Related Experiment Video
Updated: Aug 22, 2025
11:50
EEG Mu Rhythm in Typical and Atypical Development
Published on: April 9, 2014
25.9K
Intellectually able adults with autism spectrum disorder show typical resting-state EEG activity
Qianliang Li1,2, Ricarda F Weiland3, Ivana Konvalinka1
1Section for Cognitive Systems, Department of Applied Mathematics and Computer Science (DTU Compute), Technical University of Denmark, 2800, Kongens Lyngby, Denmark.
Scientific Reports
|November 8, 2022
Summary
Electroencephalography (EEG) biomarkers did not reliably distinguish adults with autism spectrum disorder (ASD) from controls. Despite high validation accuracy, machine learning models showed poor performance on unseen data, highlighting ASD heterogeneity.
Area of Science:
- Neuroscience
- Psychiatry
- Biomarker Discovery
Background:
- Quantifiable physiological biomarkers are sought for psychiatric disorders.
- Identifying biomarkers for autism spectrum disorder (ASD) is challenging due to high heterogeneity.
- Electroencephalography (EEG) is a potential tool for biomarker discovery.
Purpose of the Study:
- To investigate the potential of EEG biomarkers for classifying ASD in adults.
- To assess the efficacy of machine learning models in distinguishing ASD from controls using EEG data.
Main Methods:
- Recorded 5-minute eyes-closed resting-state EEG from 186 adults (51% ASD).
- Utilized machine learning models with two-layer cross-validation.
- Analyzed spectral, temporal, and spatial dimensions of source-modelled EEG, generating 3443 biomarkers.
Main Results:
- No significant group-mean or group-variance differences were found in EEG features.
- Machine learning models achieved >80% accuracy during validation.
- The best model showed only 56% balanced test accuracy for ASD classification, indicating a significant performance drop.
Conclusions:
- Rigorous model evaluation is crucial, as demonstrated by the performance gap between validation and testing.
- High heterogeneity in ASD significantly impacts biomarker discovery and classification accuracy.
- Intellectually able adults with ASD exhibit typical resting-state EEG at the group level.
