Research And Application Of Personalized Recommendation System Based On Tensor In Medical Field

With the development of artificial intelligence and machine learning,every field has gradually stepped into intelligence.As a medical and health service field which has been paid attention by the state,its intelligence has been paid more and more attention in recent years.Today's recommendation system has been applied successfully in various fields,but the application in the medical field still needs to be further explored.Although the recommendation system in the medical field has more difficulties than other fields,even so scholars at home and abroad have never stopped exploring.At present,the more successful business applications of the recommendation system in the medical industry are: auxiliary diagnosis and treatment,rational drug use,medical guidance,precision medical treatment,etc.the core technologies used are mostly traditional recommendation algorithms,among which the more successful are collaborative filtering and hybrid recommendation.In recent years,tensor decomposition has also been applied to the recommendation system,even better than the traditional recommendation algorithm to some extent.Therefore,this paper chooses drug relocation as the research and application direction of this recommendation system in the medical field,and uses tensor decomposition technology to build a drug relocation recommendation algorithm based on tensor.Firstly,this paper makes a comprehensive survey of medical big data and related business application scenarios,and introduces the relationship between today's medical category and recommendation system.Tensor is derived from medical big data,recommendation system is derived from information overload,and tensor decomposition is the hub connecting the two fields.Then,the theoretical basis of tensor and the common concept and method are comprehensively integrated.From the classic traditional recommendation algorithm,the recommendation system is researched,sorted and analyzed.The application direction of this recommendation algorithm in the medical field is drug relocation,and the value of the recommendation algorithm in it is explained.The tensor model is constructed for the medical data related to drug relocation,and the normalized tensor is obtained by processing the irregular data matrix set with parafac2 method.The parallel factorization method,which is commonly used in recommendation system by tensor,is used to predict the relationship between disease and drug,mine the potential relationship between disease and drug,and finally complete the topn recommendation of drug relocation.Finally,under the open data set,select 10% and erase part of the drug disease relationship information as the test data,use the k-fold cross validation method to repeatedly test the algorithm in this paper,and compare it with the traditional collaborative filtering,which verifies that the research model has good performance in precision,recall and F1 measure,and has certain novelty and feasibility.Then,based on the original model,the confidence matrix is introduced to improve the reliability of the improved model.Finally,it shows that the algorithm in this paper provides a certain reference value for the research of drug relocation in the medical field,and better realizes the modeling and reasoning of tensor in the medical big data.