Abstract
Influenza virus infections vary widely in severity, with human challenge studies revealing substantial heterogeneity in viral shedding and immune responses. Yet, the mechanistic basis of this variability remains unresolved. Here, we applied a multivariate mechanistic modeling framework to determine how variation in viral replication and CD8(+) T cell responses shapes infection kinetics and symptom dynamics in individuals experimentally challenged with H1N1 influenza virus. The analysis identified six distinct infection clusters arising from multivariate influences, including viral dose and individualized rates of viral infectivity. Baseline T cell levels contributed minimally, and model outputs suggested a tradeoff between CD8(+) T cell efficacy and expansion. Two participants showed evidence of reinfection despite being antibody negative, and modeling their data indicated that host-pathogen interactions remained consistent with those from primary infections, though data were insufficient to distinguish between cytotoxic and IFN- γ -mediated protection. Modeling of symptom trajectories further revealed potential subjectivity in symptom reporting, independent of viral strain. Collectively, these findings demonstrate how mechanistic modeling can improve the predictability of influenza outcomes and reveal that intrinsic human variation can converge on similar infection trajectories.