EPFL · Science

New AI Technique Uncovers Hidden Microplastic Pollution in Swiss Waters

A novel AI-based method developed at EPFL has revealed significantly higher concentrations of microplastics in Lake Geneva and its tributaries than previously estimated, particularly for particles smaller than 100 micrometers.

Microplastics, plastic fragments under five millimeters, are pervasive environmental contaminants that persist for decades. While larger particles are detectable, smaller ones (under 20 microns) have been challenging to characterize, leading to an underestimation of their prevalence.

The new technique, published in Nature Partner Journals Clean Water, uses AI to identify and classify polymer particles as small as five micrometers. This allows for a more accurate assessment of microplastic pollution.

Researchers found that microplastic concentrations between 1 and 100 micrometers in Lake Geneva and surrounding rivers may have been underestimated by over 650-fold.

Analysis of 38 water samples indicated that approximately six exceeded established ecotoxicological thresholds for environmental risk, suggesting a potential concern.

The full impact of these small microplastics on ecosystems is still being studied, but concerns exist about the exposure of the entire food chain and the transport of associated chemicals and additives.

The AI methodology employs two algorithms: one to separate microplastics from biological matter and another to classify them into six major polymer types, improving accuracy over older methods.

Future research aims to track microplastic concentration changes over time in the Geneva watershed and to develop methods for detecting even smaller, nanoscale particles.

AI-samenvatting op basis van de bron.