Research And Implementation Of Intelligent Assisted Disease Diagnosis Method Based On Belief Rule-Base Reasoning

To a certain extent,the disease diagnosis expert system can assist doctors to improve the efficiency of diagnosis and treatment and alleviate the dilemma of "difficult to see a doctor".Due to the diversity,heterogeneity and long process of diagnosis and treatment information collection,patient diagnosis and treatment data are usually missing,incomplete and uncertain,which makes it difficult to acquire knowledge,incomplete rules and dispersion of rule matching in the modeling and reasoning of expert systems,resulting in the poor diagnostic accuracy of expert systems.To address the above problems,this thesis investigates the method of intelligent assisted diagnosis of diseases based on belief rule-based(BRB)inference and applies it to the diagnosis of chronic atrophic gastritis in Chinese medicine and colorectal diseases in Western medicine,and the main work is as follows:(1)Diagnosis method of chronic atrophic gastritis based on the inference of belief rule base.Firstly,the clinical diagnosis and treatment experience knowledge of famous veteran TCM doctors and the data of chronic atrophic gastritis cases are comprehensively used to diagnose and analyze the disease,and the method of parameter linear matching activation rules based on S function is proposed to increase the focus of matching degree;then the BRB diagnostic inference model is optimized by GA algorithm;finally,the SVM diagnostic model and BP neural network diagnostic model are compared with the SVM diagnostic model for chronic atrophic gastritis Finally,the inference diagnosis results of chronic atrophic gastritis disease with SVM diagnosis model and BP neural network diagnosis model are compared and analyzed,and it is verified that the S-BRB diagnosis model can diagnose chronic atrophic gastritis disease more accurately.(2)A multi-level belief rule base inference-based diagnosis method for colorectal disease classification.Based on the S function proposed in(1),a multi-level BRB inference for colorectal disease auxiliary diagnosis model is constructed for the complex mapping relationship between clinical manifestation symptoms of colorectal diseases and types of colorectal diseases.Since each type of colorectal disease has different manifestation symptoms,different belief rule bases are established according to experts’ experience to make continuous hierarchical diagnosis of colorectal diseases,and a complete rule base is gradually generated through the accumulation of multi-level classification and hierarchical diagnosis rules;then the multi-level BRB inference diagnosis model is optimized by GA algorithm;finally,the diagnosis model of colorectal diseases is optimized by combining with SVM diagnosis model and BP Finally,by comparing the results with SVM diagnostic model and BP neural network diagnostic model for colorectal disease classification and diagnosis,it is verified that the continuous hierarchical BRB diagnostic model is more accurate and effective for the diagnosis of colorectal diseases.(3)Design and development of an intelligent assisted disease diagnosis platform based on belief rule base inference.The front-end and back-end of the platform are developed using Vue framework and Spring Boot framework,including user module,login module,diagnosis module and data transmission module,respectively.All kinds of users(doctors,administrators,etc.)can login directly through the browser.The platform and core functions were tested and applied using data from chronic atrophic gastritis in Chinese medicine and colorectal diseases in Western medicine,thus illustrating the effectiveness of the developed platform and methodology.