CN-121983294-A - Knowledge graph-based method and system for analyzing clinical data of chronic patients

CN121983294ACN 121983294 ACN121983294 ACN 121983294ACN-121983294-A

Abstract

The invention relates to the technical field of medical health, and discloses a method and a system for analyzing clinical data of a chronic patient based on a knowledge graph, wherein the method comprises the steps of acquiring a clinical data set; the method comprises the steps of semantic extraction to obtain structured event data, construction of personalized knowledge graph, concept mapping to obtain conceptual knowledge graph, correlation path mining to obtain multi-order time sequence correlation path, matching with historical data to determine missing items and updating collection strategy. According to the invention, through two-stage knowledge graph construction and two-dimensional deletion detection, the depth and data integrity of chronic clinical data analysis are improved.

Inventors

SONG HUI
LIN QIFENG
XIE MINGHUI
LIN CHENG
WANG XIANJUN

Assignees

福州中康智慧科技股份有限公司

Dates

Publication Date: 20260505
Application Date: 20260407

Claims (10)

1. A method for analyzing clinical data of a chronically ill patient based on a knowledge graph, the method comprising: a1, acquiring a clinical data set of a target patient, wherein the clinical data set comprises identification information of the target patient, clinical examination index data and a clinical record text; a2, carrying out semantic extraction on the clinical record text to obtain structured event data of the target patient; a3, constructing a personalized knowledge graph of the target patient by taking the identity information as a node and the clinical examination index data and the structured event data as connecting edges; a4, performing conceptual mapping on nodes and attribute sides in the personalized knowledge graph according to a preset medical knowledge ontology to obtain a conceptual knowledge graph of the target patient; a5, carrying out association path mining on the conceptual knowledge graph to obtain a multi-order time sequence association path of the target patient; a6, matching the multi-order time sequence association path with the historical clinical event data of the target patient, and determining data missing items in the clinical data set according to a matching result so as to update the data acquisition strategy of the clinical data set.
2. The method for analysis of clinical data of a chronically ill patient based on a knowledge-graph according to claim 1, wherein the acquiring of the clinical data set of the target patient comprises: calling an original clinical data record matched with the identity information of the target patient from an electronic medical record database in the hospital information terminal; performing data cleaning on the original clinical data record to generate a standardized clinical record of the target patient; And synchronously extracting clinical examination index data and clinical record text in the standardized clinical record.
3. The method for analyzing clinical data of a chronically ill patient based on a knowledge graph according to claim 1, wherein the semantic extraction of the clinical record text to obtain the structured event data of the target patient comprises: Word segmentation processing is carried out on the clinical record text to obtain a text word segmentation sequence of the clinical record text; Carrying out named entity recognition on the text word segmentation sequence, and recognizing medical entities in the text word segmentation sequence to serve as candidate entity sets of the clinical record text; Performing relationship classification on any medical entity in the candidate entity set, and determining the semantic relationship of the candidate entity set; based on the semantic relationship, carrying out relationship instantiation on the candidate entity set to obtain an original relationship triplet of the target patient; And extracting time information of the original relation triplet from the clinical record text, and combining the time information serving as a time tag with the original relation triplet to obtain the structured event data of the target patient.
4. The method for analyzing clinical data of a chronic patient based on a knowledge graph according to claim 2, wherein the constructing a personalized knowledge graph of the target patient by using the identification information as a node and the clinical examination index data and the structured event data as a connection side comprises: Taking the identity information as a central node, and constructing an initial graph structure of the target patient; taking an event subject, an event action and an event object in the structured event data as event nodes, and carrying out instantiation association on the event nodes and the central node to obtain a first type connection edge of the initial graph structure; Taking the examination index in the clinical examination index data as an index node, and carrying out attributive mapping on the index node and the central node to obtain a second type connecting edge of the initial graph structure; in the time dimension, constructing a third type of connection edge of the initial graph structure according to the adjacent relation between the event node and the index node; and injecting the center node, the first type connecting edge, the second type connecting edge and the third type connecting edge into the initial graph structure to obtain the personalized knowledge graph of the target patient.
5. The method of claim 4, wherein said constructing a third type of connection edge of the initial graph structure in the time dimension according to the proximity relationship between the event node and the index node comprises: Synchronously acquiring a first time sequence label and a second time sequence label corresponding to the event node and the index node; Performing difference value analysis on the first time sequence tag and the second time sequence tag to obtain a time difference value of the initial graph structure; And establishing an association relation between the event node and the index node according to the time difference value to obtain a third type connecting edge of the initial graph structure.
6. The method for analyzing clinical data of a chronically ill patient based on a knowledge graph according to claim 1, wherein the performing conceptual mapping on nodes and attribute edges in the personalized knowledge graph according to a preset medical knowledge ontology to obtain a conceptual knowledge graph of the target patient comprises: acquiring a preset medical knowledge body, wherein the medical knowledge body comprises concept nodes and concept relation edges; traversing nodes to be mapped in the personalized knowledge graph, and calculating initial matching degree between attribute information of the nodes to be mapped and concept nodes in the medical knowledge body; Selecting a concept node with the highest matching degree from the medical knowledge body for the node to be mapped as a mapping target according to the initial matching degree, and obtaining a node mapping relation of the personalized knowledge graph; Traversing attribute edges to be mapped in the personalized knowledge graph, and carrying out association matching on the attribute edges to be mapped and the conceptual relationship edges to obtain an attribute edge mapping relationship of the personalized knowledge graph; based on the node mapping relation and the attribute edge mapping relation, carrying out conceptual conversion on the personalized knowledge graph to obtain a conceptual knowledge graph of the target patient.
7. The method for analyzing clinical data of chronic patients based on knowledge graph according to claim 6, wherein calculating the initial matching degree between the attribute information of the nodes to be mapped and the concept nodes in the medical ontology comprises: acquiring node type identifiers, associated nodes and attribute information of the nodes to be mapped, and acquiring concept type identifiers, concept association modes and definition characteristics of the concept nodes; Determining a type matching factor between the node to be mapped and the concept node according to the consistency of the node type identifier and the concept type identifier; Evaluating the semantic matching degree between the nodes to be mapped and the concept nodes according to the attribute information and the definition characteristics; based on the association node and the concept association mode, quantifying the structural matching degree between the node to be mapped and the concept node; Determining a first weight factor of the type matching factor, a second weight factor of the semantic matching degree and a third weight factor of the structural matching degree according to a clinical expert experience library, wherein the clinical expert experience library comprises historical case samples and expert annotation matching results; According to the type matching factor, the semantic matching degree and the structural matching degree, calculating initial matching degree between attribute information of the node to be mapped and concept nodes in the medical knowledge body, wherein a calculation formula of the initial matching degree is as follows: ; In the formula, For the initial degree of matching between the nodes to be mapped and the concept nodes, For the type of the matching factor to be used, For the degree of matching of the semantics, For the degree of matching of the structures described, For the first weight factor to be the first weight factor, For the second weight factor to be the second weight factor, Is the third weight factor, and satisfies α+β+γ=1.
8. The knowledge-graph-based chronic disease patient clinical data analysis method according to claim 1, wherein the performing association path mining on the conceptual knowledge graph to obtain a multi-order time sequence association path of the target patient comprises: Taking the nodes with time attribute information in the conceptual knowledge graph as time sensitive nodes; Taking the time sensitive node as a starting point, and carrying out path expansion in the conceptual knowledge graph according to a time sequence to obtain an initial association path set of the target patient; performing time sequence consistency test on the initial association paths in the initial association path set to obtain a time sequence consistency path set of the target patient; According to a preset time window parameter, performing time sequence adjacency screening on the time sequence identical path set to obtain a time sequence adjacency path set of the target patient; And carrying out path length statistics on the time sequence adjacent path set, and marking a path with a statistical result reaching a preset order threshold as a multi-order time sequence associated path of the target patient.
9. The knowledge-graph-based chronic disease patient clinical data analysis method according to claim 1, wherein the matching the multi-order time-series correlation path with the historical clinical event data of the target patient, determining the data missing items in the clinical data set according to the matching result, so as to update the data acquisition strategy of the clinical data set, comprises: Acquiring historical clinical event data of the target patient; Performing node level matching on nodes in the multi-order time sequence association path and the historical clinical event data to obtain a first type data missing item of the clinical data set; Performing edge level matching on the concept relation edges in the multi-order time sequence association paths and the historical clinical event data to obtain second-class data missing items of the clinical data set; And updating the data acquisition strategy of the clinical data set according to the first type data missing items and the second type data missing items so as to optimize the subsequent data acquisition of the target patient.
10. A knowledge-graph-based clinical data analysis system for chronically ill patients, for implementing the knowledge-graph-based clinical data analysis method of claim 1, the system comprising: the system comprises a data acquisition module, a semantic extraction module, a map construction module, a concept mapping module, a path mining module and a data updating module, wherein: the data acquisition module is used for acquiring a clinical data set of a target patient, wherein the clinical data set comprises identification information of the target patient, clinical examination index data and a clinical record text; The semantic extraction module is used for carrying out semantic extraction on the clinical record text to obtain the structured event data of the target patient; The map construction module is used for constructing a personalized knowledge map of the target patient by taking the identity information as a node and the clinical examination index data and the structured event data as connecting edges; The concept mapping module is used for performing concept mapping on the nodes and the attribute sides in the personalized knowledge graph according to a preset medical knowledge ontology to obtain a conceptual knowledge graph of the target patient; The path mining module is used for carrying out association path mining on the conceptual knowledge graph to obtain a multi-order time sequence association path of the target patient; The synchronous updating module is used for matching the multi-order time sequence association path with the historical clinical event data of the target patient, and determining data missing items in the clinical data set according to a matching result so as to update the data acquisition strategy of the clinical data set.

Description

Knowledge graph-based method and system for analyzing clinical data of chronic patients Technical Field The invention relates to the technical field of medical health, in particular to a method and a system for analyzing clinical data of chronic patients based on a knowledge graph. Background In the prior art, only a unified knowledge graph is constructed, and a personalized example of a patient and a medical standard concept are not distinguished, so that the graph cannot consider individual specificity and medical standardization; the side construction of the knowledge graph only considers semantic association, does not consider the adjacent relation of the time dimension, and cannot capture a time sequence causal chain in the chronic disease development process; The data missing detection is only aimed at the node entity, and is not aimed at the relation edge, so that the logic association missing among clinical events cannot be found; the lack of a closed-loop mechanism for dynamically optimizing the acquisition strategy according to the missing detection result cannot realize continuous improvement of data quality. Disclosure of Invention The invention provides a method and a system for analyzing clinical data of chronic patients based on a knowledge graph, which are used for solving the problems in the background technology. In order to achieve the above object, the present invention provides a method for analyzing clinical data of a chronic patient based on a knowledge graph, comprising: a1, acquiring a clinical data set of a target patient, wherein the clinical data set comprises identification information of the target patient, clinical examination index data and a clinical record text; a2, carrying out semantic extraction on the clinical record text to obtain structured event data of the target patient; a3, constructing a personalized knowledge graph of the target patient by taking the identity information as a node and the clinical examination index data and the structured event data as connecting edges; a4, performing conceptual mapping on nodes and attribute sides in the personalized knowledge graph according to a preset medical knowledge ontology to obtain a conceptual knowledge graph of the target patient; a5, carrying out association path mining on the conceptual knowledge graph to obtain a multi-order time sequence association path of the target patient; a6, matching the multi-order time sequence association path with the historical clinical event data of the target patient, and determining data missing items in the clinical data set according to a matching result so as to update the data acquisition strategy of the clinical data set. In a preferred embodiment, the acquiring a clinical dataset of a target patient comprises: calling an original clinical data record matched with the identity information of the target patient from an electronic medical record database in the hospital information terminal; performing data cleaning on the original clinical data record to generate a standardized clinical record of the target patient; And synchronously extracting clinical examination index data and clinical record text in the standardized clinical record. In a preferred embodiment, the semantic extraction of the clinical record text to obtain structured event data of the target patient includes: Word segmentation processing is carried out on the clinical record text to obtain a text word segmentation sequence of the clinical record text; Carrying out named entity recognition on the text word segmentation sequence, and recognizing medical entities in the text word segmentation sequence to serve as candidate entity sets of the clinical record text; Performing relationship classification on any medical entity in the candidate entity set, and determining the semantic relationship of the candidate entity set; based on the semantic relationship, carrying out relationship instantiation on the candidate entity set to obtain an original relationship triplet of the target patient; And extracting time information of the original relation triplet from the clinical record text, and combining the time information serving as a time tag with the original relation triplet to obtain the structured event data of the target patient. In a preferred embodiment, the constructing the personalized knowledge graph of the target patient with the identification information as a node and the clinical examination index data and the structured event data as connecting edges includes: Taking the identity information as a central node, and constructing an initial graph structure of the target patient; taking an event subject, an event action and an event object in the structured event data as event nodes, and carrying out instantiation association on the event nodes and the central node to obtain a first type connection edge of the initial graph structure; Taking the examination index in the clinical examination index data as an