AI for Monitoring, Risk Assessment, and Mitigation of Pharmaceutical Pollution in Agricultural Systems

Pharmaceuticals and antibiotics are increasingly recognized as emerging pollutants in agriculture, introduced through veterinary drug use, land application of sewage sludge and manure, and irrigation with treated wastewater. These inputs can cause contamination of soils, surface waters, and crops, with implications for soil health, food safety, and antimicrobial resistance (AMR). Conventional monitoring is limited because of sparse sampling, difficulty in capturing the complex, non-linear processes governing contaminant fate across the water–soil–crop system. This review synthesizes recent advances in artificial intelligence (AI) for detecting, modeling, and mitigating pharmaceutical pollution in agriculture. AI-assisted LC/HRMS improves screening and monitoring by predicting retention time and collision cross-section. Additionally, IoT-enabled sensing with learning-based signal processing supports high-frequency surveillance of reclaimed irrigation water. Machine-learning models provide compound- and site-specific predictions of persistence and mobility by linking molecular descriptors with soil properties, hydrology, and management. Trained ML models, including LARS and lightweight OMP, resulted in very high accuracy, demonstrating AI’s ability to estimate physicochemical properties of pharmaceuticals. AI also optimizes advanced oxidation, photocatalytic, and biological treatments, with reported removal efficiencies greater than 95%. For risk assessment and AMR surveillance, multimodal models integrate contaminant loads, environmental quality metrics, metagenomic resistance markers, and management indicators to flag high-risk reuse scenarios. Key needs include harmonized datasets, transferable models, and interpretable, regulatory-ready AI frameworks.