Prediction of the Ebola Virus Epidemic using Data-Driven Modeling: A Focus on the historical Western African Ebola Virus Epidemic

Ebola virus disease is among the deadliest diseases in human history and numerous measures have been established to contain it. However, in the absence of a vaccine to combat the disease, the established measures may not serve their utmost purpose. The reliance on data-driven modeling offers hope toward combating this deadly zoonotic viral disease through early detection. Decision trees, gradient boosting, random forests, and deep neural network were used as machine learning algorithms to forecast the next Ebola cases and deaths in Guinea, Liberia, and Sierra Leone. The forecasting was done for the next nineteen years from the last year of registering Ebola cases and deaths (2016) in these three countries to 2035. Predictions from decision trees, gradient boosting, and random forest revealed rises in Ebola cases and deaths up to 2035. Predictions from deep neural networks revealed a surge in Ebola cases and deaths up to 2035. The forecasted Ebola cases in the three countries were remarkably higher than the forecasted Ebola deaths. These findings propose data-driven modelling as an effective tool to model the Ebola outbreak implying that prediction of Ebola outbreaks can be reached by relying on machine learning.