Water Quality Monitoring and Assessment Using Machine Learning: A Review of Formulation, Modeling Approaches, and Explainable Artificial Intelligence

Water pollution poses significant risks to human health and environmental sustainability, highlighting the need for accurate water quality assessment and prediction. This review examines the application of machine learning (ML) in Water Quality Index (WQI) assessments, focusing on WQI formulation, predictive modelling approaches, and explainable artificial intelligence (XAI) techniques. A structured literature review is conducted using major scientific databases, including ScienceDirect, Springer, and other relevant sources, following a systematic study selection process. The review analyzes commonly used water quality parameters and highlights how the deterministic structure of WQI influences machine learning modelling, often leading to high predictive performance that reflects predefined formulations rather than independent pattern learning. A comprehensive comparison of single, hybrid, and ensemble ML models is presented, showing that hybrid approaches generally provide improved robustness and accuracy in complex water quality scenarios. In addition, the role of XAI methods in enhancing model interpretability and supporting transparent decision-making is discussed. Key challenges, including limited generalization, model complexity, and interpretability constraints, are identified, and future research directions are proposed to develop more reliable and practical AI-based water quality monitoring systems. Overall, this review provides insights into the integration of machine learning and WQI, emphasizing the importance of balancing predictive accuracy with interpretability for sustainable water resource management.