Related Experiment Video
Updated: Mar 5, 2026
08:17
A Semantic Priming Event-related Potential ERP Task to Study Lexico-semantic and Visuo-semantic Processing in Autism Spectrum Disorder
Published on: April 12, 2018
11.2K
Diagnosis of autism through EEG processed by advanced computational algorithms: A pilot study
Enzo Grossi1, Chiara Olivieri1, Massimo Buscema2
1Autism Research Unit, Villa Santa Maria Institute, Italy, Via IV Novembre 22038 Tavernerio (CO).
Computer Methods and Programs in Biomedicine
|March 23, 2017
Summary
This study shows that a new Artificial Neural Network system (MS-ROM/I-FAST) can accurately detect autism spectrum disorder (ASD) using EEG data. The system achieved 100% accuracy in identifying autistic individuals, suggesting a potential for early, non-invasive diagnosis.
Area of Science:
- Neuroscience
- Artificial Intelligence
- Biomedical Engineering
Background:
- Autism Spectrum Disorder (ASD) is characterized by atypical brain organization, potentially reflected in EEG patterns.
- Artificial Neural Networks (ANNs), specifically Multi-Scale Ranked Organizing Map coupled with Implicit Function as Squashing Time (MS-ROM/I-FAST), offer advanced feature extraction from EEG.
- Previous studies demonstrated MS-ROM/I-FAST's efficacy in identifying Mild Cognitive Impairment and Alzheimer's Disease.
Purpose of the Study:
- To evaluate the effectiveness of the MS-ROM/I-FAST methodology in distinguishing individuals with ASD from typically developing individuals.
- To explore the potential of EEG-based machine learning for ASD detection.
Main Methods:
- The study included 15 individuals with ASD and 10 typically developing controls.
- A 60-second artefact-free EEG segment was analyzed using MS-ROM/I-FAST and TWIST feature selection.
- Supervised machine learning classifiers were employed to analyze the invariant feature vector derived from EEG data.
Main Results:
- The machine learning system achieved 100% accuracy in classifying ASD cases versus controls using a training-testing protocol.
- Leave-One-Out cross-validation yielded accuracies between 84% and 92.8%.
- ANN weight matrix similarities were age-invariant, indicating detection of underlying brain disconnection signatures rather than age-related patterns.
Conclusions:
- This pilot study suggests that MS-ROM/I-FAST can effectively identify EEG signatures of ASD.
- The findings open new possibilities for developing non-invasive diagnostic tools for early ASD detection.
- Further research is warranted to validate these promising results in larger cohorts.
