AUTOREGRESSIVE INTEGRATED MOVING AVERAGE (ARIMA) AND STATE-SPACE MODELING OF COVID-19 TRANSMISSION AND MORTALITY TRENDS IN GHANA

The COVID-19 pandemic revealed critical vulnerabilities in Ghana’s public health response, particularly the absence of robust, adaptive forecasting tools capable of guiding real-time interventions. To address this gap, the study evaluated the effectiveness of Autoregressive Integrated Moving Average (ARIMA) and state-space models in forecasting COVID-19 transmission and mortality trends in Ghana from 2020 to 2024. Using a retrospective quantitative design with secondary data covering 171,000 confirmed cases and 2,047 deaths, the research applied ARIMA to capture short-term infection dynamics and state-space models to uncover latent mortality patterns. ARIMA parameters (AR(1) = 0.65, MA(1) = -0.40, p < 0.01) and model fit indices (AIC = 1520, RMSE = 210) demonstrated solid short-term forecasting power. However, the state-space model outperformed ARIMA across all metrics (cases: AIC = 1480, RMSE = 190; deaths: AIC = 940, RMSE = 38), with key estimates-trend (β) = 5 and state variance = 80-proving its strength in modeling hidden drivers like under-reporting and intervention delays. The study found a strong positive correlation (r = 0.96) between cases and deaths, while regression analysis (R² = 0.93, p = 0.008) confirmed that case counts significantly predicted mortality. These results imply that integrating both models enhances accuracy, realism, and policy relevance in pandemic forecasting. The study recommends embedding state-space models into national surveillance systems and expanding modeling capacity through hybrid and AI-enhanced approaches for future epidemic preparedness.