Research On Diversity-enhanced Physician Recommendation In Online Medical

In recent years,with the aggravation of the problem of “high expense and difficulties in medical care”,the Chinese government spares no affects to promote the healthcare reform.With the development of Internet,the traditional medical industry starts to cooperate with Internet to promote the healthcare reform.Patients can choose physicians with online appointment on the medical websites.Under these circumstances,these online medical websites recommend hundreds of physicians,And they usually provide little information about the competency of the physicians recommended.So,it is difficulty to patients evaluate physician competency correctly.In this case,patients prefer to choose the public hospitals to guarantee the quality of treatment.But considering the uneven distribution of high quality medical resources,this preference contributes to the overload of large public hospitals and the idling of community hospitals.Therefore,this thesis proposes a novel diversity-based physician recommendation for triage system.This model solves two problems in physician recommendation.Firstly,the number of alternative physicians and hospitals increase rapidly,but the data which can evaluate the competency of physicians is still very scare.To overcome this challenge,an improved matrix factorization with the consideration of implicit similarity relationship is proposed to evaluate the physician competency precisely.The experimental results demonstrated that the improved matrix factorization can predict the missing values more accuracy and stable than the traditional collaborative filtering methods.Secondly,to balance the patients’ preference and hospital staff,this thesis constructs a utility-based function based on the patients’ preference and a diversity-based function focusing on the triage.And then two heuristic algorithms,a swap algorithm and a greedy algorithm,are proposed to coordinate the demand of utility and diversity,which will increase the kinds of alternative hospitals and difference between physicians in the recommendation rank.The experiments show that these two heuristic algorithms can improve the diversity of the physician recommendation system than both the traditional heuristic algorithms and state of art.