In one sentence
Looking at how a child’s brain-wave patterns fluctuate from moment to moment — not only their average levels — better separated autistic from non-autistic children in this research sample, though accuracy is still far from a clinical test.
What the researchers did
The team analysed resting-state EEG (a recording of the brain’s electrical activity) from 190 children: 94 autistic and 96 typically developing, drawn from a multi-site Autism Center of Excellence project. They compared three ways of describing brain-wave power across five frequency bands: traditional trial-averaged power, features from all individual trials, and “across-trial variability” — how much activity fluctuates across moments. Four machine-learning models were trained and checked with repeated cross-validation.
What they found
- Across-trial variability features reached about 70.7% classification accuracy and outperformed traditional mean EEG features.
- Variability in the delta (1–4 Hz) and gamma (30–40 Hz) bands was especially informative.
- Patterns held across models and cross-validation folds, and key variability measures correlated with behavioural scores in autistic participants.
What this means for families and therapists
- This is research on a possible biomarker direction, not a clinic-ready diagnostic tool. Do not treat EEG variability as a stand-alone autism test.
- For therapists, the idea that neural “stability vs flexibility” matters may eventually connect to sensory and learning differences — but that link still needs careful clinical translation.
- Parents can ask whether research teams or specialist centres are studying EEG biomarkers; routine clinical EEG for this purpose is not established.
Limitations and what we don't know yet
- Accuracy around 71% is useful for research comparisons, not diagnosis.
- The sample, while multi-site, is still modest and not representative of all autistic people.
- Only resting-state EEG was analysed; task-based and long-term change remain open questions.
- Authors frame this as a framework for future biomarker work, not a ready clinical assay.
This is a plain-language summary of Harnessing Trial-to-Trial Variability of EEG Spectral Characteristics to Understand Autism by Shama D.M., Su M., Beeler-Duden S. et al., Journal of autism and developmental disorders (2025). Source license: CC-BY-NC-ND-4.0. It is not medical advice — talk to a qualified clinician before changing therapy.