关键词: font-size:medium, ">2型糖尿病；BP神经网络；MCMC；多元回归；相关性分析

Abstract: Objective： With TCM syndrome elements as the core， combined with the distribution of TCM syndrome types of type 2 diabetes， the relationship between TCM syndrome types and western medicine indexes and the relationship between TCM syndrome elements and western medicine indexes were sought by filling in the missing data and combining correlation analysis. Methods：A classification prediction model based on BP neural network is established. Taking TCM syndrome elements as the input and syndrome types as the output of the neural network， a BP neural network model with 16 inputs and single output was established. The missing data of TCM syndrome types obtained in the survey were few， so the direct elimination method was adopted， and the training error was 0.011 57 through the training of statistical data， indicating that the model could show the corresponding relationship between TCM syndrome elements and syndrome types. Secondly， due to the high degree of missing of other data， a multi-filling model based on MCMC was established to fill the data of height， weight， HbAIC， fasting insulin， fasting blood glucose and fasting C-peptide. Then， a multiple regression model was established to solve the relevant data analysis of the relationship between the indicators of traditional Chinese medicine and western medicine. Results and Conclusion： The correlation analysis shows that there is a relationship between type 2 diabetes western medicine indicators and TCM syndrome elements， and also between type 2 diabetes western medicine indicators and part of TCM syndrome types.

Key words: font-size:medium, ">Type 2 diabetes； BP neural network； MCMC；Multiple regression； Correlation analysis