Prediction of emergence from the minimally conscious state at 6 months using acute-phase clinical and event-related potential indicators in patients with prolonged disorders of consciousness: a single-center retrospective cohort study

Objective To identify acute-phase predictors of long-term outcome in patients with disorders of consciousness (DoC) and establish a multimodal prediction model. Methods This retrospective study enrolled 227 patients with prolonged DoC at the Second Affiliated Hospital of the University of South China between August 2019 and December 2025. The primary endpoint was emergence from the minimally conscious state (eMCS) at 6 months after onset, defined as functional communication or functional use of objects on the Coma Recovery Scale-Revised (CRS-R), which was assessed at 1, 3 and 6 months after onset. Clinical, imaging and neurophysiological acute-phase data (≤28 days after onset) were collected. Variables significant in univariate analysis entered multivariable logistic regression; clinical, electrophysiological and combined models were compared by the area under the receiver operating characteristic curve (AUC), calibration and clinical utility, and the final model was presented as a nomogram. Results Univariate screening identified age, acute-phase FOUR score, MMN and P300 amplitudes, cerebral herniation, etiology, hydrocephalus and st-SCS surgery (all p < 0.05); age, etiology, FOUR score and Fz P300 amplitude remained independent in multivariable analysis. Discrimination was acceptable for the clinical (AUC = 0.745, 95% CI 0.681–0.809) and multimodal (AUC = 0.772, 95% CI 0.711–0.832) models, but not significantly different (ΔAUC = 0.027, DeLong test p = 0.115); after bootstrap correction the multimodal AUC was 0.735, with adequate calibration (Hosmer-Lemeshow p = 0.634). Adding ERP measures improved reclassification relative to the clinical model (continuous NRI = 0.317, 95% CI 0.062–0.572, p = 0.015; IDI = 0.039, 0.014–0.065, p = 0.003). Performance did not differ between TBI and non-TBI patients (ΔAUC = 0.061, p = 0.346), although these comparisons were underpowered. Conclusion Acute-phase FOUR score, age, etiology and the Fz P300 amplitude were independently associated with emergence from the minimally conscious state at 6 months in prolonged DoC. A nine-indicator multimodal model showed acceptable, internally validated discrimination (AUC = 0.772; bootstrap-corrected 0.735) and modestly improved reclassification over clinical variables alone (continuous NRI = 0.317; IDI = 0.039). External validation is required before clinical use.