Research On The Method Of Medical Quality Evaluation And Precision Medical Recommendation

Medical safety and medical quality have been raised attention at home and abroad.The commonly used medical quality evaluation process requires manual intervention and is overly dependent on the experience of experts.At the same time,the analysis and evaluation system is based on random samples rather than aggregate data.Therefore,one of the defects of medical quality evaluation is that lack of objectivity and scientificity.At present,hospitals have not adopted big data analysis and other technical means to compare doctors’ medical methods objectively,so it is impossible to judge their merits and demerits.As a result,the treatment methods of good doctors cannot be promoted,which leads to the omission of hospital quality management and the reduction of management efficiency and quality.At the same time,in the process of medical treatment,the way patients choose departments or doctors is lack of objective description,comparison of diagnosis,treatment characteristics of departments and doctors’ professional expertise,which leads to hospitals can not provide timely and high-quality medical services for patients.Therefore,in order to deal with the challenges above,this thesis mainly does the following work.1.Based on big data analysis technology,this paper proposes a medical quality evaluation method ground on big data of full sample cases in hospitals.Through the analysis,the treatment time,treatment cost and treatment effect were selected as the indicators of medical quality evaluation,and the medical quality was scored through K-means cluster analysis,in order to give an objective comparison of medical quality among different doctors and departments.By quantifying the professional and technical level and medical service quality of doctors and departments,the medical quality management level of hospitals can be comprehensively improved.At the same time,the evaluation results can be used as the basis for decision making.2.The medical recommendation system generally adopts collaborative filtering technology,which collects patients’ historical records,personal preferences and other information,calculates the similarity with other patients,and uses the evaluation of similar patients to recommend doctors and departments.In this thesis,based on deep learning technology,a deep neural network recommendation method established on precision medical treatment recommendation is proposed and implemented by constructing a neural network model to obtain the feature eigenvector of patients and the neural network model to obtain the feature eigenvector of recommended objects.3.In order to timely obtains the case types that doctors and departments are good at,as well as objectively reflects the dynamic changes of the case types that doctors and departments are good at,this thesis proposes to use cases patients with features as environment,to recommend a doctor or a department as action,to treat the quality score from a doctor or a department as a reward feedback.Constructing state representation model,policy model and reward prediction model respectively.On the basis of these three models,a trained medical recommendation model will be built to implements a precise doctor recommendation method which based on a deep reinforcement learning.