Researchers have developed a machine learning approach that can optimize the treatment of livestock manure and predict how valuable nutrients such as phosphorus are distributed during processing. The breakthrough could help transform agricultural waste into safer and more useful resources, while reducing environmental pollution.
The study, published in Biochar, focuses on hydrothermal treatment, a technology that converts biowaste into a solid material called hydrochar and a nutrient-rich liquid. Unlike traditional methods, hydrothermal treatment does not require drying, works efficiently at different temperatures, and can recycle essential nutrients like phosphorus. Phosphorus is vital for crop growth but is a limited global resource, and its mismanagement often leads to water pollution and ecosystem damage.
"Livestock manure contains large amounts of phosphorus that are both an opportunity and a challenge," said lead author Xiaofei Ge from China Agricultural University. "If released into the environment, it can contaminate water and soil. But if properly recovered, it can be reused as fertilizer to support sustainable agriculture."
To address this challenge, the research team applied machine learning models including XGBoost, Decision Tree, and Random Forest to predict phosphorus behavior in both hydrochar and the liquid phase. They paid special attention to how calcium and iron ions influence phosphorus distribution. Among the models tested, XGBoost provided the most accurate predictions, with excellent agreement between computer forecasts and experimental data.
The study found that reaction time and pH strongly influenced phosphorus recovery, while the addition of calcium and iron promoted phosphorus retention in hydrochar. This process makes phosphorus less likely to leach into water and more suitable for agricultural reuse. Advanced analyses confirmed that as treatment severity increased, the forms of phosphorus in hydrochar became more stable and uniform.
By combining modern artificial intelligence with traditional environmental engineering, the researchers provide a practical tool to guide waste treatment decisions. The approach allows precise control of key conditions, enabling better nutrient recycling and higher-quality hydrochar products.
"Our findings demonstrate that machine learning can help design smarter waste treatment strategies," said co-corresponding author Sabry M. Shaheen from the University of Wuppertal. "This has major implications for sustainable farming, environmental protection, and resource recovery."
The team believes that their results can support policies and technologies for managing livestock waste worldwide, contributing to carbon neutrality goals and reducing the ecological footprint of agriculture.
===
Journal Reference: Ge, X., Zhang, T., Mukherjee, S. et al. Optimizing the conditions of biowastes hydrothermal treatment and predicting phosphorus fate in the hydrochar and liquid phase using machine learning. Biochar 7, 96 (2025). https://doi.org/10.1007/s42773-025-00485-9
===
About Biochar
Biochar is the first journal dedicated exclusively to biochar research, spanning agronomy, environmental science, and materials science. It publishes original studies on biochar production, processing, and applications—such as bioenergy, environmental remediation, soil enhancement, climate mitigation, water treatment, and sustainability analysis. The journal serves as an innovative and professional platform for global researchers to share advances in this rapidly expanding field.
