Evolutionary game theory modelling of COVID 19 vaccination dynamics and epidemic outcomes in Apulia Italy

Abstract Introduction The COVID-19 pandemic highlighted the critical role of individual decision‑making in shaping epidemic outcomes, particularly with respect to vaccination uptake. Method In this study, we propose a data‑driven compartmental model that integrates epidemiological dynamics with behavioral responses to vaccination using an evolutionary game‑theoretic framework. We develop a time‑dependent SIRDVB (Susceptible, Infected, Recovery, Deceased, Vaccinated, Booster) model calibrated on regional data from Apulia, Italy, covering the period January 2021–December 2022. Vaccination and booster decisions are modeled through evolutionary game theory, with payoff differences inferred directly from observed epidemiological time series. Results The proposed model accurately reproduces observed epidemic and vaccination dynamics, achieving mean absolute percentage errors below 5% for all compartments except for active infections, which showed a 12.59% error. The estimated utility function became positive during epidemic waves, coinciding with increases in vaccination and booster uptake, highlighting a dynamic behavioral response to perceived infection risk. Scenario analyses show that moderate changes in vaccination participation substantially affect epidemic burden and mortality. A direct comparison with a baseline non‑EGT SIRDVB model demonstrates that the inclusion of behavioral feedback markedly improves model performance across all compartments, particularly for vaccination and booster dynamics. Conclusion The SIRDVB model after smoothing fitted correctly the data. These findings indicate that explicitly accounting for behavioral adaptation is essential for realistically modeling vaccination campaigns and epidemic evolution. The application of a game-theoretic framework provides a transparent and generalizable approach for integrating behavioral feedback into epidemic models, supporting more effective, context‑sensitive public health strategies.