Artificial Intelligence for Air Quality Prediction: A Review of Machine Learning and Deep Learning Applications in Atmospheric Pollution Forecasting

Abstract

Air pollution is a major global environmental and public health challenge, causing respiratory diseases, environmental degradation, and premature mortality. Accurate air quality prediction is essential for environmental management, reducing human exposure to harmful pollutants, and supporting timely mitigation strategies. Recently, Artificial Intelligence (AI) techniques, particularly Machine Learning (ML) and Deep Learning (DL), have emerged as powerful approaches for air quality forecasting because they can effectively model complex nonlinear relationships among environmental variables. This review examines representative ML and DL methods applied to air quality prediction, including Support Vector Machines (SVM), Random Forest (RF), Extreme Gradient Boosting (XGBoost), Artificial Neural Networks (ANN), Convolutional Neural Networks (CNN), Long Short-Term Memory networks (LSTM), hybrid CNN–LSTM models, Transformer-based architectures, and Graph Neural Networks (GNNs). The review compares these approaches based on predictive performance, computational requirements, interpretability, and applicability across different forecasting scenarios. The analysis shows that deep learning models, especially hybrid and attention-based architectures, generally achieve higher predictive accuracy by capturing complex spatial and temporal patterns, while traditional machine learning approaches remain valuable because of their lower computational requirements and suitability for small- and medium-sized datasets. Furthermore, this review discusses major challenges including data quality limitations, model interpretability, computational cost, and spatial generalizability. Future research directions are highlighted, including explainable artificial intelligence, physics-informed learning, federated learning, multi-task prediction, and integration with intelligent monitoring systems for next-generation air quality forecasting.