群组多轨迹模型在纵向数据研究中的应用及实例分析  

Application and case study of group-based multi-trajectory model in longitudinal data research

作　　者：王孝焱 孙秀彬 纪伊曼[1,2] 张涛 刘云霞 Wang Xiaoyan;Sun Xiubin;Ji Yiman;Zhang Tao;Liu Yunxia(Department of Biostatistics,School of Public Health,Cheeloo College of Medicine,Shandong University,Jinan250012,China;Institute for Medical Dataology,Shandong University,Jinan250002,China)

机构地区：[1]山东大学齐鲁医学院公共卫生学院生物统计学系,济南250012 [2]山东大学健康医疗大数据研究院,济南250002

出　　处：《中华流行病学杂志》2024年第11期1590-1597,共8页Chinese Journal of Epidemiology

基　　金：国家重点研发计划(2021YFF0704101);国家自然科学基金(82222064);基于移动设备的老年人群生活方式干预研究(21320012002309)。

摘　　要：纵向队列的发展为识别和监测影响疾病病程或健康状况的多种生物标志物及行为等因素创造了条件。然而,传统统计方法通常只能利用单变量纵向数据的信息进行研究,无法充分利用多变量纵向数据信息。群组多轨迹模型(GBMTM)是近年来提出的研究多变量发展轨迹的一种方法,通过影响目标结局的多个指标来识别遵循相似轨迹的潜在人群亚组,在处理多变量纵向数据中具有独特优势。本研究阐述GBMTM的基本原理,并运用一项基于智能穿戴设备的老年人健康管理研究的数据探索多种生活相关指标与高血压的关系,展示GBMTM的具体应用,以期促进其在纵向队列研究中的应用。The development of longitudinal cohorts has made the identification and surveillance of multiple biological markers and behavioral factors which influence disease course or health status become possible.However,traditional statistical methods typically use univariate longitudinal data for research,failing to fully exploit the information from multivariate longitudinal data.The group-based multi-trajectory model(GBMTM)emerged as a method to study the developmental trajectory of multivariate data in recent years.GBMTM has distinct advantages in analyzing multivariate longitudinal data by identifying potential subgroups of populations following similar trajectories by multiple indicators that influence the outcome of interest.In this study,we introduced the application of GBMTM by explaining the fundamental principles and using the data from a health management study in the elderly by using smart wearing equipment to investigate the relationship between multiple life-related variables and hypertension to promote the wider use of GBMTM in longitudinal cohort studies.

关 键 词：纵向数据 群组多轨迹模型 发展轨迹 队列研究 

分 类 号：R195.1[医药卫生—卫生统计学]