Machine Learning Model for Predicting the Temporal Lassa Fever Confirmed Cases in Nigeria

Lassa fever continues to pose a major public health threat in Nigeria, marked by recurrent outbreaks and high case fatality rates. The absence of robust predictive models has significantly impeded accurate trend forecasting, thereby limiting timely resource deployment and the implementation of effective preventive measures. This study seeks to bridge that gap by developing a comprehensive predictive framework for estimating confirmed Lassa fever cases in Nigeria. The research utilizes a combination of quantitative analysis and computational modeling techniques, leveraging weekly epidemiological data on Lassa fever cases from the Nigeria Centre for Disease Control (NCDC), spanning the year 2020. The dataset, which includes both suspected and confirmed cases, was cleaned and restricted to confirmed cases for the purpose of this analysis. Key steps included feature selection and dimensionality reduction to enhance model efficiency and accuracy. Three predictive models—Random Forest, Linear Regression, and Gradient Boosting—were developed and assessed using standard evaluation metrics such as Mean Absolute Error (MAE), Root Mean Square Error (RMSE), and R-squared (R²). These models were designed to forecast future confirmed Lassa fever cases. The findings highlight the critical role of temporal variables, particularly weeks and months, in shaping transmission patterns. These features were shown to significantly influence the trends in confirmed cases.