A Segmented SIR-D Mathematical Model for Coronavirus Propagation Dynamics (COVID-19) in Peru

The present study proposes the use of a segmented SIR-D mathematical model to predict the evolution of epidemiological populations of interest in the COVID-19 pandemic (Susceptible [S], Infected [I], Recovered [R] and dead [D]), information that is often key to guiding decision-making in the fight against epidemics. In order to obtain a better model calibration and a lower prediction error in the short term, we performed the model segmentation in 6 stages of periods of 14 days each. At each stage, the epidemiological  that define the system of equations are empirically estimated by linear regression of the epidemiological surveillance data that the Peruvian Ministry of Health collects and reports daily. This strategy showed better model calibration compared to an unsegmented SIR-D model.