Online Patient Counseling Data Based Online Doctor Recommend System Research

Nowadays,the Internet platform meeting with the demand of real time online medical consultation and light interrogation becomes a new direction of medical industry development in future.However,in some medical consultation and guidance system,patient counseling is still in a passive mode.Patients find doctors by hospital or department and wait for doctors' reply.At the same time,online doctors have to go through all or a little screening of the disease information,to find their own appropriate patients' questions.In view of the above problems,this article researched on the doctor recommend system for online patient consultation,which recommended doctor through consulting text mining and improved the efficiency,accuracy and effectiveness of online medical consultation service.The main research contents include:(1)Recommended frame design.The solution of doctor recommendation problem could be divided into two phases.Firstly,the recommendation problem was converted to three layer classification problem.In this phase,construct tree structure diagram of department on the web page first of all,then divide patient into the first layer of department using the classification algorithm,on this basis,divide patient into the second layer of department and then into the third layer of illness.In the second phase which called collaborative filtering recommendation based on clustering,find out the similar resolved cases of patient condition under department by clustering,then recommend those doctors to solve the condition and recommend patient to doctor.(2)According to the characteristics of short consulting text,this paper built a text expression model based on word vector and a vector space model based on TF-IDF,solved the problem of high dimension and sparse representation of consulting texts effectively.(3)Designed a text multi-classification algorithm based on the maximum entropy algorithm and a doctor collaborative filtering recommendation algorithm based on the fast density peak clustering algorithm.(4)Taking into account of the timeliness of the recommendation,this paper considered doctors' activeness when designing the recommendation algorithm,ranked doctors by giving priority to active ones and designed activeness computational method.Finally,this article realized a doctor recommend system based on the consulting text of Good Doctor.In order to verify the validity of algorithms already designed,the paper compared the effectiveness and computational efficiency of algorithm through experiments.