Online Medical Expert Recommendation Method Based On Deep Learning

With the development of online medical service platform,patients can find more medical information resources and obtain better medical treatment.However,it is difficult for patients to discover the most suitable doctors from the complex information resources.There are many imperceptible but very important links in these medical data,so it has certain research value to find the internal links in the data for disease prediction and doctor recommendation.Effective analysis and mining of Electronic Health Records(EHR)is also of great significance for patients to obtain timely and accurate treatment.Traditional medical diagnosis largely relies on the professional knowledge and practical experience of doctors,so "the good / bad treatment is likely to be caused by a good / bad doctor,and a good / bad doctor has a higher/lower evaluation by the patient(s)".As far as patients are concerned,the most suitable doctor is actually to predict the doctor's accurate performance of the patient's disease,so it is reasonable to use the patient's evaluation information to recommend the most suitable doctor for the specific disease.In recent years,as a data modeling method in the era of big data,deep learning technology has become a research hotspot,and has had a profound impact in various fields of application.This paper proposes an approach based on deep learning named Doc Inf,which is to seek the most effective doctor for a specific disease.In this paper,we build a doctor disease heterogeneous information network based on the comments data of patients on the treatment effect of doctors,and get the doctor's feature representation by embedding features,so as to explore the potential relationship between diseases and doctors.In order to obtain more subtle and abstract features of doctors and diseases,this paper uses an Auto-encoder to learn the embedded representation of the network,and constructs the latent feature representation in an unsupervised way.On this basis,we input the latent features into XGBoost algorithm to predict the doctor's experience score of special diseases.The higher the score,the better the doctor's performance.This paper evaluates the data set based on mobile medical service platform(Guahao.COM)to verify the effectiveness of the Doc Inf method.Experimental results show that,compared with other algorithms,the Doc Inf method has good performance,can effectively predict the doctor's experience score of specific diseases,and recommend more suitable doctors for patients.