Can a composite heart rate variability biomarker shed new insights about autism spectrum disorder in school-aged children?

TL;DR

This study explores a composite heart rate variability biomarker derived from various measures to distinguish children with autism spectrum disorder from peers, showing promising classification accuracy.

Contribution

It introduces a novel combination of linear and nonlinear HRV measures analyzed through machine learning to identify ASD in children.

Findings

HRV measures achieved 0.89 ROC AUC in identifying ASD.

The biomarker differentiated ASD from conduct problems and depression.

Preliminary results suggest potential for non-invasive ASD screening.

Abstract

High-frequency heart rate variability (HRV) has identified parasympathetic nervous system alterations in autism spectrum disorder (ASD). In a cohort of school-aged children with and without ASD, we test a set of alternative linear and nonlinear HRV measures, including phase rectified signal averaging, applied to a segment of resting ECG, for associations with ASD vs. other psychiatric conditions. Using machine learning, we identify HRV measures derived from time, frequency, and geometric signal-analytical domains that (1) identify children with ASD relative to peers with receiver operating curve area of .89, and (2) differentiate such children from those with conduct problems or depression. Despite the small cohort and lack of prospective external validation, these preliminary results warrant larger prospective validation studies.