Resting state EEG power spectrum and functional connectivity in autism: a cross-sectional analysis.
Pilar Garcés, Sarah Baumeister, Luke Mason and 23 others
PMID 35585637WHAT IT FOUND
Resting-state brain activity patterns did not reliably distinguish autistic people from neurotypical peers in this large sample.
A machine-learning model trained on the data failed to reproduce its accuracy in an independent group, so EEG is not a diagnostic tool here.
Key findings
01No differences in resting-state EEG power or connectivity between autistic and neurotypical groups survived validation in an independent sample.
02Machine-learning classifiers trained on EEG features performed poorly, with accuracies of 47-57% in the training set that did not replicate in the validation set.
03A trend for reduced alpha reactivity to eye opening in autistic adults was found in the training data but was not significant in the validation data.
STILL TO COME
How it was doneWhat they found
Read the rest of this summary
You get three full summaries a month, free, and we do not ask for a card. Search, the TL;DRs and your library stay unlimited either way.
What it does not show
The study only included individuals with an IQ above 75, so results do not generalise to autistic people with intellectual disability. The validation sample was underpowered to detect small effects, particularly those specific to certain age groups. The cross-sectional design prevents conclusions about developmental trajectories. Participants who could not complete the EEG recording were excluded, and this excluded group had higher symptom severity, introducing potential selection bias.
Declared interests
The study was supported by the Innovative Medicines Initiative 2 Joint Undertaking under grant agreement No 777394 for the project EU-AIMS, which receives support from the European Union's Horizon 2020 research and innovation programme and EFPIA. The EU-AIMS LEAP project also receives support from the European Federation of Autism Research and the European Commission. No specific commercial sponsor for the treatment or intervention was identified as designing or writing the manuscript.
The easy way to misread this
Do not interpret the significant machine-learning results in the training set as evidence that EEG can diagnose autism. These findings failed to replicate in the independent validation set, and the authors state the classifier performance suggests limited clinical utility.