Revisiting the detection, fate, and health risks of microplastics in the environment through artificial intelligence

Against the backdrop of global change, the widespread presence of emerging contaminants (ECs) like microplastics (MPs), coupled with the escalating impacts of climate change, collectively poses a dual threat to global ecosystems and human health. The complexity, nonlinearity, and multi-scale nature of this 'triple planetary crisis' pose significant challenges to traditional environmental research methodologies. Artificial intelligence (AI), particularly deep learning technology, is leading a paradigm shift in environmental research due to its exceptional capabilities in processing high-dimensional data, recognizing complex patterns and performing causal inference. This article systematically elaborates on the transformative role of AI across the entire MPs research chain. At the detection stage, AI-driven high-throughput spectroscopy and microfluidic technologies enable rapid, precise identification and in-situ monitoring of low-concentration, small-sized MPs. In behavioral simulation, interpretable machine learning reveals the transport mechanisms of MPs in porous media, their shape-dependent settling behavior, and the vicious cycle of feedback with climate warming. In risk assessment, AI integrates multi-source data, showing great potential for elucidating the reproductive and neurotoxic mechanisms of MPs (e.g., ferroptosis), assessing their global distribution patterns and ecological risks, and quantifying human exposure (especially in pregnant women and children) via the food chain. We propose an "Pan-microplastic AI Framework" framework that extends beyond the scope of MPs. This framework uses AI to systematically address the combined crisis of emerging contaminants (ECs), climate change, and biodiversity loss. It employs methods such as multi-modal data fusion, causal discovery, and generative simulation to support "One Health" governance, enabling both macro-policy optimization and micro-precise interventions. Despite challenges such as data quality, model interpretability, and cross-scale integration, the deep integration of environmental science and AI is undoubtedly the necessary path to providing quantitative, systematic, and intelligent solutions for building a planetary health framework and achieving harmonious coexistence between humanity and nature.