Bayesian Analysis of Covid-19 Recovery and Death Cases in Nigeria
Olawale Basheer Akanbi
*
8 Everlasting Avenue, Daada Zone, Aroro Mankinde, Ibadan, Oyo State, Nigeria.
Prince Oluwaseyi Okunade
Department of Computer and Data Science, York St John University, London Campus, United Kingdom.
*Author to whom correspondence should be addressed.
Abstract
Uncertainty in Coronavirus disease 2019 (COVID-19) infections, recoveries, and deaths complicates public health decision-making in Nigeria. This study applies a Bayesian framework to quantify this uncertainty using state-level surveillance count data (n = 37) from the Nigeria Centre for Disease Control covering March 2020 to January 2024. The Weibull distribution was illustrated for recovery cases and the Lognormal distribution for death cases, with Gamma and Inverse Gamma conjugate priors respectively. The Weibull shape parameter (k = 0.8231) was fixed at its maximum likelihood estimate, yielding a closed-form posterior convergence for the scale parameter. Convergence diagnostics confirmed satisfactory model performance (R̂ < 1.05). The posterior mean recovery rate was 103.4 per 100,000 (95% credible interval: 70.8–150.9), supporting a national recovery rate of 97.48%. The posterior mean death rate was 1.610 per 100,000 (95% credible interval: 1.106–2.425), with mortality peaking at 7.60 per 100,000 in Lagos. Findings highlight high recovery but unequal mortality burden, informing targeted interventions.
Keywords: COVID-19, Bayesian inference, Markov Chain Monte Carlo, Weibull distribution, recovery and death cases