To assess the predictive role of dynamic patterns and the interaction of physiological variables in the long-term outcome during the acute phase of coma following traumatic brain injury (TBI).

Prognosis in patients with acute TBI poses a clinical challenge. Currently, prognostic scores rely on neurological assessments in combination with physiological variables measured upon admission. However, from a multi-organ network perspective, patterns of physiological signals may provide valuable insights for predicting outcomes.

A cohort of 21 acute TBI coma patients admitted to the Neuro-ICU was stratified into two outcome groups: n=15 favorable (F); n=6 unfavorable (U), based on Glasgow Outcome Scale Extended (GOSE) at 6 months (F=GOSE>4). We retrospectively tracked multiple physiological measures, spanning from cardiovascular to metabolic, respiratory and autonomic functions, which were compared between outcome groups and before and after awakening from coma/sedation withdrawal. These measures were fed into a principal component analysis (PCA) to assess the interactions between physiological variables and the difference in the distribution of PCA scores between F and U groups (using the Kolmogorov-Smirnov test with significant p-value set at <0.05).

The PCA analysis on physiological signals showed three main components explaining around 70% of the variance. PC1 predominantly loaded on cardiovascular and inflammation measures, PC2 was primarily associated with metabolic parameters, and PC3 was linked to respiratory and coagulation factors. Notably, we observed that the scores for these three components could effectively differentiate between the F and U groups before awakening (PC1: KS=0.18; p<0.0001; PC2: KS=0.13; p<0.0001; PC3: KS=0.11; p=0.0002). These differences remained statistically significant after awakening between the U and F groups (p<0.017 for all PCs).