Early Diabetes Risk Prediction Based on Deep Learning Methods

Diabetes is a relatively common chronic disease, and there is a long asymptomatic stage. With the continuous development of deep learning methods, it has been widely used in various fields, and its application to the medical field has gradually become a hot spot. To prevent and reduce the incidence of diabetes, and for effective diagnosis and prediction of diabetes, a diabetes prediction model based on one dimensional convolutional neural network (lDCNN) is proposed. In order to verify the effectiveness of our proposed method, we compared it with the naive Bayes classifier and random forest. We analyzed the three methods according to several performance indicators of the confusion matrix. The experimental results verify that our proposed method has more advantages in terms of accuracy and precision, achieving the accuracy of 97.02%, which can effectively predict the diabetes. The proposed 1DCNN-based method has high clinical application value and promotion significance, providing ideas for the mining and analysis of other disease data in the medical industry today.