Air Quality Prediction and Public Health Risk Assessment Using Machine Learning: A Case Study of Dhaka and Rajshahi in Bangladesh

Air pollution is a major public health threat in Bangladesh, particularly in rapidly urbanizing cities like Dhaka and Rajshahi. Dhaka consistently ranks among the world’s most polluted cities due to unregulated industrial emissions, heavy vehicular traffic, and widespread biomass burning. This chapter provides a comprehensive analysis of urban air quality using advanced machine learning models—SARIMA, XGBoost, Random Forest, and Long Short-Term Memory (LSTM)—to forecast PM2.5, PM10, and CO concentrations through 2030. Among the models evaluated, Random Forest and XGBoost demonstrated superior temporal prediction accuracy. Forecasts indicate that PM2.5 concentrations may rise to 56.27 µg/m3 in Dhaka and 33.33 µg/m3 in Rajshahi, far exceeding the WHO guideline of 5 µg/m3. PM10 levels are projected to reach 143.90 µg/m3 in Dhaka and 132.39 µg/m3 in Rajshahi, surpassing the 15 µg/m3 WHO guideline. CO concentrations are expected to climb to 10.605 ppm in Dhaka and 7.964 ppm in Rajshahi, posing significant cardiovascular and neurological risks. By integrating these forecasts with exposure-response health risk models, the study estimates a 5–7% increase in cardiopulmonary morbidity and premature mortality in Dhaka and a 3–5% increase in Rajshahi by 2030, with children and the elderly facing the greatest risk. The findings underscore the urgent need for evidence-based air quality management, real-time monitoring, and comprehensive public health strategies. This chapter presents an interdisciplinary framework for predictive environmental health modeling, offering a replicable approach for assessing urban air pollution in other low- and middle-income countries.