Feed

COVID-19 presents a wide spectrum of clinical outcomes, and early identification of patients at high-risk for a worse disease prognosis remains a critical challenge. In this study, we performed a six-week longitudinal proteomic analysis of 197 plasma samples collected weekly from 44 hospitalized COVID-19 patients. Temporal profiling revealed 828 proteins with significant time-dependent abundance changes, clustering into an early acute-phase and inflammatory response group that gradually declined toward the cohort's overall mean profile over time and a group of late vesicle-mediated transport and antigen binding that were decreased in the first weeks. Global correlation mapping identified six coregulated protein modules, notably including insulin-like growth factor binding proteins. Proteomic divergence between survivors and nonsurvivors became pronounced from week 4 onward. The proteins HPR, IGFALS, PF4, LPA, and COLEC11 achieved AUCs > 0.72 for mortality prediction in the first 2 weeks, and an SVM model combining these markers yielded an AUC of 0.801. These findings demonstrate that plasma proteome trajectories capture dynamic host responses and generate candidate prognostic markers, offering a promising strategy for risk stratification and guiding targeted interventions in COVID-19.