Construction and Application of Prediction Model of Diabetes Treatment Effect Based on Improved CART Algorithm

Abstract

In this research, the rehospitalisation of diabetic patients was taken as the standard to judge the treatment effect. With the help of CART algorithm improved on the basis of Principal Component Analysis, the researcher explored the relationship between multiple data involved in the diabetic treatment program and its treatment effect and built a prediction model of diabetic treatment effect. The study focused on the treatment data from 130 American hospitals. 9000 sets of data were used as the training set, and 1000 sets of data were used as the test set to generate the decision tree, and then the researcher pruned the generated decision tree. The accuracy of the newly established prediction model was greatly improved by 21% and the running time was largely reduced by 5.352s compared with the old model established on the unimproved CART algorithm. The results of the study well verified the feasibility and high efficiency of the new prediction model established on the improved CART algorithm, and provided a new idea for improving the treatment effect and efficiency of diabetes.