Assessment of the Autism Spectrum Disorder Based on Machine Learning and Social Visual Attention: A Systematic Review.
Maria Eleonora Minissi, Irene Alice Chicchi Giglioli, Fabrizia Mantovani and 1 others
PMID 34101081WHAT IT FOUND
Eye-tracking combined with machine learning correctly separated autistic children from typical peers in 11 studies, with accuracies ranging from 60% to 98%.
These tools are not yet validated for clinical use, so they cannot replace current diagnostic assessments.
Key findings
01Machine learning models using eye-tracking data achieved classification accuracies between 60% and 98% in distinguishing children with ASD from typically developing peers.
02Despite promising research results, none of the implicit measures or biomarkers used in these studies have been validated for clinical use.
03The studies included in this review generally had small sample sizes, which affected the reliability of the discrimination accuracy in the machine learning models.
STILL TO COME
How it was doneWhat they foundWhat it means for SLPs
Read the rest of this summary
You get three full summaries a month, free, and we do not ask for a card. Search, the TL;DRs and your library stay unlimited either way.
What it does not show
The included studies had small sample sizes, which reduces the reliability and generalizability of the reported accuracies. None of the eye-tracking or machine learning methods reviewed are currently validated for clinical use. There was significant heterogeneity in the machine learning models, stimuli types, and data processing methods across the 11 studies, making direct comparison difficult. The review focused on children aged 2-10 years, so findings may not apply to adolescents or adults.
Declared interests
The work was funded by the Ministerio de Economía, Industria y Competitividad, Gobierno de España. No other conflicts of interest are reported in the provided text.
The easy way to misread this
Do not assume these high accuracy rates (up to 98%) mean eye-tracking can diagnose ASD in your clinic today. The paper states these biomarkers are not validated for clinical use, and the high accuracies came from small, controlled research studies that may not reflect real-world diagnostic complexity.