3种成分数据方法在24h活动行为时间使用流行病学研究中的应用比较  

Comparison of Three Compositional Data Methods Applied to the 24h Movement Time-Use Epidemiological Research

作　　者：刘晴 谭健怡 黄宝莹 任志强 程金群 黄珍惠 徐华富 李红娟[3] 郜艳晖 LIU Qing;TAN Jianyi;HUANG Baoying;REN Zhiqiang;CHENG Jinqun;HUANG Zhenhui;XU Huafu;LI Hongjuan;GAO Yanhui(Guangdong Provincial Institute of Sports Science,Guangzhou 510663,China;Guangdong Pharmaceutical University,Guangzhou 510310,China;Beijing Sport University,Beijing 100084,China;Jinan University,Guangzhou 510632,China)

机构地区：[1]广东省体育科学研究所,广东广州510663 [2]广东药科大学,广东广州510310 [3]北京体育大学,北京100084 [4]暨南大学,广东广州510632

出　　处：《中国体育科技》2022年第10期47-55,共9页China Sport Science and Technology

基　　金：国家社会科学基金项目(ALA190015)。

摘　　要：通过文献资料调研和实例分析比较等距对数比转换(isometric logratio transformation,ilr)、Dirichlet分布和Flexible Dirichlet(FD)3种成分数据方法优缺点,旨在选择最优方法以阐明人群24 h活动行为时间使用比例分布。研究发现,相比于Dirichlet和FD方法,ilr更能准确描述研究人群24 h活动行为时间使用比例分布。未来的研究仍需继续探索成分数据方法,促进时间使用流行病方法学研究及应用。The study compared the advantages and disadvantages of three compositional data methods, including the isometric logratio transformation(ilr), Dirichlet distribution, Flexible Dirichlet(FD) and conducted case analysis. The aims were to choose the best method to clarify the distribution of 24 h movement time-use ratio in population. The study found that compared with Dirichlet and FD methods, ilr could more accurately describe 24 h movement time-use ratio distribution of population in this study. Future research explored compositional data methods was still necessary for promoting the research and application of time-use epidemiological methodology.

关 键 词：身体活动 时间使用流行病学 成分数据 等距对数比转换 Dirichlet分布 柔性狄利克雷 

分 类 号：G804.49[文化科学—运动人体科学]