Forecasting Under-5 Mortality Rate in Somalia to 2030: a comparative analysis of univariate and multivariate ARIMAX models.

Background: Accurate forecasting of the Under-5 Mortality Rate (U5MR) is essential for monitoring health progress and informing policy interventions in conflict-affected regions. In Somalia, achieving the Sustainable Development Goal (SDG) 3.2 target of reducing child mortality to at least 25 deaths per 1,000 live births by 2030 remains a formidable challenge due to persistent socio-political instability and economic shocks. Methods: Utilizing a two-phase analytical approach, the study first evaluated 11 time-series models (6 single and 5 hybrid) based on historical U5MR data from 1960 to 2023. In the second phase, a multivariate ARIMAX framework (1989-2023) was implemented to incorporate exogenous drivers, including total conflict-related fatalities, GDP per capita, and immunization coverage. Model performance was validated using MAPE, sMAPE, and Theil's Results: The univariate analysis identified the TBATS and hybrid ARIMA-TBATS models as the superior forecasting tools, with the latter selected for its robustness in handling Somalia's high data volatility. The multivariate ARIMAX model revealed that conflict fatalities are a highly significant predictor of mortality ( Conclusion: The findings underscore a concerning stagnation in child mortality reduction. Achieving international benchmarks in Somalia requires a dual focus on intensifying health interventions and ensuring national security stability. These data-driven insights offer critical evidence for policymakers to design resilient strategies for child survival in the region.