The research of BP Neural Network based on One-Hot Encoding and Principle Component Analysis in determining the therapeutic effect of diabetes mellitus

Abstract

The BP neural network is a hybrid algorithm based on One-Hot Encoding and Principle Component Analysis (PCA). In order to make the distance calculation between variables more reasonable, the hybrid algorithm first reduces the dimension of input variables by means of PCA, and then processes the principal component variables with One-Hot Encoding. Thereafter, the BP neural network is established with all the specially processed data. The evaluation system of the diabetes therapeutic effect is established on this hybrid algorithm and the system can reduce the burden of doctors and enhance the treatment efficiency. Compared with the evaluation system established by directly using BP neural network, the BP neural network evaluation system based on One-Hot Encoding and PCA is faster and more accurate in the evaluation of the therapeutic programs of diabetes. It also provides a technological base for the evaluation of therapeutic programs of other diseases.