Forecasting Maternal Mortality in Sub-Saharan Africa to 2035: Evidence from the Global Burden of Disease 2023 Study

Abstract Maternal mortality continues to be an enduring measure of global health disparity, with Sub-Saharan Africa disproportionately contributing to the global burden of maternal mortality despite a reduction over the past three decades. Forecasts of future trends play a critical role in health policy decision-making. No prior study has used a unified statistical, machine learning, and deep learning comparative approach to model maternal mortality in Sub-Saharan Africa, based on the most recent Global Burden of Disease (GBD) estimates. In this study, GBD 2023 maternal mortality ratio (MMR) estimates for Sub-Saharan Africa (1980-2023, 44 yearly observations) were analyzed to compare eight models under three different methods: statistical methods (Holt damped-trend exponential smoothing, ARIMA, log-linear regression), machine learning (Random Forest, Gradient Boosting, XGBoost, Support Vector Regression), and deep learning (Long Short-Term Memory). Model selection was performed using a walk-forward, expanding-window out-of-sample validation for 10 years (2014-2023) based on RMSE, MAE, and MAPE criteria, and the best-selected model was re-estimated on the full series to forecast MMR through 2035. Between 1980 and 2023, MMR decreased from 674.5 to 396.2 deaths per 100,000 live births (EAPC 2013-2023: -2.17%/year). The lowest and nearly identical RMSEs and MAPEs were obtained for ARIMA and Holt’s damped-trend exponential smoothing (RMSE = 2.74 and 2.75; MAPE = 0.55% and 0.52%), clearly outperforming all four ML models, especially LSTM (RMSE = 388.33, MAPE = 88.52%). The best-fitting ARIMA model predicts MMR to rise to 344.7 (95% confidence interval [CI]: 301.7 – 387.7) by 2030 and 313.0 (95% CI: 231.6 – 394.4) by 2035, or about five times higher than the SDG 3.1 target. The classical statistical approach was clearly better than the ML and DL approaches in the case of short annual time series, and Sub-Saharan Africa will miss the SDG 3.1 target by 2030 and beyond 2035.