Analysis of Healthcare Data to Predict Malaria Incidence in Benue State

Malaria remains the leading cause of morbidity in Benue State, and predictive modelling of incidence patterns can improve resource allocation and preventive intervention timing. This study used healthcare records and environmental data to build predictive models for monthly malaria incidence in Benue State, North Central Nigeria. Monthly malaria case counts from 35 Primary Healthcare Centre facilities across Makurdi, Otukpo, and Katsina-Ala Local Government Areas were merged with rainfall, temperature, and relative humidity data from the Nigeria Meteorological Agency for 2017 to 2022. Negative binomial regression and gradient boosting models were compared for predictive performance. The gradient boosting model achieved an RMSE of 48.3 cases per facility per month and an R-squared of 0.74 on the test set, outperforming the regression baseline. Rainfall in the preceding month was the strongest predictor, followed by maximum temperature and healthcare facility catchment population. Seasonal decomposition revealed consistent incidence peaks in August and September following the rainy season onset. Facilities in riverine LGAs showed incidence rates 3.1 times higher than upland facilities after controlling for population. Gaps in facility reporting affected approximately 14 percent of monthly records. Recommendations include deployment of this predictive framework as an alert system for the Benue State Ministry of Health, targeted pre-season insecticide net distribution to high-risk catchment areas, and integration of satellite-derived vegetation index data as additional environmental predictors to improve model accuracy.

Keywords: malaria prediction, gradient boosting, healthcare data, Benue State, environmental variables