CN-122025176-A - Liver, gall, pancreas and spleen nursing decision-making system based on knowledge graph

CN122025176ACN 122025176 ACN122025176 ACN 122025176ACN-122025176-A

Abstract

The invention relates to the technical field of pre-diagnosis and health management, in particular to a liver, gall, pancreas and spleen nursing decision-making system based on a knowledge graph, which comprises a term collecting module, a semantic recognition module, a path positioning module, a node verification module and a decision-making reasoning module. According to the invention, the structural extraction of symptom expression and nursing behavior is realized by collecting nursing record content and constructing a semantic input sequence, the expression standardization level of terms is improved by combining segment identification generation term pairs, a semantic comparison mechanism is formed by keyword cross mapping, the matching association between terms and path nodes is enhanced, the path structure closure of the node pairs which do not accord with a logic sequence is ensured by path semantic structure verification and elimination, the accurate butt joint between a nursing scene and knowledge reasoning is realized by path entry marks and context mapping, so that nursing information expression has systematicness and standardability, the path generation process has continuity and logicality, and the decision record has traceability and calculability.

Inventors

CHENG SHI
ZHOU JIANMEI
HUANG LIN

Assignees

南通大学

Dates

Publication Date: 20260512
Application Date: 20260130

Claims (9)

1. The liver, gall, pancreas and spleen nursing decision-making system based on the knowledge graph is characterized by comprising: the term collection module collects the contents of liver, gall, pancreas and spleen nursing records, classifies structured information of nursing behavior records and symptom expression records, extracts syntax units in each text segment, and constructs a nursing semantic input sequence; The semantic recognition module analyzes a syntactic structure and marks part-of-speech features based on the nursing semantic input sequence, screens a combined segment comprising noun phrases and verb phrases, records standard terms and generates a nursing term pair list; The path positioning module reads the established liver, gall, pancreas and spleen standard path node labels in the knowledge graph according to the nursing term pair list, extracts keywords in the standard terms and keywords in the node labels to execute synonym cross mapping, counts the cross mapping group and establishes a semantic matching comparison table; The node verification module analyzes the path semantic types represented by each group of node labels based on the semantic matching comparison table, marks the node pairs with logic jump failure, extracts the node pairs which are not marked as structure closed sequences, and outputs a structure closed path fragment set; And marking each path as an independent item according to the structure closed path segment set by the decision-making reasoning module, marking the corresponding nursing situation of the starting node and the tail node of each path, mapping the corresponding nursing situation to a knowledge graph reasoning engine interface, and establishing a liver, gall, pancreas and spleen nursing decision record.
2. The hepatobiliary and pancreatic care decision system based on a knowledge graph according to claim 1, wherein said standard terms specifically refer to said combined fragments comprising both a symptom status description noun phrase and a career action verb phrase.
3. The system for liver, gall, pancreas and spleen care decision-making based on a knowledge graph of claim 1, wherein the logic jump failure node pair specifically refers to a path semantic structure of judging whether adjacent node pairs in a path sequentially meet symptom terms, judging terms and nursing terms, and marking the logic jump failure node if any node pair violates a corresponding structure sequence.
4. The liver, gall, pancreas and spleen care decision system based on the knowledge graph of claim 1, wherein the care semantic input sequence comprises a syntactic structure tree, part-of-speech tagging features and semantic classification labels, the care term pair list comprises symptom behavior term pairs, syntactic combination fragment indexes and structural semantic mapping labels, the semantic matching comparison table comprises node label keywords, standard term keywords and synonym mapping relations, the structural closed path fragment set comprises path sequence numbers, node sequence labels and semantic structure identifiers, and the liver, gall, pancreas and spleen care decision records comprise care path entry numbers, path start care situations, path end care situations and reasoning interface mapping numbers.
5. The knowledge-based hepatobiliary pancreatic care decision system of claim 1, wherein said term-aggregation module comprises: The term identification sub-module collects liver, gall, pancreas and spleen nursing record contents, after splitting according to record paragraphs, each segment of contents is compared with a nursing term library, according to the vocabulary entry and the belonging classification label in a glossary, matched terms are compared for marking, and marking position mapping is carried out according to the original text sequence, so that a term matching marking list is obtained; The syntactic structure extraction submodule obtains a corresponding text segment according to the term matching labeling list, performs word segmentation and part-of-speech labeling on verbs, nouns and prepositions in sentences according to a fixed grammar rule, and integrates word sequences and attachment levels according to the subordinate relations among dependency structure identification components to obtain a syntactic structure node sequence; the semantic sequence construction submodule classifies and sorts the syntactic components based on the syntactic structure node sequence and based on the main verb type, adds semantic identification by combining term matching labeling information, carries out structural conversion according to nursing behavior classes and symptom expression classes, and generates a nursing semantic input sequence after the semantic components are recombined.
6. The knowledge-based hepatobiliary pancreatic care decision system of claim 1, wherein said semantic recognition module comprises: The part-of-speech tagging submodule acquires the nursing semantic input sequence, performs word segmentation on each semantic text record, detects terms, recognizes part-of-speech category according to a general part-of-speech tagging set, determines a word formation form through word and context word order relation, tags part-of-speech features and positions of the terms, and generates a part-of-speech feature tagging sequence; The phrase combination extraction submodule identifies continuous occurrence of the combination of the part of speech and the word term of the part of speech and the combination of the word term of the verb class based on the part of speech feature labeling sequence, performs unified standard treatment on the internal word sequence of the combined phrase, and filters through a front-back attachment structure to screen and obtain a candidate phrase combination fragment set; and the term pair screening submodule judges whether each group of combined fragments simultaneously comprises noun phrases describing symptom states and verb phrases indicating nursing behaviors according to the candidate phrase combined fragment set, and if the noun phrases and the verb phrases exist simultaneously, the term pair screening submodule determines standard term combinations and generates a nursing term pair list.
7. The knowledge-based hepatobiliary pancreatic care decision system of claim 1, wherein said path localization module comprises: The path tag reading submodule acquires the nursing term pair list, reads the liver, gall, pancreas and spleen standard path node data constructed in the knowledge graph, extracts tag content under each node, extracts semantic field terms in the tags according to the sequence of the nodes, establishes a reference index structure between the terms and the node numbers, and generates a path node semantic term set; The keyword cross mapping submodule extracts the content of each term in the nursing term pair based on the path node semantic term set, analyzes the term vector relation between the term and the node semantic term, performs merging matching on the term pairs with semantic approach values smaller than a set semantic threshold, and counts the mapping quantity of each group of term pairs in each node label to obtain a semantic cross mapping group set; the semantic comparison construction submodule integrates the semantic mapping quantity between the term pair and the node labels according to the semantic cross mapping group set, constructs a one-to-many semantic pairing structure between the term pair and the node path labels, and establishes a semantic matching comparison table by taking a mapping strength value as a sequencing benchmark.
8. The knowledge-graph-based hepatobiliary pancreatic care decision system of claim 1, wherein said node verification module comprises: the semantic type analysis submodule acquires the label data of each node in the semantic matching comparison table, extracts the semantic term content of the term corresponding to each group of labels, marks each node term as a symptom class, a judgment class or a nursing class, and generates a path node semantic type label set; the path structure verification submodule builds an ordered combination of node labels in a path structure based on the path node semantic type label set, sequentially judges the structure mode of adjacent node pairs, and marks the corresponding node pairs as logic jump failure node pairs if the semantic sequence of symptom terms, judgment terms and nursing terms is not met, so as to obtain a logic jump failure node pair list; And the closed path extraction submodule eliminates all marked node pairs from the original path sequence according to the logic jump failure node pair list, reserves unmarked continuous node combinations, outputs the continuous node combinations according to the path sequence and establishes a structure closed path fragment set.
9. The knowledge-based hepatobiliary pancreatic care decision system of claim 1, wherein said decision inference module comprises: The path item labeling sub-module acquires the structure closed path fragment set, distributes unique item numbers to each path according to the sequence of the path fragments, positions the position indexes of the initial node and the tail node of the path in the original structure, establishes a mapping relation between the item numbers and the node indexes, and generates a path item labeling comparison table; The nursing situation mapping sub-module extracts semantic term contents contained in the initial node and the tail node of each path according to the path item labeling comparison table, labels the initial node as an input situation and the tail node as a response situation according to nursing context classification labels corresponding to the semantic terms, and builds situation structure mapping to obtain a path nursing situation correspondence table; And the map reasoning generation submodule inputs each path segment and corresponding initial and tail nursing situations into a map reasoning engine interface based on the path nursing situation correspondence table, judges node communication, outputs path segment information which can be mapped in a map and establishes liver, gall, pancreas and spleen nursing decision records.

Description

Liver, gall, pancreas and spleen nursing decision-making system based on knowledge graph Technical Field The invention relates to the technical field of pre-diagnosis and health management, in particular to a liver, gall, pancreas and spleen nursing decision-making system based on a knowledge graph. Background The technical field of pre-diagnosis and health management mainly relates to the monitoring and intervention of individual health states before or at early stage of occurrence of diseases by means of data acquisition, medical evaluation, health file analysis and the like, and carries out risk prediction, early screening and dynamic management on chronic diseases and major diseases, wherein the method comprises the steps of physiological parameter acquisition, health evaluation model construction, health intervention path establishment, fusion and intelligent analysis of multi-source health data, fusion of clinical medicine, bioinformatics, artificial intelligence, big data and other cross technologies, and forms a health management system framework characterized by intelligence, individuation and continuity, wherein a liver, gall, pancreas and spleen nursing decision system is a system applied to liver, gall, pancreas and spleen related diseases nursing process and used for assisting nursing staff in health state evaluation and nursing plan establishment, generally carries out nursing scheme selection by judging through nursing staff experience based on paper document record information, or carries out data and nursing suggestion by adopting an electronic health record mode, and comprises the steps of carrying out manual judgment on nursing levels according to test indexes and main complaints, combining clinical guideline text extraction related key points, carrying out decision making and regulating of a reference procedure record nursing plan and the like. The existing liver, gall and pancreas and spleen nursing technology mainly relies on subjective interpretation of paper documents or electronic health records by nursing staff, systematic induction and accurate classification of nursing information are difficult to achieve in actual operation, structural association is lacking between symptom expression and nursing behaviors, so that health state assessment is easily affected by experience level, nursing path selection lacks consistency and logic, clear nursing decision flow is difficult to form, nursing key point extraction relies on text retrieval and manual interpretation, the problems of low efficiency and poor accuracy exist, disease course data and nursing opinion are loosely associated, dynamic adjustment of nursing schemes and multi-round decision response are difficult to adapt to nursing complexity and diversity caused by disease evolution in long-term application, and knowledge-based data mining and continuous optimization cannot be effectively supported. Disclosure of Invention In order to solve the technical problems in the prior art, the embodiment of the invention provides a liver, gall, pancreas and spleen nursing decision-making system based on a knowledge graph. The technical scheme is as follows: In one aspect, a system for determining liver, gall, pancreas and spleen care based on a knowledge graph is provided, the system comprising: the term collection module collects the contents of liver, gall, pancreas and spleen nursing records, classifies structured information of nursing behavior records and symptom expression records, extracts syntax units in each text segment, and constructs a nursing semantic input sequence; The semantic recognition module analyzes a syntactic structure and marks part-of-speech features based on the nursing semantic input sequence, screens a combined segment comprising noun phrases and verb phrases, records standard terms and generates a nursing term pair list; The path positioning module reads the established liver, gall, pancreas and spleen standard path node labels in the knowledge graph according to the nursing term pair list, extracts keywords in the standard terms and keywords in the node labels to execute synonym cross mapping, counts the cross mapping group and establishes a semantic matching comparison table; The node verification module analyzes the path semantic types represented by each group of node labels based on the semantic matching comparison table, marks the node pairs with logic jump failure, extracts the node pairs which are not marked as structure closed sequences, and outputs a structure closed path fragment set; And marking each path as an independent item according to the structure closed path segment set by the decision-making reasoning module, marking the corresponding nursing situation of the starting node and the tail node of each path, mapping the corresponding nursing situation to a knowledge graph reasoning engine interface, and establishing a liver, gall, pancreas and spleen nursing decision record. As a further aspect