Raising the bar for submission of machine learning studies in environmental geochemistry and health

Environmental Geochemistry and Health (EGAH) has observed a pronounced and accelerating surge in manuscripts that apply machine learning (ML) techniques to contamination datasets spanning soil, water, biota, and other environmental matrices. Since mid-2022, the journal has received approximately 160 submissions which had ML as a central methodological component, based on title-level identification. The true number is likely higher when considering manuscripts that employ ML without explicitly signalling it in the title. The growth has been steep: from just 3 such submissions in the second half of 2022, to 8 in 2023, 25 in 2024, and 87 in 2025-roughly a 30-fold increase in annual submissions over three years-with a further 35 received in the first five months of 2026 alone. These submissions encompass supervised and unsupervised learning algorithms, including ensemble methods (such as Random Forest and Gradient Boosting), neural network architectures, and support vector machines. We draw a practical distinction between these approaches and classical multivariate statistical techniques such as principal component analysis and multiple linear regression, while recognising that the boundary is not always sharp.