A study published in Biochar X on 17 August 2026 by Wenmei Tao's team at Kunming University of Science & Technology shows that elemental descriptors predict biochar radical properties and reveal compositional patterns associated with radical accumulation and oxygen-centered radical formation. Biochar's persistent free radicals can promote pollutant removal but can also generate reactive oxygen species linked to oxidative stress and disruption of soil microorganisms. The researchers collected 263 paired records of radical concentration and g-Factor from published studies, using carbon, hydrogen, oxygen, hydrogen-to-carbon ratio, and oxygen-to-carbon ratio as model inputs, with data randomly divided into 80% for training and 20% for testing. Six approaches were compared: XGBoost, gradient boosting, support vector regression, a shallow neural network, random forest, and an ensemble averaging predictions from four models. The shallow neural network achieved the highest test R² values, reaching 0.7797 for radical concentration and 0.7647 for g-Factor, with areas under receiver operating characteristic curves of at least 0.9600 for concentration and above 0.9780 for g-Factor. Model interpretation associated lower hydrogen-to-carbon ratios and lower oxygen contents with greater radical accumulation, whereas higher oxygen contents and oxygen-to-carbon ratios favored higher g-Factors. Laboratory correlations supported these rankings, with oxygen content and oxygen-to-carbon ratio showing Spearman correlations of 0.905 and 0.882 with g-Factor.