Reevaluating principal component analysis in geroscience: A call for nonlinear approaches in AI-based evaluations

Fuellen et al. (2025) highlighted the essential role of explainable AI methods, particularly principal component analysis (PCA), in evaluating interventions for aging and longevity. However, this paper raises significant concerns regarding PCA's linear and parametric nature, which can misrepresent complex, nonlinear data common in geroscience research. As biological relationships often defy simplistic interpretations, reliance on PCA may obscure vital insights, leading to potential misinterpretations of intervention effects. To enhance accuracy in analyses, this study advocates for the adoption of nonlinear and nonparametric methods, such as Spearman's rank correlation and Kendall's tau. By reconsidering their methodological approaches, researchers can foster more accurate and informed evaluations of aging-related interventions.