This study offers perspectives into COVID-19 dynamics by employing a phenomenological model representing multiple epidemiological waves. It aims to support decision-making for health authorities and hospital administrators, particularly in optimizing intensive care unit (ICU) bed management and implementing effective containment measures. Given the intricate complexity of ICU environments, utilizing a mathematical model to anticipate occupancy is highly beneficial and might mitigate mortality rates associated with COVID-19. The study focuses on the evolution of intensive care patient numbers across multiple epidemiological waves in Italian regions. Our methodology involves the application of a low-complexity phenomenological model with an efficient optimization procedure. ICU occupancy data from five populous Italian regions are utilized to demonstrate the model’s efficacy on describing historical data and providing forecasts for two-week intervals. Based on the analyzed ICU occupancy data, the study confirms the efficacy of the proposed model. It successfully fits historical data and offers accurate forecasts, achieving an average relative RMSE of 0.51% for the whole fit and 0.93% for the predictions, across all regions. Beyond the immediate context, the model low complexity and efficient optimization make it suitable to diverse regions and diseases, supporting the tracking and containment of future epidemics.