Feed

Accurate assessment of residual cognition in patients with disorders of consciousness (DoC) remains challenging due to their limited behavioral responsiveness. Statistical learning (SL), defined as the ability to extract regularities from sensory input, reflects neural processing associated with residual cognition. This study examined how SL is preserved across different levels of consciousness and whether neural markers of SL can inform diagnosis and prognosis in patients with DoC.

We employed an auditory paradigm with structured trisyllabic pseudowords and recorded electroencephalography (EEG) from 46 patients and 25 healthy controls. Neural entrainment to statistical regularities was quantified using frequency-tagging analysis.

The within-group effect pattern was observed in the healthy control group and in the minimally conscious state (MCS) group, whereas the unresponsive wakefulness syndrome (UWS) group did not show a significant effect. Moreover, stronger neural learning predicted better behavioral scores in patients with MCS, but not in patients with UWS. Further analyses showed that SL-related effect correlated significantly with behavioral scores and 6-month outcomes in the traumatic group, whereas no such association was observed in the non-traumatic group. Importantly, SL-based EEG models effectively distinguished healthy controls from patients and further differentiated MCS from UWS patients. Combining the SL model with clinical behavioral performance could enable a more precise assessment of patients' level of consciousness.

These findings suggest that SL may require a minimal level of consciousness in the integration of structured regularities and that it may serve as an index for clinical diagnosis and prognosis in patients with DoC.