Research And Application Of Personalized Doctor Recommendation For Intelligent Guidance Service

With the continuous improvement of the smart medical system,the intelligent guidance system is gradually replacing the service mode of manual guidance,and has become a new direction for the future development of the medical industry.At present,for most of the domestic intelligent guidance services,patients are still in a passive service mode and they need to find a doctor according to the hospital’s department classification or according to the form of human body parts.Due to lack of professional medical knowledge,patients are prone to problems such as greater blindness and excessive concentration of medical experts when choosing a doctor,they cannot choose the right doctor for medical treatment according to their illness,which makes their medical treatment inefficient and the doctor resources cannot be rationally used.Therefore,this paper focuses on the in-depth discussion and research on the personalized doctor recommendation in the intelligent guidance service,and recommends the most suitable doctor for the patient’s disease by deeply digging up the patient’s commands.While bringing efficient medical services to patients,it can effectively improve patient satisfaction and rational use of hospital resources,bringing maximum value to patients and hospitals.This study first builds a knowledge base in the vertical medical field,and then implements coarse-grained department recommendation and fine-grained doctor recommendation.The main research contents are as follows:(1)Constructed a medical vertical information knowledge base covering chief complaint information database,disease knowledge base and doctor information resource pool.Specifically,it mainly uses asynchronous crawler technology to extract structured and semi-structured chief complaint data from the four medical professional websites with the highest Baidu index,and builds a complaint information database covering 48 secondary departments.Extract disease symptom knowledge to construct a disease knowledge base,and design a symptom weight calculation method based on Bayesian algorithm.Build a dedicated doctor information resource pool for guidance recommendations to provide data support for doctor recommendation.(2)It is proposed to convert the department recommendation problem into the chief complaint text multi-category problem,and to implement the coarse-grained recommendation of the entire doctor recommendation process.Firstly,the chief complaint text expression model is designed,and the chief complaint text is converted into a vector representation.Then,in order to realize the accurate recommendation of the department according to the patient chief complaint,this paper focus on the multi-classification model of the chief complaint text based on the TextCNN algorithm.(3)A personalized doctor recommendation method is proposed to achieve fine-grained doctor recommendation.In order to achieve the purpose of the patient looking for the doctor and the doctor looking at the patient,this paper proposes a complaint text entity recognition model based on BiLSTM-CRF algorithm,and further proposes a two-way maximum matching algorithm based on the disease knowledge base to optimize the recognition results.At the same time,in order to dig deeply into the patient’s disease condition,the patient’s intention identification method is studied,which effectively ensured the accurate matching of the patient’s disease and the doctor’s expertise.Finally,in order to achieve personalized doctor recommendation for patients,this paper proposes a personalized doctor recommendation algorithm based on collaborative filtering,which consist of patient social information similarity,complaint text similarity,doctor position factor and doctor score.(4)The personalized doctor recommendation system is designed and implemented.Meanwhile,we developed the interface based on the Flask framework,which is used to connect the intelligent guiding robot,and verifies the feasibility of the personalized doctor recommendation program from the perspective of practical application.