| Type 2 diabetes is a common metabolic disease,usually accompanied by decreased cognitive ability and the risk of dementia also increased significantly.At present,the Montreal Cognitive Assessment Scale(MoCA)is used to assess the cognitive function abnormal of patients,but this method is susceptible to educational level,cultural differences,examiner skills and experience.Lack of an easy to operate,quantitative assessment of the cognitive function of the method to avoid the above problems to bring interference.Functional magnetic resonance imaging(fMRI)has a higher temporal and spatial resolution compared to existing brain imaging techniques.Using fMRI can extract a large number of high dimensional image features to study the relationship of the brain and memory and attention.At present,there are few studies on the prediction of cognitive function in patients with type 2 diabetes mellitus based on fMRI.Therefore,this paper presents a new solution idea,which extracts the characteristics from fMRI and establishes the cognitive function prediction model by machine learning method.The experimental data from the Henan Provincial People’s Hospital,which includes 95 cases of fMRI images and clinical information.The main process of the experiment is to extract the features from the image,use the Elastic Net method to reduce the feature of the image features and clinical features.Support vector machine(SVM)and Logistic Regression(Logistic Regression),LR)two algorithms construct the cognitive function prediction model.In the process of model training,the model parameters are determined by ten fold cross validation.The correct rate of support vector machine classification model is 93.24%,AUC is 0.9781,and the recall rate is 92.11%.The correct rate of logistic regression model is 95.95%,AUC is 0.9971,and the recall rate is 94.74%.By comparison,it is found that the regression effect of the logistic regression model is better.Compared with the traditional method of assessing cognitive impairment in patients with type Ⅱ diabetes mellitus,the prediction model is established by extracting the feature from the image,and it has the advantage of being disturbed by external factors.At the same time,The new method of the evaluation of cognitive dysfunction is given.in addition to the experiment by the elastic network model to reduce the dimension after the selection of imaging features,can be type Ⅱ diabetes clinical diagnosis and treatment provides an important clue. |
