Artificial intelligence (AI) and machine learning (ML) are increasingly transforming water and wastewater treatment by enabling robust prediction, control, and optimization of complex physicochemical and biological processes that conventional mechanistic models struggle to capture. This review provides a comprehensive and methodologically structured synthesis of AI/ML applications across major treatment operations, including coagulation–flocculation, membrane filtration, adsorption, disinfection byproduct (DBP) formation, and wastewater treatment systems such as water quality monitoring, test design and laboratory scale tests. Using a systematic PRISMA-guided literature selection and scientometric analysis of publications from 1998 to 2025, fifty-five studies were critically evaluated to assess algorithmic trends, performance characteristics, and domain-specific applicability. Traditional ML algorithms (KNN, SVM, RF), deep learning architectures (CNN, RNN, LSTM, NARX), and metaheuristic optimization tools (GA, PSO, GEP) were examined alongside hybrid and ensemble models used to address nonlinear, multivariate water-quality relationships. Results show that AI/ML models consistently outperform empirical and mechanistic baselines in predicting coagulant dosage, membrane fouling, permeate flux, pollutant adsorption efficiency, and DBP formation, while emerging approaches such as soft sensors, and IoT-enabled monitoring are enabling real-time operational decision support. The review highlights future research opportunities in digital twins, physics-informed ML, transfer learning, explainable AI (XAI), graph neural networks (GNNs), and autonomous process control using reinforcement learning. By consolidating algorithmic mechanisms, and application domains, this work provides a rigorous and forward-looking perspective on the role of AI/ML in developing resilient and energy-efficient treatment systems.
Biglarijoo et al. (Mon,) studied this question.