Towards Sustainable Health: A Time Series Analysis of Health-Related Sustainable Development Goal Indicators (SDG 3, 6) for Bangladesh Using ARIMA and Machine Learning Models

Abstract IntroductionBangladesh’s progress toward the Sustainable Development Goals (SDGs) remains closely tied to health and environmental outcomes, particularly SDG 3 (Good Health and Well-being) and SDG 6 (Clean Water and Sanitation). Despite substantial advancements in recent decades, persistent challenges threaten continued progress. This study forecasts key health and environmental indicators to assess Bangladesh’s trajectory toward the 2025 and 2030 targets, providing evidence to guide policy and action.MethodsThis secondary data-based observational study analyzed yearly estimates (2000–2024) of 14 SDG indicators for Bangladesh. Time series analyses were performed using ARIMA models and machine learning techniques (Linear Regression, Random Forest, XGBoost). Model performances were evaluated using mean absolute error (MAE), root mean squared error (RMSE), and mean absolute scaled error (MASE) between ARIMA and ML models and Diebold Mariano Test (α = 0.05) was used for model comparisons among the machine learning models. Forecasts were compared with SDG targets for 2025 and 2030. Data were obtained from the Sustainable Development Report and the Bangladesh SDG Tracker.ResultsBased on forecasted trends, indicators fall into three categories. Good: Targets (2025 and 2030) achieved or on track, requiring only continued efforts (MMR, under-5 mortality). Fair: Slightly behind but achievable with minor adjustments (neonatal mortality, vaccination, safe water, adolescent fertility, water consumption). Bad: Persistently poor with no improvement, needing urgent action (NCD mortality, road traffic mortality, sanitation, handwashing, well-being, water quality). Machine learning models corroborate these findings, emphasizing the urgency and want of further working in these areas.ConclusionThe comparative analysis reveals that no single model is universally superior; Linear Regression excels for strong linear trends, ARIMA for well-structured time series, and XGBoost for volatile, non-linear indicators, providing an evidence-based framework for optimal model selection in SDG forecasting. Bangladesh is on track to achieve some SDG 3 and 6 targets, but critical gaps remain in non-communicable diseases, road safety, adolescent health, sanitation, and water quality. Implementation of strategically planned policy and continuous monitoring are essential to overcome the challenges and achieve the SDG goals by 2030.