AI-Driven Bayesian Model Enhances Malaria Forecasting for Ghana

Malaria remains a significant public health challenge in sub-Saharan Africa, with effective disease quantification hampered by inconsistent and noisy surveillance data. In Ghana, specifically, health facility records from 2014 to 2023 reveal complex, non-linear patterns in hospital admissions that existing models struggle to accurately capture, particularly regarding stochastic variability and reliable uncertainty bounds. This research addresses these limitations by introducing a sophisticated Bayesian nonlinear inference framework.

The newly developed framework integrates a cubic baseline with a damped oscillatory kernel, with parameters estimated using an affine-invariant ensemble Markov Chain Monte Carlo sampler. This methodology is particularly robust, designed to effectively handle limited data while accurately modeling parameter uncertainty. It generates probabilistic forecasts, offering a more nuanced understanding of future disease trends for distinct age groups: children under five years and individuals aged five years or more.

The study demonstrates the strong empirical adequacy of the model, achieving R2 values of approximately 0.99 for both age cohorts, with residual errors below 2%. A district-level analysis further highlights significant spatial heterogeneity in malaria dynamics across Ghana, with urban centers showing much lower coefficients of variation compared to peripheral districts. These findings underscore the localized nature of malaria transmission and the need for tailored interventions.

Crucially, the framework provides probabilistic forecasts for 2024-2026, predicting a gradual resurgence in malaria cases. For children under five, cases are projected to rise from 137,000 to 149,000, and for older individuals, from 348,000 to 375,000, with increasing uncertainty over time. This predictive capability offers a critical advantage for proactive public health planning.

By delivering these data-driven insights and probabilistic forecasts, this Bayesian framework furnishes Ghana's national malaria control strategy with a powerful and principled tool. It enables health authorities to better anticipate malaria fluctuations, thereby strengthening decision-making and resource allocation to combat the disease more effectively across the country.