Diagnostic model generated by MRI-derived brain features in toddlers with autism spectrum disorder.
Toddler brain thickness feeds a simple model that tells autism from delay, giving a science-backed push for earlier ABA starts.
01Research in Context
What this study did
The team built a computer model that reads toddler MRI scans. They wanted to see if brain thickness could tell autism from developmental delay.
They used a method called random forest. It looks at many spots on the cortex and learns which pattern fits each group.
What they found
Cortical thickness beat volume and surface area. The model sorted toddlers with autism from those with delay better than chance.
This means a quick brain scan could give a clear yes-or-no signal before age three.
How this fits with other research
Brugnaro et al. (2024) pushed the idea further. They swapped toddler MRI for adult fMRI and random forest for a deep-learning model called STDCformer. Both papers show machines can flag autism from brain data, but the new work needs older brains and resting-state scans.
Estes et al. (2011) looked at the same age group years earlier. They tied smaller basal-ganglia volume to more repetitive play. Xiang et al. now show thickness, not volume, gives cleaner labels. The older study asked why kids act different; the new one asks how to spot autism faster.
Shokouhi et al. (2012) and Chien et al. (2015) also hunted thickness clues, but in teens and adults. Their links to social quirks line up with Xiang’s toddler work, showing thickness is a steady ruler across ages.
Why it matters
You can’t order an MRI during a clinic visit, yet this proof shows biology can back your behavioral diagnosis. Share the paper with pediatricians who still wait-and-see. Push for earlier screening slots. The sooner a toddler is flagged, the sooner you start ABA, and the more brain plasticity you have to shape learning.
Want CEUs on This Topic?
The ABA Clubhouse has 60+ free CEUs — live every Wednesday. Ethics, supervision & clinical topics.
Join Free →Show the paper to your referral source and ask, ‘Can we flag these kids before age three?’
02At a glance
03Original abstract
Autism spectrum disorder (ASD) is a complex neurodevelopmental disorder mainly showed atypical social interaction, communication, and restricted, repetitive patterns of behavior, interests and activities. Now clinic diagnosis of ASD is mostly based on psychological evaluation, clinical observation and medical history. All these behavioral indexes could not avoid defects such as subjectivity and reporter-dependency. Therefore researchers devoted themselves to seek relatively stable biomarkers of ASD as supplementary diagnostic evidence. The goal of present study is to generate relatively stable predictive model based on anatomical brain features by using machine learning technique. Forty-six ASD children and thirty-nine development delay children aged from 18 to 37 months were evolved in. As a result, the predictive model generated by regional average cortical thickness of regions with top 20 highest importance of random forest classifier showed best diagnostic performance. And random forest was proved to be the optimal approach for neuroimaging data mining in small size set and thickness-based classification outperformed volume-based classification and surface area-based classification in ASD. The brain regions selected by the models might attract attention and the idea of considering biomarkers as a supplementary evidence of ASD diagnosis worth exploring. Autism Res 2017, 0: 000-000. © 2016 International Society for Autism Research, Wiley Periodicals, Inc. Autism Res 2017, 10: 620-630. © 2016 International Society for Autism Research, Wiley Periodicals, Inc.
Autism research : official journal of the International Society for Autism Research, 2017 · doi:10.1002/aur.1711