A systematic review on machine learning deep learning and IoT for water quality prediction

Abstract Water quality degradation driven by population growth, industrialization, and climate change poses significant risks to public health and aquatic ecosystems, necessitating accurate and timely prediction frameworks. This systematic review synthesizes recent advances in Water Quality Prediction (WQP) using Machine Learning (ML), Deep Learning (DL), and Internet of Things (IoT) technologies. Following PRISMA guidelines, 289 records published between 2020 and 2025 were retrieved from IEEE Xplore, ScienceDirect, MDPI, and Google Scholar, of which 81 studies met the inclusion criteria. The reviewed literature spans diverse datasets from rivers, lakes, reservoirs, and drinking water systems, with predicted parameters including dissolved oxygen, pH, turbidity, biochemical oxygen demand, and water quality indices. In studies, most models report high predictive performance, although the performance metrics vary across datasets and environmental contexts. DL architecture is increasingly adopted for spatiotemporal modeling and large-scale data integration. IoT-enabled systems are employed for real-time monitoring and early warning, in conjunction with ML/DL models. This study emphasizes the role of ML, DL and IoT in supporting sustainable water resource management by enabling accurate real time WQP and contributing to global sustainability goals including SDG 6 Clean Water and Sanitation. Future research should focus on Edge AI, Federated Learning and explainable AI, models to improve secure, reliable, adaptive, and scalable WQP systems.