Estimating undocumented Covid-19 infections in Cuba by means of a hybrid mechanistic-statistical approach

TL;DR

This paper estimates the true extent of undocumented Covid-19 infections in Cuba using a hybrid model, revealing significant underreporting and early epidemic peaks, with implications for understanding disease spread.

Contribution

It introduces a hybrid mechanistic-statistical method to accurately estimate undocumented infections and epidemic dynamics in Cuba.

Findings

60% of infections were undocumented

Epidemic peaked ten days earlier than reported

Reproduction number declined after 80 days

Abstract

We adapt the hybrid mechanistic-statistical approach of Ref. [1] to estimate the total number of undocumented Covid-19 infections in Cuba. This scheme is based on the maximum likelihood estimation of a SIR-like model parameters for the infected population, assuming that the detection process matches a Bernoulli trial. Our estimations show that (a) 60% of the infections were undocumented, (b) the real epidemics behind the data peaked ten days before the reports suggested, and (c) the reproduction number swiftly vanishes after 80 epidemic days.