Canonical correlation analysis (CCA) is interesting and important, since it provides an overall picture of associations between risk factors and outcome variables. Ridderstolpe et al. present a study that uses CCA to examine the relationship between sets of multiple risk factors and sets of multiple clinical outcomes. The authors say that this method has some limitations, since retrospective studies using existing data may be compromised by many factors, including missing data, inexact definitions, and the relationship between cause and effect in the variables observed. They discuss the importance of having a thorough understanding of the statistical models to apply them in practice. They also discuss the reason they chose CCA as a method of data study.

This paper is a complete source of forthcoming research on the construction of clinically accepted combinations of output variables of statistically robust prediction based on relevant sets of risk factors.