Complications Of Type 2 Diabetes Based On Deep Learning Research On Forecasting System

The use of deep learning algorithms to diagnose diseases has become a research hotspot in the field of artificial intelligence.Diabetes,as one of the three major diseases endangering human health,is caused by the increase of blood glucose levels.If left uncontrolled for a long time,it will cause damage to different organs of the body and cause life-threatening complications.The complications of diabe tes have the characteristics of high morbidity,high treatment cost,high mortality,low cure rate,and people are not easy to detect.There is no cure for complications in medicine.Therefore,diabetes complications must be prevented at the source.The me dical information system of most hospitals is only a management tool.How to mine the massive medical data and assist doctors in diagnosis has become a trend.Under the above background,a prediction system of type Ⅱ diabetes complications is constructed in combination with deep learning network.It can be divided into the following three aspects.(1)Preprocessing of the data set.A total of2,000 electronic medical records and paper medical records of complications of type Ⅱ diabetes in a tertiary hospital in the past three years have been sorted and collected.And a series of processing such as feature selection,integration,reduction,missing value processing and discretization are performed on the data set.Analyzed with SPSS4.0 statistical software,with P≤0.05 as the statistical significance,OR>1.0 as the risk factor,16 indicators were sel ected as the input of the model,and the nephropathy,ophthalmopathy and neuropathy of type Ⅱ diabetes were used as the output of the model.(2)Model construction and evaluation.Taking the Deep Belief Network(DBN)as the basic algorithm for constructing the model,in order to solve the problem that the number of neurons in the hidden layer of the DBN model is difficult to determine,a DBN network optimized by the ion group algorithm is proposed,and it is combined with the support vector machine algorithm(SVM),SVM model optimized by grid search for comparison.The classification accuracy,receiver curve and confusion matrix are used as the evaluation criteria of the model.Experiments show that support vector machines have certain advantages for classification problems.Its test accuracy rate is 81.8%,while the accuracy rate of Deep Belief Network(DBN)is second only to SVM at 81.6%;the use of grid search to further improve the parameters of SVM The accuracy rate of optimization is as high as 83%;the ion swarm algorithm further optimizes the number of neurons in the hidden layer of DBN and the learning rate,and its accuracy rate reaches the highest91%.Obviously,the effect of the particle swarm optimization DBN network in the concurrent prediction of type Ⅱ diabetes is significantly better than the other two algorithms.(3)Build a web-based forecasting system.Combine Labview development platform with MATLAB software.To build a log-in system for predicting complications of type 2 diabetes,input the relevant medical data of a type 2 diabetes patient,and the predictive model can be used to determine which type of complications the patient is in.And ass ist the doctor to make the corresponding diagnosis,and propose the corresponding treatment plan and the evaluation of the patient’s condition.In general,the establishment of a web version of the login system can effectively and accurately prevent the occurrence of diabetes complications.On the one hand: For patients,not only can avoid the damage to the body caused by repeated examinations,but also save time and money costs.On the other hand: it brings convenience to the doctor’s diagnosis and can prevent the doctor from misjudgment and missed judgment.So as to achieve the purpose of early detection and early treatment.