability that symptom and disease appear together, represents the sum of the probabilities that all symptoms in the candidate set and disease appear together.

[0010] Furthermore, for the remote medical service system based on past medical records described above, for the disease set appearing in the electronic medical record, if a disease negative factor appears, it will be deleted from the candidate set; otherwise, if the candidate set associated through symptoms already includes , then its weight value will be increased: ; If the candidate set associated through symptoms does not include , then it will be added to the disease candidate set, and its weight is times the average weight.

[0011] Furthermore, for the remote medical service system based on past medical records described above, after obtaining the corresponding disease weights in the patient's electronic medical record, according to the disease weight ranking, the top three disease types are selected to match the corresponding experts for consultation of the patient.

[0012] Furthermore, for the remote medical service system based on past medical records described above, during the matching process, match according to the appointment quantity and video time of the experts; Meet the constraint conditions: There is an exclusive requirement for experts, that is, the same expert cannot be matched to more than one patient at the same time. For expert k, there is only one patient before and after being assigned to each patient. Suppose patient a has seen expert k, then there is Suppose is a binary variable. When and only when both patient a and patient b are assigned to expert k, and patient a exactly precedes patient b, it is equal to 1, otherwise it is equal to 0. The constraint condition is: ; ; There are time requirements for patients, that is, the higher the satisfaction of patients is, the faster they can receive medical services from experts. To cooperate with the expert allocation to achieve shorter waiting times for patients, the number of expert appointments is c, and the medical service time is t. The waiting time of patients is: ; k1, k2, and k3 are three experts respectively, is the average waiting time.

[0013] Furthermore, the described remote medical service system based on past medical records includes: converting the constraint conditions into a mathematical model of AGAP, and solving it through a genetic algorithm to obtain the medical service time.

[0014] A remote medical service method based on past medical records, which is implemented through a remote medical service system based on past medical records. The method includes the following steps: S1: The patient initiates a remote medical service application in the service layer; S2: The system obtains the patient's electronic medical record through the exchange layer and stores it in the resource layer; S3: In the resource layer, entity extraction is performed on the electronic medical record according to the knowledge graph to obtain the patient's symptoms and disease types in the electronic medical record; S4: Match the video time of the expert according to the symptoms and disease types; S5: After the appointment time arrives, initiate a video service to the patient and the expert, and implement remote medical services in the application layer.

[0015] The beneficial effects of the present invention are: Through in-depth analysis and processing of electronic medical record data, targeted medical services can be provided for patients. Services such as remote consultation and monitoring can be adjusted according to the specific situation of patients, providing medical services that better meet individual needs; Using the disease classification knowledge graph can accurately and quickly identify and classify symptoms and diseases, improving the accuracy of disease diagnosis; at the same time, the introduction of the knowledge graph reduces the time for doctors to manually analyze medical record data, improving the efficiency of medical services; The system can intelligently match the most suitable expert team according to the disease weight of the patient, ensuring the reasonable allocation of medical resources and improving the pertinence and effect of treatment; Using genetic algorithms and mathematical models to optimize the diagnosis and treatment process can enable the system to dynamically adjust and optimize in terms of resource allocation, service time arrangement, etc., thereby improving the satisfaction of patients. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 is a schematic diagram of a remote medical service system based on past medical records of the present invention; Figure 2 This is a schematic flowchart of a remote medical service method based on past medical records according to the present invention.

[0017] Figure 3 This is a schematic diagram of the construction process of a knowledge graph according to an embodiment of the present invention; Figure 4 This is a schematic diagram of entity recognition based on rules and machine learning according to an embodiment of the present invention; Figure 5 This is a schematic diagram of the framework of a remote medical service system according to an embodiment of the present invention. Detailed implementation manners

[0018] Next, the technical solutions in the embodiments of the present application will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative efforts belong to the scope of protection of the present application.

[0019] In the description of the present application, the terms "first" and "second" are only used for descriptive purposes and cannot be construed as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features. Thus, the features defined with "first" and "second" may explicitly or implicitly include one or more of the described features. In the description of the present application, "a plurality" means two or more, unless otherwise specifically defined.

[0020] In the description of the present application, the term "for example" is used to mean "serving as an example, illustration, or explanation". Any embodiment described as "for example" in the present application is not necessarily construed as being more preferred or having more advantages than other embodiments. In order for any person skilled in the art to implement and use the present invention, the following description is given. In the following description, details are set forth for purposes of explanation. It should be understood that those skilled in the art can recognize that the present invention can be implemented without using these specific details. In other instances, well-known structures and processes are not elaborated in detail to avoid unnecessary details from obscuring the description of the present invention. Therefore, the present invention is not intended to be limited to the embodiments shown, but is consistent with the broadest scope that conforms to the principles and features disclosed in the present application.

[0021] A remote medical service system based on past medical records includes an application layer, a service layer, a resource layer, an exchange layer, and a data collection layer; Application layer: used to provide remote medical services, including but not limited to remote consultation, remote specialist diagnosis, remote outpatient service, remote ward round, and remote monitoring; Service layer: provides system registration services, application services, and remote video services; Resource layer: is the storage center and processing center of data; Exchange layer: conducts data exchange with the outside; The exchange layer stores the electronic medical record data obtained from the outside in the resource layer. After the service layer receives the remote medical service initiated by the patient, it analyzes and processes the patient's electronic medical record data and matches the corresponding experts for consultation on the patient.

[0022] In this embodiment, the application layer of the remote collaborative diagnosis and treatment system consists of two parts: remote diagnosis and treatment service application and remote diagnosis and treatment service supervision; access the remote collaborative diagnosis and treatment system through a unified remote collaborative diagnosis and treatment service portal website to carry out appointment, application, and diagnosis and treatment services. Through the remote diagnosis and treatment service supervision module, various supervision functions such as analysis, statistics, and decision-making of the operation conditions of the remote collaborative diagnosis and treatment systems at all levels can be realized; The service layer of the remote collaborative diagnosis and treatment system provides registration services, remote services, storage services, electronic medical record filing services, etc., for directly interacting with the remote diagnosis and treatment business logic layer through the remote diagnosis and treatment data transmission object, concentrating the system business logic processing. The message exchange and message transmission between services run through each service layer; The resources provided by the resource layer of the remote collaborative diagnosis and treatment system include structured data, unstructured data, structured document data, application service resources, etc., mainly used to support the management and coordination of the cross-regional remote diagnosis and treatment service work, support the cross-regional remote diagnosis and treatment service work, auxiliary decision-making work, and the development of data statistical analysis services; The exchange layer of the remote collaborative diagnosis and treatment system includes enterprise service bus ESB, service integration SMP, general file transfer GTP, data integration ETL, unified media control unit MCU. The exchange layer is mainly used to meet the cross-hospital and cross-regional information exchange and collaborative applications of clinical information, for the registration, application, authorization, management, and monitoring of medical service resources, to achieve service-based information resource sharing and exchange, to meet the collection of business information based on the data specifications of the health care industry, and to provide file-based data exchange services for external systems, to meet the data integration requirements of data collection, processing, and conversion during the construction of the remote collaborative diagnosis and treatment data warehouse.

[0023] In one embodiment, a disease classification knowledge graph is constructed in the resource layer; The knowledge graph includes a concept layer and a data layer. The concept layer is constructed from international and domestic standard medical terminology sets, as well as case annotation and analysis. At the same time, clear specification explanations are given to the concepts and the relationships between the concepts. After medical experts evaluate the concept system and relationships, the concept layer is formed; The data layer is based on the concept layer and obtains descriptions of disease-related knowledge from multiple sources, including structured, semi-structured, and unstructured texts. It uses semi-automatic or automatic methods to label entities and their relationships. After fusing entities and their relationships from multiple sources, the data layer is formed.

[0024] The knowledge graph can be formally expressed as , where the set represents entities, events, or attributes in the knowledge graph, and the relationship set represents that there are n types of relationships between entities. The triple relationship set , expressed as data form, is , where e1 is the head entity, e2 is the tail entity, and r represents their relationship. The above knowledge graph describes the symptoms, treatments, and detection data involved in the form of triples according to relevant descriptions such as disease symptoms, characteristics, and detection data. The specific construction process is as Figure 3 shown: In another embodiment, as Figure 4 shown, entity recognition and relationship extraction are performed using both rule-based and machine learning methods. After data integration and deduplication, structured triples are obtained. In the rule-based method, first, by inducting semi-automatically labeled corpora, words or phrases that often appear between the same relationships are automatically extracted as relationship marker patterns, and part of the rule description is in BNF format for automatic extraction. Second, the text is classified, and different rules are used for extraction according to categories. The text from different sources is analyzed, and corresponding extraction rules are designed. Triples in semi-structured and structured texts are extracted according to the defined rules, and the extracted relationships are matched with relevant relationships according to the knowledge description system. In machine learning, the experiment uses manually labeled clinical practice texts for common diseases as the training set, and semi-automatically labeled textbooks and disease-related clinical practice texts as the test set, and triples are obtained through training in an 8:2 ratio.

[0025] In the data layer, entity recognition and automatic relationship extraction are performed through an entity recognition model. During the entity recognition process, strict S and relaxed R metrics are used to evaluate the results. The automatically recognized result is denoted as , and the gold standard result is denoted as ; In the strict metric, when and only when the start position, end position, and category of an entity are exactly the same as the gold standard, that is , based on this, the strict intersection , strict precision , recall and recognition accuracy are defined: ; ; ; In the relaxation index, a strict intersection is defined if and only if an entity has a start position with respect to the gold standard ≤ the end position of the entity and the categories are the same. , the strict standard accuracy , the recall rate and the recognition accuracy : ; ; 。

[0026] In one embodiment, the electronic medical record data of a patient is read in the resource layer, and the electronic medical record data is processed to obtain recognizable symptoms and diseases, and link relationships are established with the nodes in the knowledge graph. A matching relationship with the knowledge graph nodes is determined by a similarity calculation method. For a given recognized entity , the n most similar entities in the knowledge graph are calculated according to the symptom and disease categories, forming a candidate entity set for the entity , the similarity between the entity and is calculated, and the entity with the highest similarity is selected as the entity matched with the knowledge graph; The similarity is calculated according to the edit distance, Jaccard coefficient, and longest common substring: ; ; ; where lev represents the entity edit distance, bigram represents the calculation of the Jaccard coefficient, represents the string length, represents the longest common substring between two entities; The average value of the three is used to represent the similarity Sim between two entities: 。

[0027] After entity linking, a symptom set that matches the electronic medical record and the knowledge graph is obtained and a disease set that includes frequency statistics and sorting ; and , the weights of diseases are predicted by comprehensively considering the symptom set and the disease set; In the knowledge graph, for the symptoms​ Match in the knowledge graph to obtain a set containing n candidate diseases , and the set of disease candidates corresponding to all symptoms is denoted as D. For a disease in the candidate set , there are m symptom sets associated with it in the knowledge graph. Let represent the set where the symptoms of the current electronic medical record match the symptoms of a disease in the knowledge graph, and we get: ; Calculate the disease weight of the electronic medical record through weight calculation. It is necessary to consider the symptoms included in a disease candidate set, and also consider the distribution of a symptom in all diseases in the knowledge graph. Based on the joint probability of the two, calculate the weight for disease : ; where and represent the number of diseases included in and D respectively. is the probability that symptom and disease appear together. represents the sum of the probabilities that all symptoms in the candidate set and disease appear together.

[0028] For the set of diseases appearing in the electronic medical record, if a disease negation factor appears, delete it from the candidate set. Otherwise, if the candidate set associated through symptoms already contains , then increase its weight value: ; If the candidate set associated through symptoms does not contain , then add it to the disease candidate set, and its weight is times the average weight.

[0029] In this embodiment, assume that a patient's electronic medical record contains the following information: Past medical history: hypertension; Current symptoms: chest pain and shortness of breath; Now it is necessary to determine the possible current disease types of the patient based on these medical histories and symptoms.

[0030] In the resource layer of the system, these disease conditions and disease data of the patient are read and analyzed; the system will process the data using the following steps: Knowledge graph matching: Hypertension may be related to heart diseases (such as coronary artery disease), kidney diseases, etc.

[0031] Chest pain and shortness of breath may be symptoms of heart disease, lung problems (such as pulmonary embolism or pneumothorax).

[0032] Match similar symptoms and disease nodes: The system matches the symptoms in the electronic medical record with the disease nodes in the knowledge graph.

[0033] For chest pain and shortness of breath, the system may link them to disease nodes of heart diseases such as angina pectoris and heart failure.

[0034] Calculate the weights of candidate diseases: The system considers the addition of hypertension as a past medical history to the risk of heart disease.

[0035] At the same time, the relevance between hypertension-related complications in the knowledge graph and these symptoms is also analyzed.

[0036] Weighted evaluation and prediction: If the data in the system knowledge graph shows that when patients with hypertension experience chest pain and shortness of breath, there is a high probability of heart disease, such as angina pectoris or heart failure, then the weights of these diseases will be higher.

[0037] The system combines all this information and calculates the predicted weights for each possible disease for this patient.

[0038] Output prediction: Based on the predicted weights, the system may suggest angina pectoris or heart failure as the most likely disease types for the current visit.

[0039] In one embodiment, after obtaining the corresponding disease weights in the patient's electronic medical record, according to the sorting of the disease weights, the top three disease types are selected to match the corresponding experts for consultation with the patient.

[0040] Furthermore, in the described telemedicine service system based on past medical records, during the matching process, matching is performed according to the appointment quantity and video time of the experts; Meet the constraint conditions: There is an exclusive requirement for experts, that is, the same expert cannot be matched to more than one patient at the same time. For expert k, there is only one patient before and after being assigned to each patient. Suppose patient a has seen expert k, then there is , suppose is a binary variable. When and only when both patient a and patient b are assigned to expert k, and patient a exactly precedes patient b, it is equal to 1, otherwise it is equal to 0. The constraint condition is: ; ; There are time requirements for patients, that is, the faster they can obtain medical services from experts, the higher their satisfaction. To cooperate with the expert allocation to achieve shorter waiting times for patients, the number of appointments for experts is c, and the medical service time is t. The waiting time for patients is: ; k1, k2, and k3 are three experts respectively, is the average waiting time.

[0041] It includes: transforming the constraint conditions into a mathematical model of AGAP, and solving it through a genetic algorithm to obtain the medical service time.

[0042] In one embodiment, a remote medical service method based on past medical records includes the following steps: S1: The patient initiates a remote medical service application in the service layer; S2: The system obtains the patient's electronic medical record through the exchange layer and stores it in the resource layer; S3: In the resource layer, entity extraction is performed on the electronic medical record according to the knowledge graph to obtain the patient's symptoms and disease types in the electronic medical record; S4: Match the video time of the expert according to the symptoms and disease types; S5: After the appointment time arrives, initiate a video service to the patient and the expert, and implement remote medical services in the application layer.

[0043] In this embodiment, the patient creates a medical service request by accessing the remote medical service portal of the medical institution or using a mobile application. The patient can enter personal information, symptom descriptions, expected service types, etc., and select an appointment time. The remote medical service system of the medical institution interacts with other medical information systems through the exchange layer, automatically obtains the patient's electronic medical record data, and securely stores this data in the resource layer. The data may include historical medical records, test results, imaging materials, etc. The analysis tools in the resource layer will process the electronic medical record data, use natural language processing algorithms to perform entity recognition and relationship extraction on the text information in the medical record. The system uses the constructed disease classification knowledge graph to standardize the extracted entities (such as symptoms and disease types) and match them to the corresponding nodes in the knowledge graph. In the service layer, the system selects a suitable medical expert for matching according to the symptoms and disease type information provided by the patient, combined with the number of appointments and video time of the expert. The system can perform intelligent matching according to factors such as the expert's professional field, work schedule, and past case processing efficiency. When the appointment time arrives, the system will automatically create a secure video call to connect the patient and the matched medical expert. The medical expert evaluates the patient's situation through the video interface and provides corresponding diagnoses and suggestions. The video service can be carried out in real time or recorded by the system for subsequent review.

[0044] Although the preferred embodiments of the present invention have been described, those skilled in the art can make additional changes and modifications once they learn the basic inventive concept. Therefore, the appended claims are intended to be construed to include the preferred embodiments as well as all changes and modifications falling within the scope of the present invention.

[0045] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalent technologies, the present invention is also intended to include these modifications and variations.

Claims

1. A remote medical service system based on past medical records, characterized in that: Includes application layer, service layer, resource layer, exchange layer and data collection layer; Application layer: used to provide telemedicine services, including but not limited to teleconsultation, telespecialist diagnosis, teleoutpatient clinic, teleward rounds, and telemonitoring; Service layer: provides system registration services, application services and remote video services; Resource layer: It is the data storage and processing center; Exchange layer: exchanges data with the outside world; The exchange layer stores the electronic medical record data obtained from the outside in the resource layer. After the service layer receives the remote medical service initiated by the patient, it analyzes and processes the electronic medical record data of the patient and matches the corresponding experts to consult with the patient.

2. A remote medical service system based on past medical records according to claim 1, characterized in that: Constructing a disease classification knowledge graph in the resource layer; The knowledge graph includes a concept layer and a data layer. The concept layer is constructed from international and domestic standard medical terminology sets and case annotations and analysis. At the same time, clear and standardized descriptions are given to concepts and the relationships between concepts. Medical experts evaluate the concept system and relationships to form the concept layer. The data layer includes a description of disease-related knowledge obtained from multiple sources based on the concept layer, including structured, semi-structured and unstructured texts, and annotates entities and their entity relationships in a semi-automatic or automatic manner. After the entities and their relationships from multiple sources are integrated, a data layer is formed.

3. A remote medical service system based on past medical records according to claim 2, characterized in that: The data layer includes identifying entities through an entity recognition model and automatically extracting their relationships; In the process of entity recognition, two types of indicators, strict S and relaxed R, are used to evaluate the results. The results of automatic recognition are recorded as , the gold standard result is recorded as ; In strict metrics, an entity is considered to be a perfect entity if and only if its starting position, ending position, and category are exactly the same as those of the gold standard, i.e. , on this basis, we define strict intersection , strict indicator accuracy , recall rate and recognition accuracy : ; ； ； In the relaxed index, a strict intersection is defined if and only if the starting position of an entity and the gold standard are ≤ the ending position of the entity and the category is the same. , strict indicator accuracy , recall rate and recognition accuracy : ; ; 。 4. A remote medical service system based on past medical records according to claim 2, characterized in that: The patient's electronic medical record data is read in the resource layer, and the electronic medical record data is processed to obtain identifiable symptoms and diseases, and a link relationship is established between the electronic medical record data and the nodes in the knowledge graph. The matching relationship with the knowledge graph nodes is determined by a similarity calculation method. , calculate the n entities most similar to the knowledge graph according to symptoms and disease categories, and form The candidate entity set , computing entity and The similarity between them is used to select the entity with the highest similarity as the entity to be matched in the knowledge graph; Calculate similarity based on edit distance, Jaccard coefficient and longest common substring: ； ； ； Where lev represents the entity edit distance, bigram represents the calculation of the Jaccard coefficient, Indicates the length of the string. Represents the longest common substring between two entities; The average value of the three is used to measure the similarity Sim between two entities: 。 5. A remote medical service system based on past medical records according to claim 4, characterized in that: After entity linking, we get the symptom set that matches the electronic medical record and the knowledge graph. and a collection of diseases with frequency statistics and rankings ;and , comprehensively consider the symptom set and the disease set to predict the weight of the disease; In the knowledge graph, the symptoms Match in the knowledge graph to obtain a set of n candidate diseases , the candidate disease set corresponding to all symptoms is recorded as D, for a disease in the candidate set , there are m symptom sets in the knowledge graph Associated with it, Represents the set of symptoms in the current electronic medical record that match the symptoms of a disease in the knowledge graph, and obtains: ; The disease weight of the electronic medical record is obtained by weight calculation. The symptoms contained in the disease candidate set should be considered, and the distribution of a symptom in all diseases in the knowledge graph should also be considered. The disease is classified based on the joint probability of the two. Calculate weights: ； in and Respectively and the number of diseases contained in D, Symptoms in the knowledge graph and disease The probability of co-occurrence, Represents all symptoms and diseases in the candidate set The sum of the probabilities of co-occurrence.

6. A remote medical service system based on past medical records according to claim 5, characterized in that: For disease collections appearing in electronic medical records If a disease negation factor appears, it will be deleted from the candidate set. Otherwise, if the candidate set associated with the symptom already contains , then its weight value will be increased: ； If the candidate set associated by symptoms does not contain , then add it to the disease candidate set, and its weight is the average weight times.

7. A remote medical service system based on past medical records according to claim 6, characterized in that: After obtaining the corresponding disease weights in the patient's electronic medical record, the disease weights are sorted and the top three disease types are selected to match the corresponding experts for consultation with the patient.

8. A remote medical service system based on past medical records according to claim 7, characterized in that: During the matching process, the experts are matched based on their number of appointments and video call times; Satisfy the constraints: There is an exclusivity requirement for experts, that is, the same expert cannot be matched to more than one patient at the same time. For expert k, there is only one patient before and after each patient is assigned. If patient a is assigned to expert k, then ,set up is a binary variable, equal to 1 if and only if both patient a and patient b are assigned to expert k, and patient a is assigned to patient b just before patient b, otherwise equal to 0, the constraints are: ； ； Patients have time requirements, that is, the sooner they get medical services from experts, the higher their satisfaction. The shorter the waiting time of patients can be achieved by coordinating the allocation of experts. The number of expert appointments is c, the medical service time is t, and the patient's waiting time is: ； k1, k2, and k3 are three experts. is the average waiting time.

9. A telemedicine service system based on past medical records according to claim 8, comprising: The constraints are transformed into the mathematical model of AGAP, which is solved by genetic algorithm to obtain the time of medical service.

10. A remote medical service method based on past medical records, characterized in that: The method is implemented by a telemedicine service system based on past medical records as described in any one of claims 1 to 9, and the method comprises the following steps: S1: The patient initiates a telemedicine service application in the service layer; S2: The system obtains the patient's electronic medical record through the exchange layer and stores it in the resource layer; S3: In the resource layer, entities are extracted from the electronic medical records based on the knowledge graph to obtain the patient's symptoms and disease types in the electronic medical records; S4: Matching the video call time of experts according to symptoms and disease types; S5: Initiate video service to patients and experts after the appointment time arrives, and realize telemedicine service in the application layer.

Images