The Establishment And Preliminary Evaluation Of Intelligent Auxiliary Diagnosis System For Vertigo

【Background】Vertigo is a common clinical symptom of a patient who feels that he or she is subject to rotation due to motor hallucinations or spatial imagery.Experts on vertigo in Germany and the United States found in epidemiological investigations that 1 in 5 adults suffer from vertigo.The prevalence also increases with age,and the prevalence of women is higher than that of men.Domestic vertigo experts also show through clinical data that vertigo is one of the common clinical symptoms.In the past three decades,with the in-depth clinical diagnosis of vertigo made by Chinese doctors,some progress has been made,but clinicians(even neurologists)generally feel that they lack confidence in the diagnosis and treatment of vertigo patients,it is difficult to many patients to obtain Effective diagnosis which delays the course of disease,so early differential diagnosis of vertigo is essential.With the continuous development of artificial intelligence technology,more and more are applied to the clinical diagnosis and treatment process.In the field of intelligent assisted diagnosis of dizziness,foreign experts have also developed some systems,such as "Vertigo","Carrusel","ONE","CADINO" These systems have their own characteristics,which lays the foundation for the research and implementation of intelligent assisted diagnosis system for vertigo.At present,the experts in China who have designed and applied the vertigo intelligent auxiliary diagnosis system,but we have not yet obtained the application.【Objective】Establish an intelligent auxiliary diagnosis system for dizziness,conduct preliminary verification of the system through clinical cases,evaluate its application value in clinical diagnosis and treatment,and help clinicians,especially grassroots clinicians,to establish a good dizziness consultation in the process of using the system Thinking,and facilitate the collection of disease-related data of patients with vertigo,and provide data support for the continuous improvement of the intelligent auxiliary diagnosis system for post-vertigo.【Methods】By establishing a knowledge base for vertigo diagnosis and treatment,the standard for diagnosis and treatment of all diseases that cause vertigo is structured.To the vertigo diagnosis and treatment expert,through the form of questionnaire survey,clarify the common types of vertigo and the important parameters and weights of diagnosis,tabulate the parameters and weights,set the disease with the diagnosis label,and establish the diagnosis model.At the same time,among the outpatient cases of vertigo in Xijing Hospital,290 cases were recorded with complete and supported by etiological evidence,and they were tabulated with the parameters established in the previous period.After modeling and sorting cases,enter the software development stage,express medical knowledge to software developers in various ways,formulate design rules for interface parameters and page elements,and software developers assist in completing architectural design and language development.In view of the large number of diagnosis and treatment parameters and the long use time,the scientific version and clinical version of the auxiliary diagnosis system were developed at the same time.After the software development is completed,the system is tested in multiple directions,and a dizziness diagnosis man-machine competition is held to verify the diagnosis efficiency of the system.【Results】1.DeepDoc vertigoEstablished an intelligent auxiliary diagnosis system for vertigo-"DeepDoc Vertigo",a medical expert system based on experience.The system can provide diagnosis of eight common causes of vertigo diseases,including: BPPV,VM,VN,VP,BVP,PPPD,MD,SH.and 23 other diseases with special labels.The system is used by downloading and installing APP via mobile phone.2.Systematic clinical validationThe system is tested and the diagnosis coincidence rate is 78-100%.Further verify the effectiveness of the system through human-machine competition.The results of the human-machine competition for BPPV diagnosis show that the diagnosis rate of DeepDoc diagnosis is significantly higher than that of specialists and general physicians in primary hospitals.The human-machine competition for diagnosing the cause of dizziness showed that the diagnostic agreement rate of DeepDoc was significantly higher than the average diagnosis rate of primary doctors.【Conclusion】Based on the relevant guidance,expert experience,and clinical cases of dizziness diagnosis and treatment,the intelligent assisted diagnosis system DeepDoc for dizziness is designed and implemented.This system can provide decision support for clinicians to diagnose dizziness diseases,and can help clinicians to establish good diagnosis and treatment thinking of dizziness diseases.Further standardizing the inquiry process can also facilitate the collection of medical records of patients with vertigo,and provide a data source for the big data analysis required for the subsequent upgrade and improvement of the intelligent auxiliary diagnosis system for vertigo.At the same time,this study also initiated a useful exploration of domestic "artificial intelligence +dizziness",and provided a reference for the development of informatization and intellectualization of medical diagnosis in China.