Functional Principal Component Analysis as an Alternative to Mixed-Effect Models for Describing Sparse Repeated Measures in Presence of Missing Data

被引:0
|
作者
Segalas, Corentin [1 ]
Helmer, Catherine [2 ]
Genuer, Robin [1 ]
Proust-Lima, Cecile [2 ]
机构
[1] Univ Bordeaux, INSERM, BPH, U1219,INRIA, Bordeaux, France
[2] Univ Bordeaux, INSERM, BPH, U1219, Bordeaux, France
关键词
functional principal component analysis; missing data; mixed models; sparse functional data; LONGITUDINAL DATA; DROP-OUT; DEATH;
D O I
10.1002/sim.10214
中图分类号
Q [生物科学];
学科分类号
07 ; 0710 ; 09 ;
摘要
Analyzing longitudinal data in health studies is challenging due to sparse and error-prone measurements, strong within-individual correlation, missing data and various trajectory shapes. While mixed-effect models (MM) effectively address these challenges, they remain parametric models and may incur computational costs. In contrast, functional principal component analysis (FPCA) is a non-parametric approach developed for regular and dense functional data that flexibly describes temporal trajectories at a potentially lower computational cost. This article presents an empirical simulation study evaluating the behavior of FPCA with sparse and error-prone repeated measures and its robustness under different missing data schemes in comparison with MM. The results show that FPCA is well-suited in the presence of missing at random data caused by dropout, except in scenarios involving most frequent and systematic dropout. Like MM, FPCA fails under missing not at random mechanism. The FPCA was applied to describe the trajectories of four cognitive functions before clinical dementia and contrast them with those of matched controls in a case-control study nested in a population-based aging cohort. The average cognitive declines of future dementia cases showed a sudden divergence from those of their matched controls with a sharp acceleration 5 to 2.5 years prior to diagnosis.
引用
收藏
页数:14
相关论文
共 24 条
  • [1] Principal component models for sparse functional data
    James, GM
    Hastie, TJ
    Sugar, CA
    [J]. BIOMETRIKA, 2000, 87 (03) : 587 - 602
  • [2] The effect of missing data on sample sizes for repeated measures models
    Johnson, M
    Davis, P
    [J]. PROCEEDINGS OF THE TWENTY-THIRD ANNUAL SAS USERS GROUP INTERNATIONAL CONFERENCE, 1998, : 1247 - 1251
  • [3] Sparse logistic functional principal component analysis for binary data
    Rou Zhong
    Shishi Liu
    Haocheng Li
    Jingxiao Zhang
    [J]. Statistics and Computing, 2023, 33
  • [4] Sparse logistic functional principal component analysis for binary data
    Zhong, Rou
    Liu, Shishi
    Li, Haocheng
    Zhang, Jingxiao
    [J]. STATISTICS AND COMPUTING, 2023, 33 (01)
  • [5] Linguistic pitch analysis using functional principal component mixed effect models
    Aston, John A. D.
    Chiou, Jeng-Min
    Evans, Jonathan P.
    [J]. JOURNAL OF THE ROYAL STATISTICAL SOCIETY SERIES C-APPLIED STATISTICS, 2010, 59 : 297 - 317
  • [6] Functional principal component models for sparse and irregularly spaced data by Bayesian inference
    Ye, Jun
    [J]. JOURNAL OF APPLIED STATISTICS, 2024, 51 (07) : 1287 - 1317
  • [7] Hierarchical sparse functional principal component analysis for multistage multivariate profile data
    Wang, Kai
    Tsung, Fugee
    [J]. IISE TRANSACTIONS, 2021, 53 (01) : 58 - 73
  • [8] ANALYSIS OF SEMIPARAMETRIC REGRESSION-MODELS FOR REPEATED OUTCOMES IN THE PRESENCE OF MISSING DATA
    ROBINS, JM
    ROTNITZKY, A
    ZHAO, LP
    [J]. JOURNAL OF THE AMERICAN STATISTICAL ASSOCIATION, 1995, 90 (429) : 106 - 121
  • [9] The use of MIXED models in the analysis of animal experiments with repeated measures data
    Wang, Z
    Goonewardene, LA
    [J]. CANADIAN JOURNAL OF ANIMAL SCIENCE, 2004, 84 (01) : 1 - 11
  • [10] Commentary: Linear mixed-effect models in longitudinal data analysis: Shaken not stirred
    Bruno, Vito Domenico
    Benedetto, Umberto
    [J]. JOURNAL OF THORACIC AND CARDIOVASCULAR SURGERY, 2019, 158 (02): : 341 - 342