ABSTRACT
The study presents data science models for a real-time forecast of COVID-19 size and spread in Nigeria. Firstly, an exploratory and comparative study of the disease spread in Nigeria and some other African nations are carried out. Then variants of support vector machine (SVM) using the Gaussian kernel and regression machine learning models suitable for modeling count data variables are built to estimate a 15-day prediction of infection cases. The data science models built in this research give a short-term forecast of the disease’s spread which is useful in better understanding the spread patterns of the disease as well as enabling future preparedness and better management of the disease by the government and relevant authorities. The research outcome can therefore serve as an effective decision support system. This work can also serve as an alternative to the mathematical-based epidemiological models for the forecast of COVID-19 spread because of their inherent advantages of learning from historical datasets and generalizing with new sets of data which promises better results. © 2022, The Author(s), under exclusive license to Springer Nature Switzerland AG.
ABSTRACT
Infectious diseases can inflict immense losses and suffering on the human population. As at 23rd of June, 2020 COVID-19 pandemic has caused 20,919 cases, 25 deaths and 7,109 had recovered in Nigeria. Nigeria Centre for Disease Control (NCDC) is tracing COVID 19 carriers for designing effective control measures and to prevent the spread. Authors have modeled COVID-19 cases, but there is a dearth of information on estimating the total number of hidden COVID-19 carriers in the population. Adaptive cluster sample was used for exploring populations of hidden COVID-19 carriers. The data on daily cases of COVID-19 were extracted from NCDC website. Nigeria population was partitioned into 37 regions (states and FCT). We considered a model based approached in Bayesian framework to make inference about the number of COVID-19 carriers in Nigeria. The fitted model showed that all COVID-19 carriers will only be captured at once if contact tracing is combined with methodology designed in this work. © 2020-IOS Press and the authors. All rights reserved.
ABSTRACT
COVID-19 is battling with many countries in the world, including Nigeria, and it has affected various sectors. Contact tracing technique without Statisticians in the team as recommended by WHO is being used in Nigeria to curb the spread of COVID-19 virus, yet confirmed cases is on the increase daily. This study proposed the integration of Statistical techniques for improving contact tracing efforts to stop the spread of the virus. A fitted model using the R package, and Adaptive Cluster Sampling mechanism was embedded. Parameters of the model were estimated using Markov Chain Monte-Carlo (MCMC) Algorithm with Winbugs software. Trace plot and correlogram were used for MCMC diagnostics to examine the goodness of fit of the model. The fitted model was used to obtain a predictive distribution for predicting the estimated number of COVID-19 carriers in Nigeria. The model has a good fit since It converged to the representation of the target posterior within the 95% highest posterior density (HPD) interval, its chains mixed well, and autocorrelation is quite similar at each lag. Estimated number of COVID-19 carriers were well estimated and higher in each state than confirmed cases. The present contact tracing process is inefficient to track COVID-19 carriers, hence integrated contact tracing technique with the involvement of Statisticians was recommended. .