Abstract
Seasonal epidemics of influenza and respiratory syncytial virus (RSV)-bronchiolitis pose significant challenges for public health systems, requiring timely predictions for effective interventions. In this study, we applied a Gompertz model to cumulative daily primary care diagnoses in Catalonia to predict epidemic peaks and characterize the dynamics of the disease. We estimated RSV-bronchiolitis cases from all-cause bronchiolitis diagnoses and computed epidemic thresholds for the disease. Our approach allowed for peak predictions up to 32 days in advance with an error margin of one week (anticipated). The estimated magnitudes were within 35% error 28 days before the peak and mostly fell within 95% confidence intervals, except for the irregular 2022-2023 RSV-bronchiolitis post-COVID-19 pandemic season. Influenza epidemics exhibited a faster decline, resulting in more symmetrical curves, whereas RSV-bronchiolitis outbreaks were broader, with a higher initial transmission rate. The model operates in real-time without reliance on external assumptions, making it adaptable to changes in epidemiology. However, human intervention to set the models' parameters enables them to be fitted more precisely, resulting in even better performance through iterative refinement. Our findings highlight the potential of supervised real-time predictive modeling to support epidemic preparedness and optimize healthcare resource allocation.