基于BP-L_(2)混合算法的线性混合效应模型在帕金森病语音信号中的应用  

Application of Linear Mixed Effects Model Based on BP-L_(2)Hybrid Algorithm in Speech Signals of Parkinson's Disease

作　　者：罗成敏 王涛 LUO Chengmin;WANG Tao(School of Mathematics,Yunnan Normal University,Kunming 650500,China;Yunnan Key Laboratory of Modern Analytical Mathematics and Applications,Kunming 650500,China)

机构地区：[1]云南师范大学数学学院,云南昆明650500 [2]云南省现代分析数学及其应用重点实验室,云南昆明650500

出　　处：《现代信息科技》2024年第19期158-163,共6页Modern Information Technology

基　　金：国家自然科学基金资助项目(81360449);云南师范大学研究生科研训练基金项目(YJSJJ23-B105)。

摘　　要：通过对帕金森病语音信号多重共线性纵向高维数据的预处理和分析,可为相关医疗机构提供重要信息。首先采用随机森林对数据进行预处理,得到降维数据集,通过相关性分析得知该数据存在多重共线问题;然后在BP神经网络中引入L_(2)正则化改变其目标函数,解决临床数据经常存在的数据多重共线性,以便更好地拟合线性混合效应模型;最后对比分析引入BP-L_(2)混合算法前后线性混合效应模型的AIC、BIC和-LogLik指标,证明引入该算法的优势。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,L_(2)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-L_(2)hybrid algorithm,demonstrating the advantages of introducing this algorithm.

关 键 词：帕金森病 随机森林 L_(2)正则化 BP神经网络 线性混合效应模型 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]