Benchmarking single-subject detectability of cognitive event-related potentials in healthy controls and minimally conscious state patients

In healthy controls, single-subject ERP detectability is high for MMN and P300 (48/48), but drops for sentence-congruity (69%) and own-name (35%).

• Why it matters: Group-level ERP differences may not translate to reliable detection in individual patients, limiting clinical use for assessing residual cognition and prognosis in disorders of consciousness.
• What they did: A hierarchical battery of six auditory ERP paradigms (MMN, passive/active oddball P300, own name, semantic oddball, sentence N400) was tested in 48 controls and 26 minimally conscious state patients using blinded visual scoring, ROI ANOVA, cluster permutation tests, and support vector machine decoding.
• The result: In MCS, detection rates were lower (e.g., MMN 27% for short deviant and 65% when any deviant type counted; semantic oddball 8%), and binary ERP classifications were not significantly associated with recovery in 22 follow-up survivors.

AI summary of the abstract · may contain errors · not medical advice · read the paper.