At the 21st Annual System of Systems Engineering (SoSE) Conference 2026, held in Kongsberg, Norway, Abbass Chreim from the University of Lille presented the research paper "A Machine Learning-Based Framework for Early Detection of Water Quality Incidents in River Systems of Systems."
The research, carried out within the framework of the BEPROACT Interreg NWE project, presents a vision for large-scale water quality monitoring using advanced machine learning techniques. The proposed framework incorporates an adapted Long Short-Term Memory (LSTM) Mixture of Experts (MoE) architecture, specifically designed for spatio-temporal pattern recognition.
By analysing complex patterns in water quality data, the framework aims to predict dissolved oxygen levels and provide early warning signals for conditions that may lead to critical fish mortality events. Such predictive capabilities can support infrastructure owners and environmental authorities in taking proactive measures to protect aquatic ecosystems.
During his presentation, Abbass highlighted the close collaboration between the University of Lille and the Flemish Environment Agency (VMM) within the BEPROACT project. The research was co-authored by Simon Nachtergaele and Andy Louwyck (VMM), together with Prof. Rochdi Merzouki, demonstrating how cross-border collaboration brings together expertise in artificial intelligence, environmental monitoring, and infrastructure resilience.
The presentation generated valuable discussions with researchers and practitioners working on Systems of Systems Engineering and AI-driven solutions for complex societal challenges. Conferences such as SoSE provide an important platform for exchanging knowledge, fostering new collaborations, and accelerating innovation across disciplines.
BEPROACT congratulates Abbass, together with his co-authors Prof. Rochdi Merzouki, Simon Nachtergaele and Andy Louwyck (VMM), on presenting this important research and contributing to the growing body of knowledge on smart, data-driven environmental monitoring.