Medical record quality management system and method

Abstract

The present invention discloses a medical record quality management system and method, which relates to the field of medical information processing technology, including: collecting medical record data, preprocessing the collected medical record data; using NLP to parse the preprocessed medical record data, and extracting the core fields of the medical record through NER; constructing a dynamic medical knowledge graph, matching the dynamic medical knowledge graph with the core fields of the medical record; after the matching is completed, using a graph reasoning algorithm to perform a logical consistency check on the medical record, marking logical contradictions, and generating a verification result; according to the verification result, generating a medical record quality audit report. The present invention realizes the intelligentization and precision of medical record quality management by constructing a dynamic medical knowledge graph and matching it with the core fields of the medical record, and using a graph reasoning algorithm to perform a logical consistency check on the medical record.

Description

Technical Field

[0001] The present invention relates to the technical field of medical information processing, and in particular to a medical record quality management system and method. Background Art

[0002] With the rapid development of medical informatization, electronic medical records (EMR) have gradually become the main carrier for medical institutions to record and manage patient diagnosis and treatment information. Electronic medical records not only play an important role in the medical service process, but also, as the core of medical data, provide key data support for clinical decision support systems, medical research, medical insurance settlement, etc. However, the quality of medical records has long been one of the bottlenecks restricting the in-depth application of electronic medical records. At present, there are problems such as too high proportion of unstructured text, inconsistent data format, and frequent logical contradictions in medical records, which affect the accuracy, completeness, and standardization of medical record information. In addition, with the rapid development of natural language processing (NLP) technology, medical record quality verification methods based on NLP and knowledge graph technology have gradually emerged, and attempts have been made to improve the quality of medical records through automated means. For example, in the existing technology, named entity recognition (NER) technology is used to extract key fields from unstructured medical records, and medical record fields are matched and verified based on the medical knowledge base. However, these methods still rely on static knowledge bases, cannot dynamically update medical knowledge, and lack comprehensive verification of the logical consistency of medical record fields, making it difficult to meet actual application needs.

[0003] The deficiencies of existing technologies in the quality management of medical records are mainly reflected in the following two points: First, existing methods have limited processing capabilities for unstructured medical record texts, and the data preprocessing process lacks standardization, resulting in ambiguity or incorrect mapping of extracted core fields; second, existing medical knowledge bases are mostly statically constructed and cannot dynamically update medical knowledge. In addition, they only rely on simple field matching rules in logical verification and fail to fully utilize graph reasoning algorithms for deep logical consistency verification. These deficiencies make it difficult to achieve a comprehensive and accurate assessment of the quality audit of medical records, which has an adverse impact on subsequent medical research and clinical decision support. Therefore, there is an urgent need for a medical record quality management method that can combine dynamic medical knowledge graphs, natural language processing technology, and logical reasoning algorithms to improve the accuracy and logical consistency of medical records. Summary of the invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a medical record quality management method to solve the problems of irregular processing of unstructured medical record data and insufficient depth of logical verification.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0007] In the first aspect, the present invention provides a medical record quality management method, which includes: collecting medical record data and preprocessing the collected medical record data; using NLP to parse the preprocessed medical record data and extracting the core fields of the medical record through NER; constructing a dynamic medical knowledge graph and matching the dynamic medical knowledge graph with the core fields of the medical record; after the matching is completed, using a graph reasoning algorithm to perform a logical consistency check on the medical record, mark logical contradictions, and generate a verification result; based on the verification result, generate a medical record quality review report.

[0008] As a preferred solution of the medical record quality management method of the present invention, the medical record data includes unstructured data, semi-structured data and structured data.

[0009] As a preferred solution of the medical record quality management method of the present invention, the preprocessing of the collected medical record data comprises the following specific steps:

[0010] Convert the collected medical record data into UTF-8 encoding format;

[0011] Use NLP tools to split and annotate unstructured data by paragraphs and categories;

[0012] Standardize medical record data;

[0013] Perform unit unification and value conversion on structured data.

[0014] As a preferred solution of the medical record quality management method of the present invention, the NLP is used to parse the pre-processed medical record data, and the core fields of the medical record are extracted by NER. The specific steps are as follows:

[0015] Use BERT to perform word segmentation and syntactic analysis on the preprocessed medical record data to extract the grammatical structure and candidate segments of the structured data;

[0016] According to the grammatical structure and candidate segments of structured data, BERT-NER is used to extract the core fields of medical records;

[0017] The extracted core fields of medical records are standardized and mapped based on the medical terminology library, and the terms in the structured data are replaced with standardized codes.

[0018] As a preferred solution of the medical record quality management method of the present invention, the construction of a dynamic medical knowledge graph and matching the dynamic medical knowledge graph with the core fields of the medical record are specifically performed in the following steps:

[0019] Through the coding and semantic rules in the standard medical terminology library, the logical associations between the core fields of the medical record are standardized and standardized relationships are defined;

[0020] Create nodes of the dynamic medical knowledge graph based on standardized coded terms, and construct edges of the dynamic medical knowledge graph based on standard relationships;

[0021] Based on the nodes of the dynamic medical knowledge graph and the edges of the dynamic medical knowledge graph, a dynamic medical knowledge graph is constructed using a graph database;

[0022] Word2Vec embedding is used to calculate the semantic similarity between the core fields of medical records and entities in the dynamic medical knowledge graph. The expression is:

[0023] ;

[0024] in, A vector representing the core fields of a medical record In the The values ​​in the dimensions, A vector representing the core fields of the medical record, Represents the corresponding standard term vector in the dynamic medical knowledge graph The value in the kth dimension, represents the corresponding standard term vector in the dynamic medical knowledge graph, Represents the core field vector of the medical record Standard term vectors with dynamic medical knowledge graph The similarity between is the size of the vector dimension, is the index variable of the vector of core fields of medical records, is the index variable of the standard term vector;

[0025] Based on the statistical analysis of the matching results between the core fields of historical medical records and the standard terms of the medical knowledge base, the verification threshold S1 is defined;

[0026] when When ≥S1, the core fields of the medical record are considered to be consistent with medical logic and the match is successful;

[0027] when When <S1, the field is considered not to conform to medical logic and the match fails.

[0028] As a preferred solution of the medical record quality management method of the present invention, after the matching is completed, a graph reasoning algorithm is used to perform a logical consistency check on the medical record, mark the logical contradictions, and generate a check result. The specific steps are as follows:

[0029] According to the results of the core field matching of the medical record, the successfully matched core fields and their corresponding node numbers of the dynamic medical knowledge graph and the medical logical relationships between the matching nodes are extracted, and the medical record field matching results and their logical relationships are organized into triples. ,in, represents the head node in the triple, Represents the head node With the tail node The logical relationship between Represents the tail node in a triple;

[0030] According to the standardized relations in the dynamic medical knowledge graph, the relation existence rules and relation mutual exclusivity rules are defined;

[0031] Based on the relationship existence rule, relationship mutual exclusivity rule and graph inference algorithm, the logical consistency score between medical record fields is calculated. The expression is:

[0032] ;

[0033] in, Score the logical consistency between medical record fields, Indicates the number of triples that successfully match the core fields of the medical record. Indicates the head node corresponding to the core field of the medical record. Represents the standardized logical relationships defined in the dynamic medical knowledge graph. Indicates the tail node corresponding to the core field of the medical record. represents the existence rate of triples in the dynamic medical knowledge graph, Represents the head node and tail node The mutual exclusion rate, Represents the head node To the end node Path consistency, is a very small true number;

[0034] Based on the statistical analysis of historical data, define the logical consistency check threshold Q1;

[0035] when When ≥Q1, the current triple is marked as logically consistent;

[0036] when When <Q1, the current triple is marked as a logical contradiction.

[0037] As a preferred solution of the medical record quality management method of the present invention, the medical record quality audit report includes the successfully matched core pathology fields and their logical relationships, the marked logical contradictions and contradiction reasons, and the logical consistency scores between the medical record fields. .

[0038] In the second aspect, the present invention provides a medical record quality management system, including: a data acquisition module, a field extraction module, a field matching module, a logical consistency verification module and a report generation module, the data acquisition module is used to collect medical record data and pre-process the collected medical record data; the field extraction module is used to use NLP to parse the pre-processed medical record data, and extract the core fields of the medical record through NER; the field matching module is used to construct a dynamic medical knowledge graph and match the dynamic medical knowledge graph with the core fields of the medical record; the logical consistency verification module is used to use a graph reasoning algorithm to perform logical consistency verification on the medical record after matching, mark logical contradictions, and generate verification results; the report generation module is used to generate a medical record quality review report based on the verification results.

[0039] In a third aspect, the present invention provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein when the computer program is executed by the processor, any step of the medical record quality management method as described in the first aspect of the present invention is implemented.

[0040] In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements any step of the medical record quality management method as described in the first aspect of the present invention.

[0041] The beneficial effects of the present invention are as follows: by constructing a dynamic medical knowledge graph and matching it with the core fields of medical records, and using a graph reasoning algorithm to perform logical consistency verification on medical records, the intelligent and precise quality management of medical records is achieved. The construction of a dynamic medical knowledge graph breaks through the limitation that traditional static knowledge bases cannot be updated in real time, and improves the accuracy of matching the core fields of medical records with medical standard terms through semantic similarity analysis; while the graph reasoning algorithm comprehensively verifies the logical consistency in medical records by defining relationship existence and mutual exclusivity rules, automatically marks logical contradictions and generates quality audit reports, thereby significantly improving the standardization and audit efficiency of medical records, and solving the problems of non-standard processing of unstructured data and insufficient depth of logical verification in the prior art. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0043] Figure 1This is a flow chart of the medical record quality management method in Example 1.

[0044] Figure 2 This is a schematic diagram of the medical record quality management system in Example 1. DETAILED DESCRIPTION

[0045] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.

[0046] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0047] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0048] Example 1, reference Figure 1 and Figure 2 , which is the first embodiment of the present invention, and provides a medical record quality management method, comprising the following steps:

[0049] S1: Collect medical record data and pre-process the collected medical record data;

[0050] S1.1: Medical record data includes unstructured data (such as diagnosis descriptions, medical history records), semi-structured data (such as examination reports, imaging descriptions), and structured data (such as laboratory test results, vital signs records).

[0051] S1.2: Convert the collected medical record data into UTF-8 encoding format;

[0052] For example, medical record files recorded in GBK encoding are transcoded into UTF-8 format to ensure that data from different sources are correctly parsed and processed under unified encoding.

[0053] S1.3: Use NLP tools to split and annotate unstructured data by paragraphs and categories;

[0054] For example, NLP tools can be used to split “Patient’s chief complaint: headache for 3 days; past history: hypertension for 10 years” into “chief complaint” paragraph and “past history” paragraph, and mark them as corresponding categories.

[0055] S1.4: Standardize medical record data;

[0056] For example, "diabetes" in structured data is converted into the standard medical term code "E11.9" (ICD-10 code) to ensure the standardization and consistency of the term.

[0057] S1.5: Perform unit unification and value conversion on structured data.

[0058] For example, convert "hemoglobin: 13.5 g / dL" in the laboratory report to "hemoglobin: 135 g / L" to unify the data units for subsequent analysis.

[0059] S2: Use NLP to parse the preprocessed medical record data and extract the core fields of the medical record through NER;

[0060] S2.1: Use BERT to perform word segmentation and syntactic analysis on the preprocessed medical record data to extract the grammatical structure and candidate segments of the structured data;

[0061] Furthermore, first, the structured data is input into the BERT model, and contextual semantic modeling is performed on it to generate word vector representations for each word; then, the dependency syntactic analysis technology is combined to identify the grammatical structure of the structured data and the dependency relationship between sentences, and extract the important grammatical components of the sentence such as the subject, predicate, and object; candidate segments are divided according to the grammatical structure, for example, from the sentence "the patient complained of headache for 3 days accompanied by nausea", the candidate segments "main complaint: headache" and "accompanying symptoms: nausea" are extracted. This process provides an accurate semantic and syntactic basis for subsequent named entity recognition (NER) and core field extraction.

[0062] S2.2: Use BERT-NER to extract the core fields of medical records based on the grammatical structure and candidate segments of structured data;

[0063] Specifically: the structured data that has been segmented and syntactically analyzed is input into the BERT-NER model, and through its powerful contextual semantic understanding ability, each word in the candidate fragment is classified as an entity, for example, "headache" is identified as a "symptom entity" and "3 days" is identified as a "time entity"; then, combined with the medical terminology library and predefined entity categories in the field (such as symptoms, diagnosis, test results, treatment plans, etc.), the extracted core fields of the medical records are further classified and labeled, for example, "hypertension" is labeled as a "diagnosis entity" and "aspirin" is labeled as a "medication record entity"; finally, a structured set of core fields of medical records is generated to provide data support for subsequent field standardization and knowledge graph matching.

[0064] The core fields of medical records include symptoms, diagnosis, test results, treatment plans, and medication records;

[0065] S2.3: Standardize the mapping of extracted core fields of medical records based on the medical terminology library and replace the terms in structured data with standardized codes.

[0066] The medical terminology library in the field (such as ICD code, SNOMED-CT) is used to compare the core fields of the medical records extracted by the BERT named entity recognition model (BERT-NER) one by one to identify the standard medical terms corresponding to the fields; then, the non-standardized terms in the structured data are converted into standardized codes in the medical terminology library, for example, "diabetes" is mapped to the ICD code "E11.9", and "hypertension" is mapped to the SNOMED-CT code "38341003"; finally, a structured data set consisting of standardized fields and corresponding codes is generated to provide a standardized data foundation for subsequent dynamic medical knowledge graph matching and logical consistency verification.

[0067] S3: Build a dynamic medical knowledge graph and match it with the core fields of medical records;

[0068] S3.1: Standardize the logical associations between core fields of medical records and define standardized relationships through coding and semantic rules in the standard medical terminology library;

[0069] Determine the logical relationship type between core fields based on the medical terminology library, such as "disease-treatment plan" or "symptom-examination method"; use semantic rules to expand associations, such as clarifying the subordinate relationship between "antihypertensive drugs" and the specific drug "amlodipine" through the upper and lower relationships. This process provides standardized logical relationship data for the construction of dynamic medical knowledge graphs.

[0070] S3.2: Create nodes of the dynamic medical knowledge graph based on standardized coded terms, and construct edges of the dynamic medical knowledge graph based on standard relationships;

[0071] Specifically, a node refers to: based on standardized coding terms, the core fields of medical records (such as "hypertension" and "aspirin") are mapped to entity nodes in the dynamic medical knowledge graph.

[0072] Edges specifically refer to: logical association edges between nodes established in a dynamic medical knowledge graph based on standardized relationships (such as "disease-treatment plan" or "symptom-examination method").

[0073] S3.3: Based on the nodes of the dynamic medical knowledge graph and the edges of the dynamic medical knowledge graph, a dynamic medical knowledge graph is constructed using a graph database;

[0074] It should be noted that this process achieves efficient storage and query of medical knowledge. Through the graph database, the relationship between the core fields of medical records and medical terms can be quickly retrieved, supporting complex logical reasoning and path analysis. This process improves the scalability and dynamic update capabilities of the knowledge graph, and provides accurate and efficient data support for the logical consistency verification of medical records.

[0075] S3.4: Use Word2Vec embedding to calculate the semantic similarity between the core fields of medical records and entities in the dynamic medical knowledge graph. The expression is:

[0076] ;

[0077] in, A vector representing the core fields of a medical record In the The values ​​in the dimensions, A vector representing the core fields of the medical record, Represents the corresponding standard term vector in the dynamic medical knowledge graph The value in the kth dimension, represents the corresponding standard term vector in the dynamic medical knowledge graph, Represents the core field vector of the medical record Standard term vectors with dynamic medical knowledge graph The similarity between is the size of the vector dimension, is the index variable of the vector of core fields of medical records, is the index variable of the standard term vector;

[0078] It should be noted that by using Word2Vec embedding to calculate the semantic similarity between the core fields of medical records and entities in the dynamic medical knowledge graph, accurate matching between the core fields of medical records and standard terms is achieved. This method can effectively capture the semantic relationship of text, avoid incorrect mapping due to differences in expression or lexical ambiguity, improve the accuracy and reliability of medical record field standardization, and provide a high-quality data foundation for subsequent logical verification and dynamic medical knowledge graph construction.

[0079] S3.5: Based on the statistical analysis of the matching results between the core fields of historical medical records and the standard terms of the medical knowledge base, define the verification threshold S1;

[0080] when When ≥S1, the core fields of the medical record are considered to be consistent with medical logic and the match is successful;

[0081] when When <S1, the field is considered not to conform to medical logic and the match fails.

[0082] It should be noted that medical logic refers to the reasonable correlation between the core fields of medical records in medical knowledge and clinical practice, including the causal relationship, correlation and mutual exclusion between fields such as diseases, symptoms, examinations, treatment plans, and medications. For example, a certain disease should correspond to a reasonable treatment plan, certain drugs cannot be used at the same time, or certain test results must conform to specific pathological conditions. Medical logic ensures the scientificity, standardization and consistency of the content of medical records through standardized rules and knowledge systems, and is the key foundation for ensuring the quality of medical care and the scientificity of clinical decision-making.

[0083] S4: After the matching is completed, the graph reasoning algorithm is used to perform a logical consistency check on the medical records, mark the logical contradictions, and generate the verification results;

[0084] S4.1: Based on the results of the core field matching of the medical record, extract the successfully matched core fields and their corresponding node numbers of the dynamic medical knowledge graph and the medical logical relationships between the matching nodes, and organize the medical record field matching results and their logical relationships into triples ,in, represents the head node in the triple, Represents the head node With the tail node The logical relationship between Represents the tail node in a triple;

[0085] It should be noted that this process realizes the structured association between medical record data and medical knowledge graph. This organizational method can intuitively express the medical logical relationship between medical record fields, facilitate subsequent logical consistency verification and reasoning analysis, and improve the efficiency and accuracy of data processing, providing an efficient and structured foundation for medical record quality management.

[0086] S4.2: Based on the standardized relations in the dynamic medical knowledge graph, define the relation existence rules and relation mutual exclusivity rules;

[0087] Furthermore, the relationship existence rule: Check whether the logical relationship between the core fields of the medical record exists in the dynamic medical knowledge graph. For example, if the drug corresponding to the disease "hypertension" in the medical record is "ibuprofen", but the relationship "hypertension-treatment plan-ibuprofen" does not exist in the dynamic medical knowledge graph, it is marked as a logical contradiction.

[0088] Relationship mutual exclusivity rule: Check whether there is a mutually exclusive relationship between medical record fields. For example, if the medical record records both "hypertension" and "hypotension", and the dynamic medical knowledge graph defines "hypertension" and "hypotension" as mutually exclusive diseases, it will be marked as a logical conflict.

[0089] S4.3: Based on the relationship existence rule, relationship mutual exclusivity rule and graph inference algorithm, the logical consistency score between medical record fields is calculated. The expression is:

[0090] ;

[0091] in, Score the logical consistency between medical record fields, Indicates the number of triples that successfully match the core fields of the medical record. Indicates the head node corresponding to the core field of the medical record. Represents the standardized logical relationships defined in the dynamic medical knowledge graph. Indicates the tail node corresponding to the core field of the medical record. represents the existence rate of triples in the dynamic medical knowledge graph, Represents the head node and tail node The mutual exclusion rate, Represents the head node To the end node Path consistency, is a very small true number;

[0092] Furthermore, path consistency The expression is:

[0093] ;

[0094] in, Represents the starting node To the end node A specific path, represents the number of triplets, is a very small positive number;

[0095] S4.4: Based on the statistical analysis of historical data, define the logical consistency check threshold Q1;

[0096] when When ≥Q1, the current triple is marked as logically consistent;

[0097] when When <Q1, the current triple is marked as a logical contradiction.

[0098] It should be noted that by defining the logic consistency check threshold Q1 based on historical data statistical analysis, the logic check standard in different scenarios can be dynamically adapted. When the logic consistency score Q of the medical record triple is higher than the threshold Q1, it is marked as logically consistent; when it is lower than Q1, it is marked as a logical contradiction. This method effectively improves the accuracy and reliability of logic verification and provides a scientific basis for medical record data quality management and medical decision-making.

[0099] S5: Generate a medical record quality review report based on the verification results.

[0100] S5.1: The medical record quality review report includes the successfully matched core pathology fields and their logical relationships, the marked logical contradictions, the reasons for the contradictions, and the logical consistency scores between the medical record fields. .

[0101] It should be noted that the medical record quality review report presents the core medical record fields that have been successfully matched and their logical relationships, the marked logical contradictions and their specific reasons, and the logical consistency scores between the medical record fields. , providing a comprehensive and structured assessment of the quality of medical record data. The report can accurately locate logical problems in medical records, assist medical personnel to quickly correct errors, improve the standardization and scientificity of medical records, and provide reliable data support for subsequent clinical decision-making, data analysis and knowledge graph optimization.

[0102] This embodiment also provides a medical record quality management system, including: a data collection module, a field extraction module, a field matching module, a logic consistency verification module and a report generation module.

[0103] A data collection module is used to collect medical record data and pre-process the collected medical record data;

[0104] The field extraction module is used to parse the pre-processed medical record data using NLP and extract the core fields of the medical record through NER;

[0105] The field matching module is used to build a dynamic medical knowledge graph and match the dynamic medical knowledge graph with the core fields of the medical record;

[0106] The logic consistency check module is used to check the logic consistency of medical records using a graph reasoning algorithm after the matching is completed, mark logical contradictions, and generate verification results;

[0107] The report generation module is used to generate a medical record quality review report based on the verification results.

[0108] This embodiment also provides a computer device, which is suitable for the medical record quality management method, including: a memory and a processor; the memory is used to store computer executable instructions, and the processor is used to execute computer executable instructions to implement the medical record quality management method proposed in the above embodiment.

[0109] The computer device may be a terminal, and the computer device includes a processor, a memory, a communication interface, a display screen and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device may be a touch layer covered on the display screen, or a key, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse, etc.

[0110] This embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, the method for implementing the medical record quality management proposed in the above embodiment is implemented; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (Static Random Access Memory, referred to as SRAM), electrically erasable programmable read-only memory (Electrically Erasable Programmable Read-Only Memory, referred to as EEPROM), erasable programmable read-only memory (Erasable Programmable Read Only Memory, referred to as EPROM), programmable read-only memory (Programmable Red-Only Memory, referred to as PROM), read-only memory (Read-Only Memory, referred to as ROM), magnetic storage, flash memory, magnetic disk or optical disk.

[0111] In summary, the present invention realizes the intelligent and precise quality management of medical records by: constructing a dynamic medical knowledge graph and matching it with the core fields of medical records, and using a graph reasoning algorithm to perform logical consistency verification on medical records. The construction of a dynamic medical knowledge graph breaks through the limitation that the traditional static knowledge base cannot be updated in real time, and improves the accuracy of matching the core fields of medical records with medical standard terms through semantic similarity analysis; and the graph reasoning algorithm comprehensively verifies the logical consistency in medical records by defining relationship existence and mutual exclusivity rules, automatically marks logical contradictions and generates quality audit reports, thereby significantly improving the standardization and audit efficiency of medical records, and solving the problems of non-standard processing of unstructured data and insufficient depth of logical verification in the prior art. It should be noted that the above embodiments are only used to illustrate the technical solution of the present invention and not to limit it. Although the present invention is described in detail with reference to the preferred embodiments, ordinary technicians in the field should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the spirit and scope of the technical solution of the present invention, which should be included in the scope of the claims of the present invention.

Claims

1. A medical record quality management method, characterized in that: include: Collect medical record data and pre-process the collected medical record data; Use NLP to parse the preprocessed medical record data and extract the core fields of the medical record through NER; Build a dynamic medical knowledge graph and match it with the core fields of medical records; After the matching is completed, the graph reasoning algorithm is used to perform logical consistency verification on the medical records, mark logical contradictions, and generate verification results; Generate a medical record quality audit report based on the verification results; After the matching is completed, the graph reasoning algorithm is used to perform a logical consistency check on the medical record, mark the logical contradictions, and generate the verification results. The specific steps are as follows: According to the results of the core field matching of the medical record, the successfully matched core fields and their corresponding node numbers of the dynamic medical knowledge graph and the medical logical relationships between the matching nodes are extracted, and the medical record field matching results and their logical relationships are organized into triples. ,in, represents the head node in the triple, Represents the head node With the tail node The logical relationship between Represents the tail node in a triple; According to the standardized relations in the dynamic medical knowledge graph, the relation existence rules and relation mutual exclusivity rules are defined; Based on the relationship existence rule, relationship mutual exclusivity rule and graph inference algorithm, the logical consistency score between medical record fields is calculated. The expression is: ； in, Score the logical consistency between medical record fields, Indicates the number of triples that successfully match the core fields of the medical record. Indicates the head node corresponding to the core field of the medical record. Represents the standardized logical relationships defined in the dynamic medical knowledge graph. Indicates the tail node corresponding to the core field of the medical record. represents the existence rate of triples in the dynamic medical knowledge graph, Represents the head node and tail node The mutual exclusion rate, Represents the head node To the end node Path consistency, is a very small true number; Based on the statistical analysis of historical data, define the logical consistency check threshold Q1; when When ≥Q1, the current triple is marked as logically consistent; when When <Q1, the current triple is marked as a logical contradiction.

2. The medical record quality management method according to claim 1, characterized in that: The medical record data includes unstructured data, semi-structured data and structured data.

3. The medical record quality management method according to claim 2, characterized in that: The specific steps of preprocessing the collected medical record data are as follows: Convert the collected medical record data into UTF-8 encoding format; Use NLP tools to split and annotate unstructured data by paragraphs and categories; Standardize medical record data; Perform unit unification and value conversion on structured data.

4. The medical record quality management method according to claim 3, characterized in that: The NLP is used to parse the pre-processed medical record data, and the core fields of the medical record are extracted through NER. The specific steps are as follows: Use BERT to perform word segmentation and syntactic analysis on the preprocessed medical record data to extract the grammatical structure and candidate segments of the structured data; According to the grammatical structure and candidate segments of structured data, BERT-NER is used to extract the core fields of medical records; The extracted core fields of medical records are standardized and mapped based on the medical terminology library, and the terms in the structured data are replaced with standardized codes.

5. The medical record quality management method according to claim 4, characterized in that: The construction of the dynamic medical knowledge graph matches the dynamic medical knowledge graph with the core fields of the medical record. The specific steps are as follows: Through the coding and semantic rules in the standard medical terminology library, the logical associations between the core fields of the medical record are standardized and standardized relationships are defined; Create nodes of the dynamic medical knowledge graph based on standardized coded terms, and construct edges of the dynamic medical knowledge graph based on standard relationships; Based on the nodes of the dynamic medical knowledge graph and the edges of the dynamic medical knowledge graph, a dynamic medical knowledge graph is constructed using a graph database; Word2Vec embedding is used to calculate the semantic similarity between the core fields of medical records and entities in the dynamic medical knowledge graph. The expression is: ； in, A vector representing the core fields of a medical record In the The values ​​in the dimensions, A vector representing the core fields of the medical record, Represents the corresponding standard term vector in the dynamic medical knowledge graph The value in the kth dimension, represents the corresponding standard term vector in the dynamic medical knowledge graph, Represents the core field vector of the medical record Standard term vectors with dynamic medical knowledge graph The similarity between is the size of the vector dimension, is the index variable of the vector of core fields of medical records, is the index variable of the standard term vector; Based on the statistical analysis of the matching results between the core fields of historical medical records and the standard terms of the medical knowledge base, the verification threshold S1 is defined; when When ≥S1, the core fields of the medical record are considered to be consistent with medical logic and the match is successful; when When <S1, the field is considered not to conform to medical logic and the match fails.

6. The medical record quality management method according to claim 5, characterized in that: The medical record quality audit report includes the successfully matched core pathology fields and their logical relationships, the marked logical contradictions and contradiction reasons, and the logical consistency scores between medical record fields. .

7. A medical record quality management system, based on the medical record quality management method according to any one of claims 1 to 6, characterized in that: include: Data collection module, field extraction module, field matching module, logical consistency verification module and report generation module, A data collection module is used to collect medical record data and pre-process the collected medical record data; The field extraction module is used to parse the pre-processed medical record data using NLP and extract the core fields of the medical record through NER; The field matching module is used to build a dynamic medical knowledge graph and match the dynamic medical knowledge graph with the core fields of the medical record; The logic consistency check module is used to check the logic consistency of medical records using a graph reasoning algorithm after the matching is completed, mark logical contradictions, and generate verification results; The report generation module is used to generate a medical record quality review report based on the verification results.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the medical record quality management method described in any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the medical record quality management method described in any one of claims 1 to 6 are implemented.

Images