Local and global temporal correlations for longitudinal data (Q1661323)

This paper focuses on the quantification of the association between a pair of random functions recorded over a time. The temporal Pearson correlation between the paired random functions \(X\) and \(Y\) is studied, and it is demonstrated that it is easy to estimate the cross-covariance between \(X\) and \(Y\) at the same time \(t\) as well as their respective variances. This facilitates the estimation of the Pearson correlation \(\rho(t)\) of \(X(t)\) and \(Y(t)\) at time \(t. \) Since a global summary of the correlations is also of interest, a global measure is proposed by averaging the temporal correlations \(\rho(t)\) over a time interval. The proposed method is applicable to various types of functional data. The efficiency of the proposed method is demonstrated on the treatment effect using the retrospective primary biliary cirrhosis medical data.