EEG-Based Diagnosis of Autism Spectrum Disorder

Autism Spectrum Disorder (ASD) encompasses a range of neurodevelopmental conditions characterised by challenges in social communication, restricted interests and repetitive behaviours. Traditional diagnosis relies on behavioural assessments often administered in specialist settings, leading to delays in identification and intervention. Electroencephalography (EEG) offers a non-invasive, low-cost means to capture brain activity with millisecond resolution, making it an attractive candidate for objective biomarker development. Recent advances have focused on extracting both linear and non-linear features from resting-state or task-based EEG, including time-frequency decompositions, entropy measures and higher-order statistics. These features serve as inputs to machine learning and deep learning classifiers, which can distinguish between ASD and typically developing profiles with high accuracy. Early detection studies have demonstrated that EEG-derived digital biomarkers present from three months of age may predict later diagnostic outcome, opening a window for timely therapeutic support. Moreover, portable EEG systems and automated signal-processing pipelines hold promise for scaling screening efforts globally, especially in low-resource settings. The integration of robust feature extraction, rigorous validation against demographic confounders and user-friendly interfaces is catalysing the translation of EEG-based tools into clinical and community contexts, with the potential to transform diagnostic pathways and optimise outcomes for individuals on the spectrum.

Research from Nature Portfolio

A foundational study employed longitudinal EEG recordings from infants at high familial risk of ASD, analysing non-linear signal features and applying statistical learning algorithms. This approach achieved sensitivity and specificity above 95% when predicting diagnostic outcome as early as three months, and correlated strongly with later behavioural severity scores. The work underscored the feasibility of extracting reliable digital biomarkers from routine EEG and highlighted the developmental trajectories of neural complexity in at-risk populations.

EEG-Based Diagnosis of Autism Spectrum Disorder publication trend

The graph below shows the total number of articles in eeg-based diagnosis of autism spectrum disorder across all publications each year (not limited to Nature Index journals).

Technical terms

Electroencephalography (EEG): Non-invasive recording of electrical activity along the scalp produced by neuronal firing.

Digital biomarker: Quantitative physiological and behavioural data collected and measured by digital devices and algorithms.

Spectrogram: Time-frequency representation of a signal showing how its spectral density varies over time.

Bispectrum: A higher-order spectral analysis technique that captures phase relationships between frequency components of a signal.

Shannon entropy: A measure of the uncertainty or complexity present in a signal’s amplitude distribution.

References

1. EEG‐Based Computer Aided Diagnosis of Autism Spectrum Disorder Using Wavelet, Entropy, and ANN. BioMed Research International (2017).
2. EEG complexity as a biomarker for autism spectrum disorder risk. BMC Medicine (2011).
3. Recurrence quantification analysis of resting state EEG signals in autism spectrum disorder – a systematic methodological exploration of technical and demographic confounders in the search for biomarkers. BMC Medicine (2018).
4. Autism Spectrum Disorder Diagnostic System Using HOS Bispectrum with EEG Signals. International Journal of Environmental Research and Public Health (2020).
5. A spectrogram image based intelligent technique for automatic detection of autism spectrum disorder from EEG. PLOS ONE (2021).