Resting-State Functional Connectivity and Cognitive Performance in Long COVID: A Randomized Controlled Trial

Cognitive impairment is a prevalent, disabling manifestation of Long COVID, but whether cognitive rehabilitation alters neural network organization is unknown. We examined whether a 12-week cognitive rehabilitation intervention modified resting-state connectivity within the central-executive (CEN), default-mode (DMN), and salience (SN) networks, and whether connectivity changes associated with cognitive performance. Participants were adults with prior SARS-CoV-2 infection, persistent Long COVID symptoms (≥3 months), and objectively confirmed mild-to-moderate cognitive impairment. Of 58 participants randomized, 39 (mean age 46.4; 69% women) completed neuroimaging (19 intervention, 20 control). The 12-week intervention combined virtual cognitive rehabilitation and small-group and individual sessions with computerized cognitive training. No significant group, time, or group-by-time interaction effects were observed for within- or between-network connectivity (all P > 0.22). Exploratory analyses identified nominal associations (P < 0.05) between connectivity changes and cognition; improved processing speed associated with increased CEN-DMN connectivity, greater SN-DMN connectivity associated with worse performance across processing speed, working memory, and executive function. None survived correction for multiple comparisons. Cognitive rehabilitation was not associated with detectable changes in resting-state network connectivity in long COVID–related cognitive impairment. SN-DMN connectivity may represent a marker of reduced network segregation underlying cognitive dysfunction, warranting study in larger, multimodal trials. Trial registration: NCT05498493 .