Mixed-Effects Regression Splines to Model Myopia Data

Abstract

Myopia is a disorder of ocular refraction with varying rates of progression. Although the disorder has a dynamic nature, prospective longitudinal studies with long term follow-ups have been remarkably few. In this paper, we show how mixed-effects regression splines with different choices of basis functions can be used to model myopia progression data in a flexible way. We show how the estimated model may be used to find prediction curves with corresponding confidence and tolerance intervals for a new myopic subject. We discuss alternative choices of the basis functions such as the truncated polynomial spline functions (2 types) and B-spline functions. Principal component functions may be used for an analysis of the variation of the curves in the population. The theory is collected together and presented in a coherent way as well as illustrated with a careful analysis of myopia progression data from a Finnish myopia study.