Forecasting Tuberculosis Incidence in Somalia: A Comparative Analysis of Single and Hybrid Time-Series Models

Tuberculosis (TB) remains a significant public health challenge, necessitating accurate forecasting methodologies to support effective control and prevention strategies. This paper explores the application and comparative performance of single and hybrid time-series models for forecasting TB incidence trends specifically in Somalia.

Annual TB incidence data from 2000 to 2022 were sourced from the World Bank to train and evaluate a comprehensive suite of 14 time-series models. This included five single models-ARIMA, ETS, TBATS, Theta, and NNAR-and nine hybrid model combinations (e.g., ARIMA-ETS, ARIMA-TBATS, ARIMA-ETS-TBATS). Model performance was assessed using Theil's U statistic, Mean Absolute Percentage Error (MAPE), Symmetric Mean Absolute Percentage Error (SMAPE), and Root Mean Square Error (RMSE).

Among the single time-series models, the TBATS model demonstrated the best fit. However, the comparative analysis revealed that the hybrid ARIMA-ETS-TBATS model outperformed other hybrid configurations. The study highlights that hybrid modeling offers enhanced forecasting accuracy compared to single models.

The resulting forecasts provide valuable insights into future TB incidence trends in Somalia. These findings underscore the importance of hybrid modeling in generating accurate data to aid informed public health decision-making and the development of targeted intervention strategies for TB control.