A diagnosis and surgery ICD coding method for medical records based on knowledge graph

Abstract

The present application provides a method for ICD coding of diagnosis and surgery of medical records based on knowledge graph, which performs semantic analysis on medical information to extract keywords corresponding to the medical information; determines the target nodes involved in the medical information based on the keywords; retrieves the pre-established knowledge graph of ICD coding, matches the target node to the multiple medical nodes, to determine at least one medical path matched by the target node; and outputs the at least one medical path and its corresponding ICD code. The technical solution of the method for ICD coding of diagnosis and surgery of medical records based on knowledge graph provided by the present application can accurately and quickly match the accurate ICD code for each patient's medical treatment process, greatly reducing the coding management cost and improving the user experience.

Description

Technical Field

[0001] The present application relates to the field of medical informationization, and specifically to a diagnosis and surgery ICD coding method for medical records based on a knowledge graph. Background Art

[0002] The International Classification of Diseases (ICD) is an international unified disease classification method developed by the WHO. It classifies diseases according to their etiology, pathology, clinical manifestations, and anatomical location, making them an orderly combination and expressing them in a coding system. ICD coding is a widely used management tool in national statistics, health commission management, hospital management, and morbidity statistics. For example, DRG / DIP (medical insurance payment) and public hospital performance appraisal are mainly managed based on ICD coding.

[0003] During the research process of conceiving and forming this application, the applicant discovered at least the following problems: doctors in hospitals need to describe in detail the patient's chief complaint, symptoms, diagnosis and treatment process, surgical records, etc. in the medical records during diagnosis and treatment. This is a description of the patient's treatment process in natural language. After the patient is discharged from the hospital, due to the need for statistical management, the medical record coder will ICD code the disease and surgery / operation in this medical record based on the above medical record information and the doctor's diagnosis / surgery description. The doctor uses medical terms to describe the disease / surgery in words, and the medical record coder performs ICD coding on the disease / surgery based on the ICD coding specifications. The two do not belong to the same system, which may cause a mismatch between the ICD codes of the actual disease and the surgery. Summary of the invention

[0004] In order to solve the above problems, this application provides a diagnosis and surgery ICD coding method for medical records based on knowledge graph, including:

[0005] S10: Establish a knowledge graph of disease and surgery ICD codes, wherein the knowledge graph of ICD codes includes a diagnosis and treatment pathway corresponding to each ICD code, and each diagnosis and treatment pathway includes multiple diagnosis and treatment nodes;

[0006] S20: in response to obtaining the diagnosis and treatment information to be matched, performing semantic analysis on the diagnosis and treatment information to extract keywords corresponding to the diagnosis and treatment information;

[0007] S30: Determine the target node involved in the diagnosis and treatment information based on the keyword;

[0008] S40: Retrieving the knowledge graph of the ICD code, matching the target node to the multiple diagnosis and treatment nodes, so as to determine at least one diagnosis and treatment pathway matched by the target node;

[0009] S50: Output the at least one diagnosis and treatment pathway and its corresponding ICD code.

[0010] Optionally, the step of establishing an ICD coding knowledge graph, wherein the ICD coding knowledge graph includes a diagnosis and treatment pathway corresponding to each ICD code, and each diagnosis and treatment pathway includes a plurality of diagnosis and treatment nodes, includes:

[0011] Collect medical data and clean the medical data, including disease information, diagnosis and treatment process, drug information, surgical records, etc.;

[0012] Identifying named entities from the cleaned medical data and assigning an identifier to each named entity; the named entities are used to represent semantic objects in the medical data;

[0013] The association relationships between the named entities are analyzed, and the identified named entities and the association relationships are integrated into a preset knowledge framework to form the knowledge graph.

[0014] Optionally, the step of establishing an ICD coding knowledge graph, wherein the ICD coding knowledge graph includes a diagnosis and treatment pathway corresponding to each ICD code, and each diagnosis and treatment pathway includes a plurality of diagnosis and treatment nodes, further includes:

[0015] Performing knowledge verification on the knowledge graph, wherein the knowledge verification includes expert verification and data cross-validation;

[0016] Regularly collect user and expert opinions and update the knowledge graph.

[0017] Optionally, in response to obtaining the medical information to be matched, the step of performing semantic analysis on the medical information to extract keywords corresponding to the medical information includes:

[0018] Extract target keywords according to the diagnosis and treatment information to be matched, and mark the part of speech of each target keyword;

[0019] Based on the target keywords marked with the part of speech, the named entities in the diagnosis and treatment information are identified and classified and labeled.

[0020] Optionally, the step of determining the target node involved in the diagnosis and treatment information based on the keyword includes:

[0021] Determine the weight of each target keyword according to a preset strategy based on the importance of the target keyword in the diagnosis and treatment information;

[0022] Based on the weight of the target keyword and the classification identifier of the named entity, determining the semantic role of each named entity in the diagnosis and treatment information to determine the association relationship between the named entities;

[0023] Connect entity nodes with association relationships in sequence to build a semantic association path;

[0024] The diagnosis and treatment nodes in the semantic association path are extracted, and the target nodes involved in the diagnosis and treatment information are determined.

[0025] Optionally, the step of retrieving the knowledge graph of the ICD code and matching the target node to the multiple diagnosis and treatment nodes to determine at least one diagnosis and treatment pathway matched by the target node includes:

[0026] Performing a query operation from the knowledge graph of the ICD code to retrieve the diagnosis and treatment nodes in the knowledge graph of the ICD code, so that the target nodes are matched with the diagnosis and treatment nodes one by one;

[0027] When the target node is successfully matched, a search is performed along the relationship chain between named entities based on the knowledge graph of the ICD code to retrieve the diagnosis and treatment path connected to the target node;

[0028] The suitability of each retrieved diagnosis and treatment pathway is calculated, and at least one diagnosis and treatment pathway whose suitability value exceeds a preset threshold is screened; the suitability is used to characterize the degree of matching between the diagnosis and treatment pathway and the target node.

[0029] Optionally, the step of performing a query operation from the knowledge graph of the ICD code to retrieve the diagnosis and treatment nodes in the knowledge graph of the ICD code and matching the target nodes with the diagnosis and treatment nodes one by one includes at least one of the following:

[0030] Comparing the entity names of the target node and the diagnosis and treatment node to perform a name-based matching calculation;

[0031] Retrieving the category of the target node, and comparing the category of the target node with the category of the diagnosis and treatment node to perform a category-based matching calculation;

[0032] Retrieving the attributes of the target node, and comparing the attributes of the target node with the attributes of the diagnosis and treatment node to perform attribute-based matching calculation;

[0033] The semantic relationship between the target nodes is analyzed, and the semantic relationship between the target nodes is compared with the semantics of the diagnosis and treatment nodes to perform a semantic-based matching calculation.

[0034] Optionally, the step of calculating the suitability of each retrieved diagnosis and treatment pathway and screening at least one diagnosis and treatment pathway whose suitability value exceeds a preset threshold comprises:

[0035] querying a medical database to determine association frequencies between diagnosis and treatment nodes in the semantic association path;

[0036] Retrieving case data from historical diagnosis and treatment data, and analyzing the diagnosis and treatment effectiveness of the semantic association path in the historical diagnosis and treatment path;

[0037] According to the association frequency and the diagnosis and treatment effectiveness, the logical coherence between the current semantic association path and the retrieved diagnosis and treatment path is calculated.

[0038] Optionally, the step of calculating the suitability of each retrieved diagnosis and treatment pathway and screening at least one diagnosis and treatment pathway whose suitability value exceeds a preset threshold further includes:

[0039] Retrieving semantic analysis content of the diagnosis and treatment information to extract the diseases, symptoms and treatment methods involved in the diagnosis and treatment information;

[0040] Based on the extracted diseases, symptoms and treatment methods, the diagnosis and treatment nodes in the semantic association path are matched with the diagnosis and treatment information, and the correlation between the diagnosis and treatment nodes in the semantic association path and the diagnosis and treatment information content is calculated.

[0041] Optionally, the step of calculating the suitability of each retrieved diagnosis and treatment pathway and screening at least one diagnosis and treatment pathway whose suitability value exceeds a preset threshold further includes:

[0042] Determine evaluation indicators for the diagnosis and treatment pathway, the evaluation indicators including relevance, logical coherence, treatment effect, and safety, and determine a corresponding weight coefficient for each evaluation indicator;

[0043] Based on the diagnosis and treatment information, the evaluation index is assigned a score, so as to calculate a weighted total score value of the diagnosis and treatment pathway according to a weight coefficient of the evaluation index;

[0044] The weighted total score of the diagnosis and treatment pathway is normalized to determine the suitability of the diagnosis and treatment pathway.

[0045] The diagnosis and surgery ICD coding method of medical records based on knowledge graph provided by the present application performs semantic analysis on the diagnosis and treatment information to extract keywords corresponding to the diagnosis and treatment information; determines the target node involved in the diagnosis and treatment information based on the keywords; retrieves the pre-established knowledge graph of ICD coding, matches the target node to the multiple diagnosis and treatment nodes to determine at least one diagnosis and treatment path matched by the target node; and outputs the at least one diagnosis and treatment path and its corresponding ICD code. The method can accurately and quickly match the diagnosis and treatment process of each patient to an accurate ICD code by performing semantic analysis on the diagnosis and treatment information and matching the target node to the multiple diagnosis and treatment nodes based on the pre-established knowledge graph to determine at least one diagnosis and treatment path matched by the target node, thereby greatly reducing the coding management cost and improving the user experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] The drawings herein are incorporated into the specification and constitute a part of the specification, illustrate embodiments consistent with the present application, and are used together with the specification to explain the principles of the present application. In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the description of the embodiments are briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without creative labor.

[0047] Figure 1 This is a flow chart of a method for diagnosis and surgical ICD coding of medical records based on a knowledge graph according to an embodiment of the present application.

[0048] The realization of the purpose, functional features and advantages of this application will be further described in conjunction with the embodiments and with reference to the accompanying drawings. The above-mentioned drawings have shown clear embodiments of this application, which will be described in more detail later. These drawings and textual descriptions are not intended to limit the scope of the concept of this application in any way, but to illustrate the concept of this application to those skilled in the art by referring to specific embodiments. DETAILED DESCRIPTION

[0049] Exemplary embodiments will be described in detail herein, examples of which are shown in the accompanying drawings. When the following description refers to the drawings, the same numbers in different drawings represent the same or similar elements unless otherwise indicated. The implementations described in the following exemplary embodiments do not represent all implementations consistent with the present application. Instead, they are merely examples of devices and methods consistent with some aspects of the present application as detailed in the appended claims.

[0050] It should be noted that, in this article, the terms "include", "comprises" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "includes a ..." does not exclude the existence of other identical elements in the process, method, article or device including the element. In addition, components, features, and elements with the same name in different embodiments of the present application may have the same meaning or different meanings, and their specific meanings need to be determined by their explanation in the specific embodiment or further combined with the context of the specific embodiment.

[0051] It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0052] First embodiment

[0053] This application provides a diagnosis and surgery ICD coding method for medical records based on knowledge graphs. Figure 1 This is a flow chart of a method for diagnosis and surgical ICD coding of medical records based on a knowledge graph according to an embodiment of the present application.

[0054] like Figure 1 As shown, in one embodiment, a diagnosis and surgery ICD coding method for medical records based on a knowledge graph includes:

[0055] S10: Establish a knowledge graph of disease and surgery ICD codes, wherein the knowledge graph of ICD codes includes a diagnosis and treatment pathway corresponding to each ICD code, and each diagnosis and treatment pathway includes multiple diagnosis and treatment nodes.

[0056] The International Classification of Diseases (ICD) is a system that classifies diseases according to certain characteristics of the disease and uses coding methods to represent them. Knowledge Graph, known as knowledge domain visualization or knowledge domain mapping map in the library and information industry, is a series of various graphics that show the development process and structural relationship of knowledge. It uses visualization technology to describe knowledge resources and their carriers, and mines, analyzes, constructs, draws and displays knowledge and their mutual connections. By combining the theories and methods of applied mathematics, graphics, information visualization technology, information science and other disciplines with metrology citation analysis, co-occurrence analysis and other methods, and using visualized graphs to vividly display the core structure, development history, frontier fields and overall knowledge architecture of the ICD coding discipline. In the application process of the knowledge graph of ICD coding, the complex medical ICD coding knowledge field can be displayed through data mining, information processing, knowledge metrology and graph drawing, revealing the dynamic development law of the medical knowledge field, and providing practical and valuable reference for the research of ICD coding discipline. For example, the diagnosis and treatment pathway is a standardized diagnosis and treatment model that aims to standardize medical behavior, improve medical quality and reduce medical costs. The diagnosis and treatment pathway is a standardized medical care plan with strict work order and accurate time requirements for the diagnosis, treatment, rehabilitation and nursing process of a single disease. The diagnosis and treatment pathway not only improves the quality of medical care and management level, but also promotes the upgrading and informatization of hospital information systems. Clinical diagnosis and treatment nodes refer to the important key periods or key factors throughout the development of the disease in medicine, that is, the phased and critical nodes in the development of the disease. Clinical nodes occupy a very important position in clinical practice and are the basis for in-depth understanding and effective intervention of the disease development process. Multiple clinical diagnosis and treatment nodes are connected in series to form a certain diagnosis and treatment pathway.

[0057] S20: In response to obtaining the diagnosis and treatment information to be matched, semantic analysis is performed on the diagnosis and treatment information to extract keywords corresponding to the diagnosis and treatment information.

[0058] Exemplarily, the diagnosis and treatment information mainly includes a patient information cluster composed of various clinical and related contents such as the patient's basic personal information, registration information, medical information, hospitalization doctor's advice information, cost information, imaging data and test results. This information plays a vital role in the medical process. It is not only used for diagnosis and treatment, but also the basis for medical institutions to provide high-quality medical services. Semantic analysis refers to the process of deeply understanding and analyzing the natural language text of diagnosis and treatment information to identify semantic information such as entities, relationships, and emotions in the text. The goal of semantic analysis is to convert natural language text into a form that can be understood and processed by computers, including entity recognition, relationship extraction, sentiment analysis, semantic role labeling, event extraction, semantic similarity calculation, etc. In the process of semantic analysis of diagnosis and treatment information, rule-based methods, statistical learning-based methods, deep learning-based methods, etc. can be used.

[0059] S30: Determine the target node involved in the diagnosis and treatment information based on the keyword.

[0060] By determining the semantic role of each identified entity based on the keywords extracted from the medical information, the target node corresponding to the medical information can be determined.

[0061] S40: Retrieve the knowledge graph of the ICD code, match the target node to the multiple diagnosis and treatment nodes, and determine at least one diagnosis and treatment pathway matched by the target node.

[0062] Establish a knowledge graph that contains the correspondence between treatment plans and coding rules, and put the doctor's fixed knowledge into the knowledge graph. The diagnosis and treatment process of a patient with a disease follows a fixed diagnosis and treatment path. For example, suppose a disease has two treatment methods, one is taking medicine and the other is surgery. Record the doctor's fixed knowledge into the knowledge graph, select the corresponding diagnosis and treatment path for cases taking medicine, and what code comes out, and select the corresponding diagnosis and treatment path for cases undergoing surgery, and then what code comes out. First, there is a knowledge graph, and the correspondence between the diagnosis and treatment path and the ICD code is established. After the patient comes in for diagnosis and treatment, the doctor determines the diagnosis and treatment path node based on the diagnosis record; and automatically identifies the at least one diagnosis and treatment path based on the diagnosis and treatment path node.

[0063] S50: Output the at least one diagnosis and treatment pathway and its corresponding ICD code.

[0064] For example, doctors are very familiar with the diagnosis and treatment pathway. When searching for and determining the ICD code, you can start from any step to select a node in the diagnosis and treatment pathway. You can go up or down. For example, if you have a CT scan today, you can start from the node record of the CT and trace it back, or you can start from the CT and go to the subsequent treatment node.

[0065] This embodiment performs semantic analysis on the diagnosis and treatment information to extract keywords corresponding to the diagnosis and treatment information; determines the target node involved in the diagnosis and treatment information based on the keywords; retrieves the pre-established ICD coding knowledge graph, matches the target node to the multiple diagnosis and treatment nodes, and determines at least one diagnosis and treatment path matched by the target node; and outputs the at least one diagnosis and treatment path and its corresponding ICD code. It can accurately and quickly match the diagnosis and treatment process of each patient to the accurate ICD code, greatly reducing the coding management cost and improving the user experience.

[0066] Optionally, the step of establishing an ICD coding knowledge graph, wherein the ICD coding knowledge graph includes a diagnosis and treatment pathway corresponding to each ICD code, and each diagnosis and treatment pathway includes a plurality of diagnosis and treatment nodes, includes:

[0067] Collect medical data and clean the medical data, including disease information, diagnosis and treatment process, drug information, surgical records, etc.;

[0068] Identifying named entities from the cleaned medical data and assigning an identifier to each named entity; the named entities are used to represent semantic objects in the medical data;

[0069] The association relationships between the named entities are analyzed, and the identified named entities and the association relationships are integrated into a preset knowledge framework to construct the knowledge graph.

[0070] For example, relevant medical data can be collected first, including various medical ICD codes, disease information, diagnosis and treatment process, drug information, etc. These data may come from medical records, clinical guidelines, drug instructions, research papers, etc. The collected data is cleaned, and incomplete, inaccurate or duplicate data is processed to ensure the quality and accuracy of the data. Key entities such as diseases, symptoms, drugs, examination items, etc. can be identified from the collected data and assigned unique identifiers. By further identifying the relationship between entities, such as the treatment method of a certain disease, the side effects of a certain drug, etc., and establishing the connection between these entities. Finally, the extracted entities and relationships can be integrated into a unified knowledge framework to form a structured knowledge graph.

[0071] Optionally, the step of establishing an ICD coding knowledge graph, wherein the ICD coding knowledge graph includes a diagnosis and treatment pathway corresponding to each ICD code, and each diagnosis and treatment pathway includes a plurality of diagnosis and treatment nodes, further includes:

[0072] Performing knowledge verification on the knowledge graph, wherein the knowledge verification includes expert verification and data cross-validation;

[0073] Regularly collect user and expert opinions and update the knowledge graph.

[0074] Exemplarily, the constructed knowledge graph is verified to ensure its accuracy and completeness. This may include working with experts, cross-validating data, etc. Since medical knowledge is constantly updated, the knowledge graph needs to be updated regularly to reflect the latest medical advances and research results. Optionally, a suitable data storage method, such as a relational database, a graph database, etc., can be selected to store the constructed knowledge graph. An interface or platform can be developed to allow users to query and access information in the knowledge graph, thereby providing a structured, easy-to-query and easy-to-analyze medical information resource.

[0075] Optionally, in response to obtaining the medical information to be matched, the step of performing semantic analysis on the medical information to extract keywords corresponding to the medical information includes:

[0076] Extract target keywords according to the diagnosis and treatment information to be matched, and mark the part of speech of each target keyword;

[0077] Based on the target keywords marked with the part of speech, the named entities in the diagnosis and treatment information are identified and classified and labeled.

[0078] For example, the medical information is preprocessed, including removing irrelevant characters, converting to a standard format (such as lowercase), and segmenting (segmenting the text into individual words or phrases). Suppose there is a medical information: "Use aspirin to treat hypertension". After preprocessing: remove irrelevant characters (such as punctuation marks), convert to a standard format, and segment into "use", "aspirin", "treatment", "hypertension".

[0079] Exemplarily, extracting keywords from the preprocessed text can be achieved through a variety of methods, including statistical methods (such as TF-IDF), semantic methods (such as part-of-speech tagging, named entity recognition), or deep learning methods (such as word embedding, neural network). For example, the TF-IDF method is used to extract keywords. The extraction results may include "aspirin", "treatment", and "hypertension".

[0080] For example, the part of speech of each word, such as noun, verb, adjective, etc., can be further identified and marked to facilitate further semantic analysis. For example, "aspirin" is marked as "noun", "treatment" is marked as "verb", and "hypertension" is marked as "noun".

[0081] For example, we can identify and classify named entities in text, such as names of people, places, organizations, and medicines, which are important and widely used in the medical field. For example, we can identify "aspirin" as a "drug name" and "hypertension" as a "disease name".

[0082] Optionally, the step of determining the target node involved in the diagnosis and treatment information based on the keyword includes:

[0083] Determine the weight of each target keyword according to a preset strategy based on the importance of the target keyword in the diagnosis and treatment information;

[0084] Based on the weight of the target keyword and the classification identifier of the named entity, determining the semantic role of each named entity in the diagnosis and treatment information to determine the association relationship between the named entities;

[0085] Connect entity nodes with association relationships in sequence to build a semantic association path;

[0086] The diagnosis and treatment nodes in the semantic association path are extracted, and the target nodes involved in the diagnosis and treatment information are determined.

[0087] Exemplarily, phrases in the text, such as verb-object phrases, modifier-predicate phrases, etc., can be identified and extracted through commonly used algorithm models. These phrases usually contain keywords. For example, the verb-object phrase "use aspirin" and the modifier-predicate phrase "treat hypertension" in "use aspirin to treat hypertension" are extracted. The weight of the keyword can be calculated according to its importance in the text, such as using TF-IDF or a word embedding method based on deep learning. For example, TF-IDF is used to calculate the weights of "aspirin", "treatment", and "hypertension" to determine their importance in the sentence.

[0088] Exemplarily, the extracted keywords are sorted or filtered to determine the most relevant keywords, which will be used for subsequent diagnosis and treatment path matching. For example, according to the weight sorting, it is determined that "aspirin" and "hypertension" are the most important keywords for subsequent diagnosis and treatment path matching. The semantic roles of entities in sentences, such as agent, patient, object, etc., can be determined by semantic analysis algorithms, which helps to understand the relationship between entities. For example, it is determined that "aspirin" plays the role of "treatment method" in the sentence, and "hypertension" plays the role of "disease".

[0089] By understanding the relationship between entities, we can more accurately understand the meaning of diagnosis and treatment information and provide a basis for subsequent medical ICD coding matching.

[0090] Exemplarily, the association relationship between entities is analyzed, which may involve direct connections between entities (such as drugs used to treat a certain disease) or indirect connections (such as the connection between symptoms and diseases). For example, by analyzing the relationship between "aspirin" and "hypertension", it can be determined that "aspirin" is a drug used to treat "hypertension". A semantic role-related weight can be assigned to each entity based on the previously extracted keyword weights and the classification of named entities (such as disease names, symptom names, drug names, etc.). For example, in the medical information "Using aspirin to treat hypertension", "aspirin" is a drug name with a higher weight, while "hypertension" is a disease name with a lower weight.

[0091] Exemplarily, a semantic role is assigned to each named entity, such as "treatment method", "disease", etc. For example, "aspirin" can be assigned the role of "treatment method", and "hypertension" can be assigned the role of "disease". Connect the entity nodes with direct or indirect associations in sequence to form a semantically associated diagnosis and treatment path. For example, a diagnosis and treatment path for myocardial infarction patients containing the nodes of "cardiac angiography" and "quick-acting heart-saving pills" means that in the treatment process of myocardial infarction patients, the "quick-acting heart-saving pills" surgical method is first used for preliminary treatment, and then "cardiac angiography" is used for follow-up treatment. Finally, based on the entity relationship analysis, the target node involved in each diagnosis and treatment information is determined. For example, in the diagnosis and treatment path of myocardial infarction patients, the target node may be the surgical treatment node "cardiac angiography" and the drug treatment node "aspirin".

[0092] Optionally, the step of retrieving the knowledge graph of the ICD code and matching the target node to the multiple diagnosis and treatment nodes to determine at least one diagnosis and treatment pathway matched by the target node includes:

[0093] Performing a query operation from the knowledge graph of the ICD code to retrieve the diagnosis and treatment nodes in the knowledge graph of the ICD code, so that the target nodes are matched with the diagnosis and treatment nodes one by one;

[0094] When the target node is successfully matched, a search is performed along the relationship chain between named entities based on the knowledge graph of the ICD code to retrieve the diagnosis and treatment path connected to the target node;

[0095] The suitability of each retrieved diagnosis and treatment pathway is calculated, and at least one diagnosis and treatment pathway whose suitability value exceeds a preset threshold is screened; the suitability is used to characterize the degree of matching between the diagnosis and treatment pathway and the target node.

[0096] Exemplarily, relevant information can be retrieved from a stored medical ICD coding knowledge graph. This typically involves performing queries in a graph database. For example, querying the graph to find treatment pathways related to "hypertension" and "aspirin". Next, the extracted target nodes (such as "hypertension" and "aspirin") are matched with nodes in the knowledge graph. The match may be based on the name, type, or other attributes of the entity. For example, match the "hypertension" node to the "disease" category in the graph, and match the "aspirin" node to the "drug" category.

[0097] Once matching nodes are found, the system searches the knowledge graph for treatment paths connecting these nodes. This may include searching along the relationship chain between entities, such as from the "disease" node to the "treatment method" node. For example, search for the treatment path from the "hypertension" node to the "drug treatment" node. After finding the treatment path, the suitability of the found treatment path can be evaluated, which may include checking the logical coherence of the path, the relevance of the nodes on the path, etc. For example, evaluate whether the treatment path from "hypertension" to "aspirin" is reasonable, that is, whether "aspirin" is indeed an effective way to treat "hypertension".

[0098] Based on the evaluation results, the most appropriate path is selected. If there are multiple paths, the one that best matches the diagnosis and treatment information is selected. This can be done by calculating the suitability value and selecting the diagnosis and treatment path corresponding to the highest suitability value that exceeds the preset threshold, so that the system can retrieve and select the diagnosis and treatment path that is most relevant to the diagnosis and treatment information to be matched from the medical ICD coding knowledge graph.

[0099] Optionally, the step of performing a query operation from the knowledge graph of the ICD code to retrieve the diagnosis and treatment nodes in the knowledge graph of the ICD code and matching the target nodes with the diagnosis and treatment nodes one by one includes:

[0100] The entity names of the target node and the diagnosis and treatment node are compared to perform a name-based matching calculation.

[0101] For example, in a name-based matching method, if the target node is "hypertension" and there is a corresponding node in the knowledge graph, matching is performed by comparing the names. The final matching result is to determine that the "hypertension" node in the knowledge graph matches the extracted target node.

[0102] Optionally, the step of performing a query operation from the knowledge graph of the ICD code to retrieve the diagnosis and treatment nodes in the knowledge graph of the ICD code and matching the target nodes with the diagnosis and treatment nodes one by one includes:

[0103] The category of the target node is retrieved, and the category of the target node is compared with the category of the diagnosis and treatment node to perform a category-based matching calculation.

[0104] For example, if the target node is "aspirin" and it is known that it is a "drug", then all nodes of the "drug" type are searched in the knowledge graph. Eventually, all nodes of the "drug" type are found, and then which node is "aspirin" is determined.

[0105] Optionally, the step of performing a query operation from the knowledge graph of the ICD code to retrieve the diagnosis and treatment nodes in the knowledge graph of the ICD code and matching the target nodes with the diagnosis and treatment nodes one by one includes:

[0106] The attributes of the target node are retrieved, and the attributes of the target node are compared with the attributes of the diagnosis and treatment node to perform attribute-based matching calculation.

[0107] For example, if the target node has specific attributes, such as the "chemical composition" or "manufacturer" of "aspirin", nodes with these attributes are searched in the knowledge graph. Finally, nodes with matching attributes are found, such as drug nodes containing the chemical composition of "aspirin".

[0108] Optionally, the step of performing a query operation from the knowledge graph of the ICD code to retrieve the diagnosis and treatment nodes in the knowledge graph of the ICD code and matching the target nodes with the diagnosis and treatment nodes one by one includes:

[0109] The semantic relationship between the target nodes is analyzed, and the semantic relationship between the target nodes is compared with the semantics of the diagnosis and treatment nodes to perform a semantic-based matching calculation.

[0110] For example, if the relationship between the target nodes is clear, such as "aspirin" is used to "treat high blood pressure", then search for nodes with this semantic relationship in the knowledge graph. Finally, find the node with "aspirin" as a "treatment method" and determine its association with the "high blood pressure" node. If there are specific rules or logical relationships between the target node and the nodes in the knowledge graph, such as "patients over 65 years old" "need regular physical examinations", matching can also be based on these rules.

[0111] Optionally, the step of calculating the suitability of each retrieved diagnosis and treatment pathway and screening at least one diagnosis and treatment pathway whose suitability value exceeds a preset threshold comprises:

[0112] querying a medical database to determine association frequencies between diagnosis and treatment nodes in the semantic association path;

[0113] Retrieving case data from historical diagnosis and treatment data, and analyzing the diagnosis and treatment effectiveness of the semantic association path in the historical diagnosis and treatment path;

[0114] According to the association frequency and the diagnosis and treatment effectiveness, the logical coherence between the current semantic association path and the retrieved diagnosis and treatment path is calculated.

[0115] Exemplarily, evaluating the suitability of the found path may include checking the logical coherence and node relevance of the path. Check whether the relationship between each node in the path conforms to medical common sense and logic, and further determine whether each node in the path is related to the content in the diagnosis and treatment information. Exemplarily, query a medical knowledge base or database, such as medical literature, clinical guidelines, drug instructions, etc., to obtain authoritative medical information about each node in the path and the relationship between them. Among the multiple authoritative medical information queried, the frequency of a certain treatment method (such as "aspirin") in the treatment of a specific disease (such as "hypertension") can be calculated, and the association frequency between diagnosis and treatment nodes can be determined. To further analyze the therapeutic effect of a certain treatment method on a specific disease, indicators such as treatment response rate and survival rate can be used. To further analyze the correlation between the disease and the treatment method, the following expression can be used, using statistical methods such as correlation coefficients:

[0116] F =

[0117] Among them, F is the first correlation coefficient, A is the disease index, B is the average disease index, C is the treatment method index, and D is the average treatment method index.

[0118] Optionally, the step of calculating the suitability of each retrieved diagnosis and treatment pathway and screening at least one diagnosis and treatment pathway whose suitability value exceeds a preset threshold further includes:

[0119] Retrieving semantic analysis content of the diagnosis and treatment information to extract the diseases, symptoms and treatment methods involved in the diagnosis and treatment information;

[0120] Based on the extracted diseases, symptoms and treatment methods, the diagnosis and treatment nodes in the semantic association path are matched with the diagnosis and treatment information, and the correlation between the diagnosis and treatment nodes in the semantic association path and the diagnosis and treatment information content is calculated.

[0121] Exemplarily, the content of the diagnosis and treatment information is analyzed in depth to determine the specific diseases, symptoms, treatment methods, etc. involved. For example, if the diagnosis and treatment information is "using aspirin to treat high blood pressure", it is necessary to clarify that "high blood pressure" is a disease and "aspirin" is a treatment method. Then match the extracted target node with the content in the diagnosis and treatment information. For example, match "high blood pressure" with the "disease" node, and match "aspirin" with the "drug" node. You can also compare whether the node attributes in the path are consistent with the content in the diagnosis and treatment information, and analyze whether the relationship between the nodes is consistent with the description in the diagnosis and treatment information. When analyzing the correlation between node attributes and diagnosis and treatment information, you can use the following expression, using statistical methods such as correlation coefficients:

[0122] E =

[0123] Among them, E is the second correlation coefficient, H is the node attribute, I is the average node attribute, J is the field content, and K is the average field content.

[0124] Optionally, the step of calculating the suitability of each retrieved diagnosis and treatment pathway and screening at least one diagnosis and treatment pathway whose suitability value exceeds a preset threshold further includes:

[0125] Determine evaluation indicators for the diagnosis and treatment pathway, the evaluation indicators including relevance, logical coherence, treatment effect, and safety, and determine a corresponding weight coefficient for each evaluation indicator;

[0126] Based on the diagnosis and treatment information, the evaluation index is assigned a score, so as to calculate a weighted total score value of the diagnosis and treatment pathway according to a weight coefficient of the evaluation index;

[0127] The weighted total score of the diagnosis and treatment pathway is normalized to determine the suitability of the diagnosis and treatment pathway.

[0128] For example, weights may be assigned according to the importance of each indicator. For example, the weight of logical coherence is 0.3, the weight of node relevance is 0.2, the weight of treatment effect is 0.2, the weight of safety is 0.15, and the weight of cost-effectiveness is 0.15. The score of each indicator may be multiplied by its weight and then added together to obtain the total score according to the following expression:

[0129] Total score = (average score of logical coherence × weight of logical coherence) + (average score of node relevance × weight of node relevance) + (average score of treatment effect × weight of treatment effect) + (average score of safety × weight of safety) + (average score of cost-effectiveness × weight of cost-effectiveness).

[0130] For example, if the logical coherence value and the node relevance value have been calculated, the suitability value of the diagnosis and treatment pathway can be calculated based on these values. In the process of determining the weights of logical coherence and node relevance in the overall suitability assessment, for example, assume that the logical coherence weight is 0.6 and the node relevance weight is 0.4. If the logical coherence score is 0.8 and the node relevance score is 0.9. Multiply the score by the corresponding weight, for example:

[0131] Logical coherence score = 0.8 × 0.6 = 0.48

[0132] Node relevance score = 0.9 × 0.4 = 0.36

[0133] Total score = 0.48 + 0.36 = 0.84

[0134] For example, the total score can be standardized to be between 0 and 1 for easy comparison. For example, the total score is divided by the maximum possible total score (such as 1) to obtain a standardized suitability score. Finally, the suitability of the diagnosis and treatment pathway can be judged based on the suitability score. For example, a score close to 1 indicates that the pathway is very suitable, while a score close to 0 indicates that the pathway may not be suitable.

[0135] In this way, a comprehensive suitability value can be calculated based on logical coherence and node relevance to help evaluate and select the best diagnosis and treatment pathway.

[0136] Second embodiment

[0137] The present application also provides a data processing system, comprising: a processor and a memory storing a computer program, and when the processor runs the computer program, the above method is implemented.

[0138] Exemplarily, during the treatment process, the doctor gives the patient appropriate treatment according to his own diagnosis and treatment process. During the diagnosis and treatment process, each diagnosis and treatment node of the doctor's diagnosis and treatment process is synchronously recorded and collected. For patients with each disease entering the doctor's diagnosis and treatment process, the same condition determines the diagnosis and treatment process according to a fixed diagnosis and treatment path. A disease may have multiple treatment methods, one diagnosis and treatment process is drug treatment, and the other is surgical treatment. In the data processing system of this embodiment, the doctor's fixed diagnosis and treatment knowledge is collected and organized to establish a knowledge graph containing the correspondence between treatment plans and coding rules. This knowledge graph can be used as the infrastructure for the data processing system to search for ICD codes.

[0139] Exemplarily, after recording the doctor's fixed knowledge into the knowledge graph, the first diagnosis and treatment path corresponding to the drug treatment can be selected according to the case of drug treatment, which corresponds to the first ICD code. According to the case of surgical treatment, the second diagnosis and treatment path corresponding to the surgical treatment can be selected, which corresponds to the second ICD code. The data processing system of the present application establishes the correspondence between the diagnosis and treatment path and the ICD code in the knowledge graph. After the patient enters the diagnosis and treatment process, the doctor diagnoses him and gives the various diagnosis and treatment nodes of the diagnosis and treatment path.

[0140] The data processing system of this embodiment can identify the doctor's case content through the large model, automatically identify and match the corresponding ICD code according to the case content, and finally record it in the case file. It can also identify and select the path on the computer according to the diagnosis and treatment path node, so as to identify the ICD code corresponding to the diagnosis and treatment path, and finally record it in the case file.

[0141] For example, the doctor may not be sure of the ICD code corresponding to each diagnosis and treatment pathway, but the doctor is very clear about the diagnosis and treatment pathway. When searching for ICD codes, users can select a node in the diagnosis and treatment pathway from any step. You can go up or down. For example, if you have a CT scan today, you can start from the CT scan node and trace back to the source, or you can start from the CT scan node and compare and search to the subsequent treatment node.

[0142] For example, for patients with myocardial infarction, heart stents may be placed after diagnosis. At this time, the doctor as a user needs to use the data processing system to query the knowledge graph and determine the corresponding ICD code. Users can search for stents in the system. For example, the three paths in the system all include stents. Different numbers of stents are placed for different patients, and different other diagnostic and treatment measures are required before and after the diagnosis and treatment node of stent placement. Then the doctor will retrieve an ICD code based on the diagnosis and treatment process path of his own patients before and after the stent placement, such as a cardiac angiography before placing two stents. The final result is based on the path relationship between the two diagnosis and treatment nodes, such as X+Y+Z. At this time, the ICD code X+Y+Z can be registered on the patient's electronic medical record.

[0143] For example, if during the query process, cardiac angiography and the placement of two stents are entered, which may overlap with other disease diagnosis and treatment pathways, the system will provide multiple diagnosis and treatment pathways and corresponding ICD codes. The doctor can continue to match upstream or downstream nodes to find an ICD code that can fully match the code for registration.

[0144] Third embodiment

[0145] The present application also provides a readable storage medium storing a computer program, which implements the steps of the above method when executed by a processor.

[0146] The diagnosis and surgery ICD coding method of medical records based on knowledge graph provided by the present application, through semantic analysis of diagnosis and treatment information, to extract keywords corresponding to the diagnosis and treatment information; based on the keywords, to determine the target nodes involved in the diagnosis and treatment information; to retrieve the pre-established knowledge graph of ICD coding, to match the target node to the multiple diagnosis and treatment nodes, to determine at least one diagnosis and treatment path matched by the target node; to output the at least one diagnosis and treatment path and its corresponding ICD code. It can accurately and quickly match the accurate ICD code for each patient's diagnosis and treatment process, greatly reducing the coding management cost and improving the user experience.

[0147] It should be noted that in the present application, step codes such as S10, S20, etc. are used for the purpose of expressing the corresponding content more clearly and concisely, and do not constitute a substantial limitation on the sequence. When implementing the step, those skilled in the art may execute S20 first and then S10, etc., but these should all be within the scope of protection of the present application.

[0148] In the embodiments of the system and storage medium provided in the present application, all technical features of any of the above-mentioned method embodiments may be included, and the expanded and explained contents of the specification are basically the same as those of the above-mentioned method embodiments, and will not be repeated here.

[0149] The embodiment of the present application further provides a computer program product, which includes a computer program code. When the computer program code runs on a computer, the computer executes the methods in the above various possible implementation modes.

[0150] An embodiment of the present application also provides a chip, including a memory and a processor, the memory is used to store a computer program, and the processor is used to call and run the computer program from the memory, so that a device equipped with the chip executes the methods in various possible implementation modes as described above.

[0151] It is understood that the above scenarios are only examples and do not constitute a limitation on the application scenarios of the technical solutions provided in the embodiments of the present application. The technical solutions of the present application can also be applied to other scenarios. For example, it is known to those skilled in the art that with the evolution of the system architecture and the emergence of new business scenarios, the technical solutions provided in the embodiments of the present application are also applicable to similar technical problems.

[0152] The serial numbers of the above-mentioned embodiments of the present application are for description only and do not represent the advantages or disadvantages of the embodiments.

[0153] The steps in the method of the embodiment of the present application can be adjusted in order, combined and deleted according to actual needs.

[0154] The units in the device of the embodiment of the present application can be merged, divided and deleted according to actual needs.

[0155] In the present application, the same or similar terminology concepts, technical solutions and / or application scenario descriptions are generally described in detail only the first time they appear. When they appear again later, they are generally not repeated for the sake of brevity. When understanding the technical solutions and other contents of the present application, for the same or similar terminology concepts, technical solutions and / or application scenario descriptions that are not described in detail later, reference can be made to the previous related detailed descriptions.

[0156] In the present application, the description of each embodiment has its own emphasis. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0157] The various technical features of the technical solution of the present application can be arbitrarily combined. In order to make the description concise, not all possible combinations of the various technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of the present application.

[0158] The above are only preferred embodiments of the present application, and are not intended to limit the patent scope of the present application. Any equivalent structure or equivalent process transformation made using the contents of the present application specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present application.

Claims

1. A diagnosis and surgery ICD coding method for medical records based on knowledge graph, characterized in that: include: S10: Establish a knowledge graph of disease and surgery ICD codes, wherein the knowledge graph of ICD codes includes a diagnosis and treatment pathway corresponding to each ICD code, and each diagnosis and treatment pathway includes multiple diagnosis and treatment nodes; S20: in response to obtaining the diagnosis and treatment information to be matched, performing semantic analysis on the diagnosis and treatment information to extract keywords corresponding to the diagnosis and treatment information; S30: Determine the target node involved in the diagnosis and treatment information based on the keyword; S40: Retrieving the knowledge graph of the ICD code, matching the target node to the multiple diagnosis and treatment nodes, so as to determine at least one diagnosis and treatment pathway matched by the target node; S50: Outputting the at least one diagnosis and treatment pathway and its corresponding ICD code; In response to obtaining the diagnosis and treatment information to be matched, the step of performing semantic analysis on the diagnosis and treatment information to extract keywords corresponding to the diagnosis and treatment information includes: Extract target keywords according to the diagnosis and treatment information to be matched, and mark the part of speech of each target keyword; Based on the target keywords marked with parts of speech, identify the named entities in the diagnosis and treatment information, and classify and identify the named entities; The step of determining the target node involved in the diagnosis and treatment information based on the keyword includes: Determine the weight of each target keyword according to the importance of the target keyword in the diagnosis and treatment information according to a preset strategy; Based on the weight of the target keyword and the classification identifier of the named entity, determining the semantic role of each named entity in the diagnosis and treatment information to determine the association relationship between the named entities; Connect entity nodes with association relationships in sequence to build a semantic association path; The diagnosis and treatment nodes in the semantic association path are extracted, and the target nodes involved in the diagnosis and treatment information are determined.

2. According to claim 1, a diagnosis and surgery ICD coding method for medical records based on knowledge graph is characterized in that: The step of establishing an ICD coding knowledge graph, wherein the ICD coding knowledge graph includes a diagnosis and treatment path corresponding to each ICD code, and each diagnosis and treatment path includes a plurality of diagnosis and treatment nodes, includes: Collect medical data and clean the medical data, including disease information, diagnosis and treatment process, drug information, surgical records, etc.; Identifying named entities from the cleaned medical data and assigning an identifier to each named entity; the named entities are used to represent semantic objects in the medical data; The association relationships between the named entities are analyzed, and the identified named entities and the association relationships are integrated into a preset knowledge framework to form the knowledge graph.

3. According to claim 2, a diagnosis and surgery ICD coding method for medical records based on knowledge graph is characterized in that: The step of establishing an ICD coding knowledge graph, wherein the ICD coding knowledge graph includes a diagnosis and treatment path corresponding to each ICD code, and each diagnosis and treatment path includes a plurality of diagnosis and treatment nodes, further includes: Performing knowledge verification on the knowledge graph, wherein the knowledge verification includes expert verification and data cross-validation; Regularly collect user and expert opinions and update the knowledge graph.

4. A method for diagnosis and surgery ICD coding of medical records based on knowledge graph according to any one of claims 1 to 3, characterized in that: The step of retrieving the knowledge graph of the ICD code and matching the target node to the multiple diagnosis and treatment nodes to determine at least one diagnosis and treatment pathway matched by the target node includes: Performing a query operation from the knowledge graph of the ICD code to retrieve the diagnosis and treatment nodes in the knowledge graph of the ICD code, so that the target nodes are matched with the diagnosis and treatment nodes one by one; The step of performing a query operation from the knowledge graph of the ICD code to retrieve the diagnosis and treatment nodes in the knowledge graph of the ICD code and matching the target nodes with the diagnosis and treatment nodes one by one includes at least one of the following: Comparing the entity names of the target node and the diagnosis and treatment node to perform a name-based matching calculation; Retrieving the category of the target node, and comparing the category of the target node with the category of the diagnosis and treatment node to perform a category-based matching calculation; Retrieving the attributes of the target node, and comparing the attributes of the target node with the attributes of the diagnosis and treatment node to perform attribute-based matching calculation; The semantic relationship between the target nodes is analyzed, and the semantic relationship between the target nodes is compared with the semantics of the diagnosis and treatment nodes to perform a semantic-based matching calculation.

5. A method for diagnosis and surgery ICD coding of medical records based on knowledge graph according to any one of claims 1 to 3, characterized in that: The step of retrieving the knowledge graph of the ICD code and matching the target node to the multiple diagnosis and treatment nodes to determine at least one diagnosis and treatment pathway matched by the target node includes: Calculating the suitability of each retrieved diagnosis and treatment pathway, and screening at least one diagnosis and treatment pathway whose suitability value exceeds a preset threshold; the suitability is used to characterize the degree of matching between the diagnosis and treatment pathway and the target node; The step of calculating the suitability of each retrieved diagnosis and treatment pathway and selecting at least one diagnosis and treatment pathway whose suitability value exceeds a preset threshold comprises: querying a medical database to determine association frequencies between diagnosis and treatment nodes in the semantic association path; Retrieving case data from historical diagnosis and treatment data, and analyzing the diagnosis and treatment effectiveness of the semantic association path in the historical diagnosis and treatment path; According to the association frequency and the diagnosis and treatment effectiveness, the logical coherence between the current semantic association path and the retrieved diagnosis and treatment path is calculated.

6. According to claim 5, a method for diagnosis and surgery ICD coding of medical records based on knowledge graph, characterized in that: The step of calculating the suitability of each retrieved diagnosis and treatment pathway and screening at least one diagnosis and treatment pathway whose suitability value exceeds a preset threshold value further includes: Retrieving semantic analysis content of the diagnosis and treatment information to extract the diseases, symptoms and treatment methods involved in the diagnosis and treatment information; Based on the extracted diseases, symptoms and treatment methods, the diagnosis and treatment nodes in the semantic association path are matched with the diagnosis and treatment information, and the correlation between the diagnosis and treatment nodes in the semantic association path and the diagnosis and treatment information content is calculated.

7. According to claim 6, a method for diagnosis and surgery ICD coding of medical records based on knowledge graph, characterized in that: The step of calculating the suitability of each retrieved diagnosis and treatment pathway and screening at least one diagnosis and treatment pathway whose suitability value exceeds a preset threshold value further includes: Determine evaluation indicators for the diagnosis and treatment pathway, the evaluation indicators including relevance, logical coherence, treatment effect, and safety, and determine a corresponding weight coefficient for each evaluation indicator; Assigning scores to the evaluation indicators based on the diagnosis and treatment information, so as to calculate a weighted total score value of the diagnosis and treatment pathway according to a weight coefficient of the evaluation indicator; The weighted total score of the diagnosis and treatment pathway is normalized to determine the suitability of the diagnosis and treatment pathway.

Images