Study On The Prediction System Of Diabetic Complications Based On Neural Network

Many of China's large and medium-sized hospitals in the construction of medical information platform has made rapid development,but most of the hospital medical information system is just a management tool for supporting daily operations of the city hospital,the doctor failed to provide more effective treatment services.But,at present the doctor in clinical diagnosis and treatment process mainly depends on their professional knowledge and experience accumulation,along with the rapid development of the diagnosis and treatment of medical knowledge and related diseases are becoming more and more complex,the doctor needs professional knowledge lags far behind the pace of medical knowledge growth,coupled with the limitations of medical science itself and the complexity of the doctor's cognitive ability the resulting current doctor faces enormous challenges in the process of clinical practice.The development of medical information urgently needs to turn from information service to intelligent service.How to make use of the massive data of disease diagnosis and treatment in hospital management system and find hidden rules and knowledge to assist doctors in diagnosis and treatment of diseases will be a significant research direction.Under the above background,this article selects the original data from diabetes related data in a hospital management system as the research goal.Diabetes is a common chronic disease that endangers people's healthy life.The complication caused by diabetes is a killer that endangers people's health and even causes death.The complications of diabetes have two characteristics,one is that the early complications are not easy to find,and the two is that once the complications are produced,the drug treatment is difficult to reverse.In view of the above problems,it is determined that the research question in this paper is to predict diabetic complications by machine learning method.The object of study is from the original data set of diabetes related to a hospital management system.Specifically,the following contents are included:(1)data preprocessing.Because the data used in the study are real medical data obtained from the hospital,in order to ensure the following analysis can be established in a well structured data,this paper has carried on the data integration,feature selection and discretization,feature reduction,error handling and other pre-processing operation data.And a preliminary statistical analysis of the processed sample data sets is carried out to further understand the characteristics of the data.(2)the construction of the prediction model of diabetic complications.According to the research problems in this investigation found that the use of the existing research in the algorithm is mainly two kinds,BP neural network and Bayesian network,based on the analysis of the characteristics of the data of diabetes complications,combined with the advantages and disadvantages of the algorithm,the BP neural network algorithm as the basic model,and according to the data of irregular diabetic complications the characteristics of the BP neural network was improved,which has a good effect in the treatment of this problem,on the basis of established diabetes complications prediction model,and use the sample data obtained above the set of comparative experiments,experiments show that the prediction of diabetes complications prediction model has better accuracy the rate of.(3)system implementation.This paper will systematize the content and design and implement the prediction system of diabetic complications.Its goal is to provide predictive functions for assisting diabetes complications,assist medical staff to make decisions,and improve the accuracy and efficiency of diagnosis and treatment.