Short-Term Prediction of COVID-19 Deaths in Argentina

This paper proposes a dynamic model for accurate short-term prediction of the coronavirus disease (COVID-19) death rates in Argentina. For this purpose, the daily number of deaths in three of the most densely populated cities in the country were observed: Buenos Aires, Rosario and Mendoza. This methodology is based on the Richards growth model, and parameters are numerically optimized using a sequential least squares programming (SLSQP) method, with parameters obtained in the fitting process being accumulated and directly applied to a daily version of the model. The model fit is analyzed by inspecting its predictions for the first three epidemiological waves in each of the studied cities. The results presented for the selected cities in Argentina confirm the model’s accuracy in depicting death trends, despite the profound noise observed in data regarding the daily number of deaths.