Coronavirus disease 2019 (COVID-19) exhibits striking clinical heterogeneity, with asymptomatic carriage and severe symptomatic disease as distinct outcomes, but the underlying molecular mechanisms remain poorly defined. This study aimed to dissect the molecular basis of COVID-19 phenotypic divergence using integrated serum proteomics and causal inference approaches. Label-free serum proteomic profiling was performed on samples from healthy controls, asymptomatic infections (active and recovery phases), and symptomatic cases (incubation, acute symptom, and recovery phases) to characterize stage-specific proteomic signatures.