Objectives: The purpose of this article is to describe and develop the predictive value of three models during the COVID-19 epidemic in Chile, providing knowledge for decision-making in health.
Methods: We developed three models during the epidemic: a discrete model to predict the maximum burden on the health system in a short time framea basic SEIR (susceptible-exposed-infected-removed) model with discrete equations; a stochastic SEIR model with the Monte Carlo method; and a Gompertz-type model for metropolitan city of Santiago.
Results: The maximum potential burden model has been useful throughout the monitoring of the epidemic, providing an upper bound for the number of cases, intensive care unit occupancy, and deaths. Deterministic and stochastic SEIR models were very useful in predicting the rise of cases and the peak and onset of case decline; however, they lost utility in the current situation due to the asynchronous recruitment of cases in the regions and the persistence of a strong endemic. The Gompertz model had a better fit in the decline since it best captures the epidemic curves asymmetry in Santiago.
Conclusions: The models have shown great utility in monitoring the epidemic in Chile, with different objectives in different epidemic stages. They have complemented empirical indicators such as reported cases, fatality, deaths, and others, making it possible to predict situations of interest and visualization of the short and long-term local behavior of this pandemic.