TY - JOUR
T1 - Spectral analysis of light-adapted Electroretinograms in neurodevelopmental disorders:
T2 - Classification with Machine Learning
AU - Constable, Paul A
AU - Pinzon-Arenas, Javier O
AU - Roberto Mercado Diaz, Luis
AU - Lee, Irene O
AU - Marmolejo-Ramos, Fernando
AU - Loh, Lynne
AU - Zhdanov, Aleksei
AU - Kulyabin, Mikhail
AU - Brabec, Marek
AU - Skuse, David H
AU - Thompson, Dorothy A
AU - Posada-Quintero, Hugo
PY - 2025/1
Y1 - 2025/1
N2 - Electroretinograms (ERGs) show differences between typically developing populations and those with a diagnosis of autism spectrum disorder (ASD) or attention deficit/hyperactivity disorder (ADHD). In a series of ERGs collected in ASD (n = 77), ADHD (n = 43), ASD + ADHD (n = 21), and control (n = 137) groups, this analysis explores the use of machine learning and feature selection techniques to improve the classification between these clinically defined groups. Standard time domain and signal analysis features were evaluated in different machine learning models. For ASD classification, a balanced accuracy (BA) of 0.87 was achieved for male participants. For ADHD, a BA of 0.84 was achieved for female participants. When a three-group model (ASD, ADHD, and control) the BA was lower, at 0.70, and fell further to 0.53 when all groups were included (ASD, ADHD, ASD + ADHD, and control). The findings support a role for the ERG in establishing a broad two-group classification of ASD or ADHD, but the model’s performance depends upon sex and is limited when multiple classes are included in machine learning modeling.
AB - Electroretinograms (ERGs) show differences between typically developing populations and those with a diagnosis of autism spectrum disorder (ASD) or attention deficit/hyperactivity disorder (ADHD). In a series of ERGs collected in ASD (n = 77), ADHD (n = 43), ASD + ADHD (n = 21), and control (n = 137) groups, this analysis explores the use of machine learning and feature selection techniques to improve the classification between these clinically defined groups. Standard time domain and signal analysis features were evaluated in different machine learning models. For ASD classification, a balanced accuracy (BA) of 0.87 was achieved for male participants. For ADHD, a BA of 0.84 was achieved for female participants. When a three-group model (ASD, ADHD, and control) the BA was lower, at 0.70, and fell further to 0.53 when all groups were included (ASD, ADHD, ASD + ADHD, and control). The findings support a role for the ERG in establishing a broad two-group classification of ASD or ADHD, but the model’s performance depends upon sex and is limited when multiple classes are included in machine learning modeling.
KW - biomarker
KW - retina
KW - autism
KW - attention deficit hyperactivity disorder
KW - sex
KW - medication
KW - feature selection
UR - http://www.scopus.com/inward/record.url?scp=85216270123&partnerID=8YFLogxK
U2 - 10.3390/bioengineering12010015
DO - 10.3390/bioengineering12010015
M3 - Article
AN - SCOPUS:85216270123
SN - 2306-5354
VL - 12
JO - Bioengineering
JF - Bioengineering
IS - 1
M1 - 15
ER -