Multi-objective Optimization of Sampling Locations for Source Detection of Antibiotic-resistant Genes in Hydrosystems

Antimicrobial resistance is a growing global challenge with significant implications for public health. Wastewater surveillance offers a promising approach to monitoring and predicting the dissemination of antibiotic-resistant bacteria and genes (ARGs) in systems like sewer systems. However, current studies often lack integration between the optimization of sampling locations and the dynamic updating of predicted source locations based on continuous measurements. This paper proposes a novel data-driven framework that bridges this gap by combining multi-objective optimization for selecting optimal sampling locations with Bayesian updating for probabilistic source detection. The framework accounts for DNA degradation, sewage dilution and measurement variability to realistically simulate ARG concentrations in hydrosystems. Through iterative updates guided by detected signals at selected sampling locations, the model refines the likelihood of each sub-catchment being the main source of multiple ARGs, enabling accurate source localization. Validation was conducted using two real-world sewer systems under varying structural and sampling constraints. Across 1,000 simulated ARG scenarios and different sampler limits, results show that using only 3 and 5 samplers, the framework can achieve over 80% detection accuracy within four iterations; using 7 and 12 samplers raises accuracy to above 90%, depending on the complexity of the networks. These findings demonstrate the scalability, robustness, and practical applicability of the framework for ARG monitoring and decision support in diverse wastewater systems.