Abstract:Objective: In view of the standardization of clinical diagnosis and treatment of the acute abdomen and the inheritance of diagnosis and treatment experience of prestigious veteran traditional Chinese medicine（TCM） doctors, a diagnosis and treatment reasoning algorithm based on association rule mining under incomplete evidence（AMIE）+ random walk was proposed to provide information services and technical support for primary doctors by recommending personalized diagnosis and treatment plans based on medical records.Method: The experience of diagnosis and treatment of acute abdomen of prestigious veteran TCM doctors and the text data of clinical diagnosis and treatment guidelines of integrated TCM and western medicine were collected to complete the task of knowledge extraction and construct acute abdomen knowledge graph based on Neo4j. On the basis of ontology-supported rule-based reasoning, the rule reasoning based on similar syndromes was used to expand the syndrome combinations whose Jaccard similarity was greater than the threshold in the syndrome recommendation results. The semantic path coverage algorithm was used to calculate the semantic similarity between the symptom nodes. The symptom nodes were divided into 10 categories, and the symptom nodes in the same category were extended. The random walk algorithm was used to search the symptom nodes connected with the syndrome, and the connection rules between the syndrome and symptom nodes were extended to realize the knowledge reasoning of AMIE+ random walk. Result: The acute abdomen knowledge graph included 1 320 nodes and 2 464 relationships. According to the link prediction evaluation index of knowledge reasoning, the reasoning results of the three algorithms in the auxiliary diagnosis and treatment of acute abdomen were compared. The AMIE+ random walk algorithm complemented the knowledge graph by extending the similar syndrome connection rules and the syndrome-symptom connection rules. Compared with the knowledge reasoning algorithm based on ontology rules, the area under the curve（AUC） was 15.18% higher and the accuracy was 30.36% higher, which achieved more accurate and effective knowledge inference.Conclusion: This study used knowledge graph technology to visualize the diagnosis and treatment of acute abdomen with TCM and western medicine, assisting primary clinicians in intuitively viewing the diagnosis and treatment process and data relationship. The proposed diagnosis and treatment reasoning algorithm can realize the personalized diagnosis and treatment plan recommendation at the level of "disease-syndrome-diagnosis-treatmentprescription", which can assist primary doctors in disease diagnosis and treatment and clinical decision-making, contribute to the knowledge sharing and application of diagnosis and treatment experience and clinical guidelines of prestigious veteran TCM doctors, improve the level of primary clinical diagnosis and treatment, and promote the normalization and standardization of the diagnosis and treatment process of acute abdomen with integrated TCM and western medicine.

Research on Viscera Syndrome Differentiation in TCM Based on Multiple Link Prediction Reasoning

Traditional Chinese Medicine Automated Diagnosis Based on Knowledge Graph Reasoning

Auxiliary Diagnosis Based on the Knowledge Graph of TCM Syndrome

Diagnosis and Treatment Reasoning of Integrated Traditional Chinese and Western Medicine Against Acute Abdomen Based on Knowledge Graph

Decision-Making System for the Diagnosis of Syndrome Based on Traditional Chinese Medicine Knowledge Graph

An Approach to Syndrome Differentiation in Traditional Chinese Medicine Based on Neural Network

Research on the Medical Knowledge Deduction Based on the Semantic Relevance of Electronic Medical Record

Diagnosis knowledge constrained network based on first-order logic for syndrome differentiation

A Traditional Chinese Medicine Syndrome Classification Model Based on Cross-Feature Generation by Convolution Neural Network: Model Development and Validation

TCM herbal prescription recommendation model based on multi-graph convolutional network

Decision tree method to extract syndrome differentiation rules of posthepatitic cirrhosis in traditional Chinese medicine

Study on intelligent syndrome differentiation neural network model of stomachache in traditional Chinese medicine based on the real world

A TCM Syndrome Differentiation Algorithm based on Deep Belief Network

Combining the External Medical Knowledge Graph Embedding to Improve the Performance of Syndrome Differentiation Model

Research on Knowledge Reasoning of TCM Based on Knowledge Graphs

Research on a Vehicle-Mounted Intelligent TCM Syndrome Differentiation System Based on Deep Belief Network

Predicting Herb-disease Associations Through Graph Convolutional Network

Comparison of Support Vector Machine, Back Propagation Neural Network and Extreme Learning Machine for Syndrome Element Differentiation

A KG-Enhanced Multi-Graph Neural Network for Attentive Herb Recommendation

Semantic-Aware Graph Convolutional Networks for Clinical Auxiliary Diagnosis and Treatment of Traditional Chinese Medicine

Feature Selection and Syndrome Prediction for Liver Cirrhosis in Traditional Chinese Medicine