An introduction to machine learning tools for the analysis of microplastics in complex matrices

Abstract

As microplastic (MP) particles continue to spread globally, their pervasive presence is increasingly problematic. Analyzing MPs in matrices as varied as soil, river water, and biosolid fertilizers is critical, as these matrices directly impact the food sources of plants, animals, and humans. Current analytical methods for quantifying and identifying MPs are limited due to labor-intensive extraction processes and the time and effort required for counting and analysis. Recently, Machine Learning (ML) has been introduced to the analysis of MPs in complex matrices, significantly reducing the need for extensive extraction and increasing analysis speeds. This work aims to illuminate various ML techniques for new researchers entering this field. It highlights numerous examples in the application of these models, with a particular focus on spectroscopic techniques such as infrared and Raman spectroscopy; tools which are used to quantify and identify MPs in complex matrices. By demonstrating the effectiveness of these computer-based tools alongside the hands-on techniques currently used in the field, we are confident that these ML methodologies will soon become integral to all aspects of microplastic analysis in the environmental sciences.