Regression based principal component analysis for sparse functional data with applications to screening growth paths

TL;DR

This paper introduces a novel regression-based principal component analysis method for sparse functional data, enabling joint assessment of growth paths over time, with applications to pediatric growth monitoring.

Contribution

It develops a new estimation algorithm for principal component analysis of sparse growth path data, which is robust, flexible, and incorporates covariates, advancing pediatric growth assessment methods.

Findings

The method performs well in simulations compared to existing approaches.

Application to Finnish teen growth data reveals new insights into puberty growth patterns.

The algorithm is computationally robust and does not rely on strong distributional assumptions.

Abstract

Growth charts are widely used in pediatric care for assessing childhood body size measurements (e.g., height or weight). The existing growth charts screen one body size at a single given age. However, when a child has multiple measures over time and exhibits a growth path, how to assess those measures jointly in a rigorous and quantitative way remains largely undeveloped in the literature. In this paper, we develop a new method to construct growth charts for growth paths. A new estimation algorithm using alternating regressions is developed to obtain principal component representations of growth paths (sparse functional data). The new algorithm does not rely on strong distribution assumptions and is computationally robust and easily incorporates subject level covariates, such as parental information. Simulation studies are conducted to investigate the performance of our proposed method, including comparisons to existing methods. When the proposed method is applied to monitor the puberty growth among a group of Finnish teens, it yields interesting insights.