Cardiovascular disease develops through gradual accumulation of risk factors and progressive vascular damage. Longitudinal studies are well suited to determine when and how these changes occur, but they introduce several analytic challenges, including repeated measurements on the same individuals, irregular or sparse follow-up schedules, missing data, and non-linear trajectories. We conducted a narrative, application-focused review categorizing methods into five major classes: traditional or marginal models, mixed-effect models, joint models, trajectory and mixture models, and functional or machine-learning approaches. For each class, we provide intuitive descriptions, typical cardiovascular applications, and a balanced discussion of assumptions, strengths, limitations, and recommended sensitivity analyses. We emphasize practical guidance for method selection, model validation, and transparent reporting. In summary, no single method addresses every research goal. The analytic strategy should fit both the clinical question and data characteristics, with clear definition of objectives, careful assessment of assumptions, appropriate handling of missing data, and validation on independent samples whenever possible. Future methodological development should focus on making hybrid models more accessible, improving integration of sparse and dense data sources, and advancing reporting standards for longitudinal cardiovascular research.
Chen et al. (Thu,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: