Prediction of Type II Diabetes Risk Based on XGBoost and 1D-CNN

In this paper, machine learning method is used to accurately predict the blood glucose level of the real physical examination data in a Grade-A Tertiary Hospital in China. The original data is preprocessed through a series of methods, and t hen some irrelevant features are removed by establishing the correlation between the feature value and the target value. Then input the feature data into XGBoost for feature construction, then build a risk prediction model for risk classification, and finally perform high and low risk regression analysis. After experiments, a more accurate prediction model of type II diabetes risk was obtained.