潜在类别混合模型及其在纵向数据轨迹分析中的应用

Latent Class Mixed Model and its Application to Longitudinal Data Trajectory Analysis

目的介绍潜在类别混合模型及其在纵向数据轨迹分析中的应用。方法以一项限制能量摄入的随机对照临床试验为例,应用潜在类别混合模型进行轨迹分析,结合贝叶斯信息准则、平均后验概率及高后验概率个体所占比例判断最佳轨迹数目及形状。结果四组三次模型最优,人群分为四类减重模式:高体重快速减重组、低体重快速减重组、高体重缓慢减重组及对照组。结论潜在类别混合模型既能识别群体中的异质性,又能考虑到类别内个体发展轨迹,有望广泛应用于纵向数据的处理分析。

Objective To introduce the latent class mixed model and its application to longitudinal data trajectory analysis.Methods A randomized controlled clinical trial that restricted calorie was used as an example to determine the number and shape of optimal trajectories based on bayesian information criterion,mean posterior probability and proportion of individuals with high posteriori probability.Results Four different weight loss trajectory groups were identified:rapid weight loss and recombination of high weight,rapid weight loss and recombination of low weight,slow weight loss,and control group.Conclusion The latent class mixed model can not only identify the heterogeneity of the population,but also take into account the developmental trajectory of the individuals in the latent class,which is expected to be widely used in longitudinal data analysis.