In this paper, entropy was studied in non-linear models including exponential, Gompertz, and logistic, to estimate epidemiological parameters of interest in data from confirmed cases of infection by COVID-19 in Peru. The data related to the spread of COVID-19 in Peru comes from the information available on the INS-Peru institutional portal (2020). The Akaike information criterion (AIC) and the residual standard error (ERR) were considered to evaluate the entropy of the models. The estimation of the parameters of the models was carried out using maximum likelihood and by the Bootstrap method. The results showed that the entropy of the models is related to the information generation rate, associated with the differential in the number of tests applied. Entropy severely affected maximum likelihood estimators. The Bootstrap estimators showed better performance against EMV with the estimated peak of confirmed cases. Bootstrap estimators were significantly affected by sample size, especially when n ≤ 10. The results of this research suggest considering the entropy and the information generation rate (differential in the application of tests for the diagnosis of COVID-19 in Peru), as well as the use of Bootstrap estimators as an alternative to estimate parameters of epidemiological models.