A new covariate selection procedure for the asynchronous functional concurrent model
A concurrent functional model is a special case where both the response Y and the covariates X depend on the same argument t and are linked pointwise. A major challenge arises in the asynchronous setting, where Y(t) and X(t) are observed at different time points for each individual, causing many standard methods to break down. Existing approaches often assume a specific structure (e.g., linear or additive), which can be restrictive and difficult to justify, especially with multiple covariates.
We propose a covariate selection method for this setting that reduces dimensionality by identifying and removing irrelevant variables without imposing structural assumptions. Building on conditional distance correlation (Wang et al., 2015), it leads to a specification test for a general model formulation.
Wang, X., Pan, W., Hu, W., Tian, Y., and Zhang, H. (2015). Conditional distance correlation. Journal of the American Statistical Association, 110(512):1726–1734.
Palabras clave: concurrent model conditional distance covariance covariates selection functional regression