New research shows deep learning can predict how quickly biochar materials break down antibiotic contaminants, providing a faster path to cleaner water and smarter environmental remediation.
Antibiotic contamination is a growing environmental and public health concern. These compounds can enter rivers, groundwater, sewage systems, and agricultural environments and persist, impacting aquatic organisms and contributing to the spread of antibiotic resistance. Biochar, a carbon-rich material made from biomass, has shown promise as a sustainable catalyst for the degradation of antibiotics. However, it remains difficult to design suitable biochars for appropriate treatment systems, as many factors simultaneously affect the performance.
The research team has now developed an interpretable artificial intelligence framework that can predict the kinetics of antibiotic degradation in biochar catalytic systems. The study, published in Biochar, combines environmental chemistry, materials science, and deep learning to identify which biochar properties and reaction conditions are most important.
“Biochar-based catalysts are very promising, but their performance is controlled by a complex interplay between pore structure, surface chemistry, residual free radicals, oxidant dosage, and contaminant concentrations,” said the corresponding author. “Our goal was to build a practical AI tool that could not only predict the kinetics of degradation, but also explain why certain systems perform better than others.”
The research team compiled a comprehensive dataset from 75 peer-reviewed studies covering multiple antibiotic classes, including tetracyclines, fluoroquinolones, and sulfonamides. They evaluated 16 input features across three main categories: biochar catalyst properties, elemental composition, and reaction conditions. Six machine learning models were tested, including Random Forest, XGBoost, LightGBM, Support Vector Regression, Multilayer Perceptron, and TabPFN, a transformer-based deep learning model designed for tabular data.
TabPFN achieved the strongest prediction performance, with a test R² of 0.91 and a root mean square error of 0.021. This performance exceeds that of tree-based, kernel-based, and traditional neural network models, demonstrating the ability of transformer-based learning to handle small but complex environmental datasets.
The model went beyond prediction and revealed important mechanistic insights. Catalyst properties contributed 59.3% of the model’s predictive power, followed by reaction conditions with 25.9% and elemental composition with 14.8%. The most influential factors include residual free radicals, total pore volume, oxidant concentration, pollutant concentration, graphite structure, average pore size, biochar dosage, and Raman ID/IG ratio.
This analysis suggests that persistent free radical-rich biochar produced at approximately 450–550 °C may promote the production of reactive oxygen species and enhance the degradation of antibiotics. A total pore volume greater than 0.23 cm3 g-1 is also associated with stronger catalytic performance. This is probably because the increased porosity enhances contaminant adsorption, oxidant transport, and access to active sites.
The study also identified the actual operating window. Moderate oxidant concentrations of approximately 0.5–5.5 mg L−1 improved the degradation, but excess oxidant may reduce efficiency due to radical scavenging. Lower contaminant concentrations, especially below 22 mg L−1, were associated with faster degradation because more active sites remained available.
To support real-world use, the researchers embedded the model in a user-friendly web-based graphical interface. Users can input catalyst properties, elemental composition, and reaction conditions to estimate antibiotic degradation rates. In external validation, the tool predicted the performance of new biochar catalysts with less than 20% error.
“This framework helps researchers screen biochar catalysts before conducting large-scale experiments,” the authors said. “This provides a faster, more explainable and more cost-effective route to optimizing antibiotic-contaminated water treatment systems.”
The findings demonstrate how interpretable AI can move environmental remediation from trial-and-error testing to data-driven catalyst design. This study provides a general strategy to improve biochar-based technologies and other complex catalytic systems used for pollution control by coupling prediction with mechanistic understanding.
===
Reference: Latif, J., Chen, N., Xie, J. et al. Deep learning-based prediction and mechanistic analysis of reaction kinetics in biochar-catalyzed antibiotic degradation. Biochar 8, 88 (2026).
https://doi.org/10.1007/s42773-026-00606-y
===
About biochar
Biochar (e-ISSN: 2524-7867) is the first journal dedicated to biochar research across agriculture, environmental science, and materials science. He publishes original research on biochar production, processing, and applications such as bioenergy, environmental remediation, soil improvement, climate mitigation, water treatment, and sustainability analysis. The journal serves as an innovative and professional platform for researchers around the world to share advances in this rapidly expanding field.
/Open to the public. This material from the original organization/author may be of a contemporary nature and has been edited for clarity, style, and length. Mirage.News does not take any institutional position or position, and all views, positions, and conclusions expressed herein are those of the authors alone. Read the full text here.
