An Analysis of Measles Incidence in Somalia Using Time Series

<ns3:p>Background Measles remains a major public health concern in Somalia, characterized by recurrent outbreaks and fluctuating vaccination coverage levels. Despite global progress toward measles elimination, fragile health systems, conflict, and immunization gaps continue to sustain transmission in the country. Understanding long-term trends and forecasting future incidence patterns are essential for evidence-based planning. This study aimed to analyze annual measles incidence in Somalia and evaluate its temporal dynamics using advanced time series modeling techniques. Methods A longitudinal time series design was employed using nationally aggregated annual data on measles incidence in Somalia covering the period 1980–2024. Stationarity of the series was assessed using the Augmented Dickey–Fuller (ADF), Phillips–Perron (PP), and KPSS tests. After first differencing, multiple univariate models—including ARIMA, ETS, TBATS, Theta, NNAR, and ARFIMA—were developed and compared. In addition, hybrid forecasting models were constructed by combining selected individual models. Model performance was evaluated using Mean Absolute Percentage Error (MAPE), Symmetric MAPE (sMAPE), and Theil’s U statistic based on training and testing datasets. Results The original measles incidence series exhibited non-stationary behavior and required first differencing (ndiffs = 1). Among the individual models, ARIMA(0,1,2) with drift demonstrated the best forecasting performance (MAPE = 7.53%), outperforming ETS, TBATS, Theta, NNAR, and ARFIMA models. Among hybrid approaches, the ARIMA–Theta combination showed the strongest performance. Forecast projections for 2025–2030 indicate a continued increasing trend in measles incidence if current patterns persist, highlighting potential future public health challenges. Conclusions The findings demonstrate that time series modeling provides robust insights into measles dynamics in Somalia. The upward forecasted trend underscores the urgent need to strengthen routine immunization coverage and outbreak preparedness strategies. The study contributes novel empirical evidence using integrated time series and hybrid modeling approaches, supporting data-driven decision-making for measles control and elimination efforts in fragile settings.</ns3:p>