Modeling and Projecting Malaria Incidence in Somalia: A Dual-Approach Framework Combining Statistical Forecast Accuracy with Mechanistic Epidemiological Insight

Malaria remains a significant public health issue in Somalia, complicated by a modeling dilemma where policymakers must choose between mechanistic models offering insight without accuracy, and statistical models providing accuracy without explanation. This study compares eight statistical time-series models with a mechanistic Susceptible-Infected-Recovered-Susceptible (SIRS) model using annual malaria incidence data from 2000–2022. The aim was to assess both quantitative forecasting accuracy and the generation of actionable epidemiological intelligence. A clear performance trade-off was observed. The Autoregressive Fractionally Integrated Moving Average (ARFIMA) model exhibited superior predictive accuracy, with a Mean Absolute Percentage Error (MAPE) of 7.44% on a hold-out test set. Conversely, the SIRS model, though less accurate (MAPE = 24.15%), offered vital mechanistic insights. It estimated a Basic Reproduction Number \(\:(R₀)\) of 1.19 and highlighted transmission and recovery rates as critical intervention parameters. The study concludes that these modeling approaches are complementary, not competing. It recommends employing the accurate ARFIMA model for short-term operational forecasting and resource allocation. Simultaneously, the insightful SIRS model should be used for long-term strategic planning and prioritizing public health interventions. This dual framework offers a more robust and comprehensive evidence base for effective malaria control in Somalia.