By preprocessing and analyzing longitudinal high-dimensional data of multicollinearity in Parkinson's disease speech signals,important information can be provided for relevant medical institutions.Firstly,the data is preprocessed using Random Forest to obtain a reduced dimensional dataset.Through correlation analysis,it is found that the data has multicollinearity issues.Then,L2 regularization is introduced into the BP Neural Networks to change its objective function,solving the problem of data multicollinearity in clinical data,so as to better fit the linear mixed effects model.Finally,a comparative analysis is conducted on the AIC,BIC,and-LogLik indicators of the linear mixed effects model before and after the introduction of the BP-L2 hybrid algorithm,demonstrating the advantages of introducing this algorithm.
关键词
帕金森病/随机森林/L2正则化/BP神经网络/线性混合效应模型
Key words
Parkinson's disease/Random Forest/L2 regularization/BP Neural Networks/linear mixed effects model
万方数据