Technical note: a functional data analysis approach to analyze the light-adapted electroretinogram in children and adolescents.

Abstract

Purpose: To present a new framework for the analysis of the light-adapted electroretinogram using Functional Data Analysis (FDA).

Methods: Light adapted full-field electroretinograms and extracted Oscillatory Potentials waveforms were analyzed using an FDA approach. Waveforms from 71 individuals with autism spectrum disorder and 98 Control participants (mean ± SD age in years): ASD (12.8 ± 4.4) and Control (13.5 ± 4.7) were reanalyzed from previous studies. Robust and powerful nonparametric permutation-based tests were developed for the mean, variance, and first derivative of the waveforms.

Results: FDA identified a significant group difference (p=0.049) for the mean of the waveforms between the a-wave minima and b-wave maxima for the 356 and 445 Td.s flash strengths. The Oscillatory Potentials mean and covariance for the mean and first derivative of the waveforms exhibited significant group differences across multiple flash strengths (p=.0244). The combined test statistic for the mean and covariance showed group differences by age and group, but not by sex.

Conclusion: By treating time-series recordings as functions, the proposed FDA framework enables a more informative assessment of group differences through the analysis of mean structure, covariance, and waveform dynamics captured by first derivatives. This new functional perspective extends beyond traditional time series methods based on peak time and amplitude to provide a powerful tool for uncovering subtle differences between groups.