EEG complexity as a biomarker for autism spectrum disorder risk
William Bosl1, Adrienne Tierney, Helen Tager-Flusberg
1Harvard Medical School, Boston, MA, USA. william.bosl@childrens.harvard.edu
BMC Medicine
|February 24, 2011
Summary
Modified multiscale entropy (mMSE) from EEG signals can identify infants at high risk for autism spectrum disorder (ASD) by age 9 months. This novel biomarker aids early detection of neurodevelopmental disorders.
Area of Science:
- Neuroscience
- Biomarkers
- Developmental disorders
Background:
- Complex neurodevelopmental disorders may present subtle early brain function signatures.
- Nonlinear complexity of electroencephalography (EEG) signals reflects neural network architecture.
- Early detection of EEG abnormalities can serve as a biomarker for cognitive impairments.
Purpose of the Study:
- To demonstrate modified multiscale entropy (mMSE) from resting-state EEG as a biomarker for normal brain development.
- To distinguish typically developing infants from those at high risk for autism spectrum disorder (ASD).
Main Methods:
- Utilized mMSE as a feature vector for classification.
- Employed a multiclass support vector machine algorithm.
- Performed classification separately for age groups from 6 to 24 months.
Main Results:
- Multiscale entropy shows different developmental trajectories in high-risk autism infants versus controls.
- Classification accuracy exceeded 80% at 9 months for distinguishing high-risk and control infants.
- Classification accuracy for boys was near 100% at 9 months, remaining high at 12 and 18 months; accuracy for girls was highest at 6 months.
Conclusions:
- mMSE from resting-state EEG shows promise as a biomarker for early ASD risk detection.
- This study is the first to use information theoretic analysis of EEG for biomarkers in infants at risk for neurodevelopmental disorders.
