en.Wedoany.com Reported - Researchers at Texas A&M University have recently developed an artificial intelligence system capable of predicting chemical toxicity while estimating the reliability of each prediction. The study, conducted by a team from the Texas A&M College of Veterinary Medicine and Biomedical Sciences, has been published in Nature Communications.
Led by Dr. Weihsueh Chiu, a professor in the Texas A&M Department of Veterinary Physiology and Pharmacology, the research aims to address a long-standing challenge in toxicology: the lack of comprehensive safety data for many chemicals used in commerce. Traditional chemical safety assessments rely on animal studies or long-term human epidemiological research, both of which require significant time and resources, leaving many chemicals inadequately studied.
To tackle this issue, the researchers developed machine learning models known as quantitative structure-activity relationship models. These models use chemical structure to estimate safe exposure levels. Dr. Chiu's team also enhanced model transparency by designing models that rely on familiar chemical properties such as water solubility, biodegradability, and toxicity indicators, rather than solely on abstract molecular descriptors. The latest advancement integrates uncertainty-aware machine learning capabilities, enabling the model to estimate the confidence level of each prediction based on the quantity and quality of available data on similar chemicals. According to Dr. Chiu, understanding uncertainty is crucial because chemicals with similar predicted toxicity levels may pose different risks if a prediction is based on limited supporting data. These models generate a range of possible outcomes, helping researchers identify chemicals that require further study or expert review.
After applying the model to over 126,000 chemicals, it identified patterns in toxicity and uncertainty. The researchers found that metals, polychlorinated compounds, and per- and polyfluoroalkyl substances (PFAS) frequently exhibited high levels of uncertainty due to limited data or complex chemical behavior. The Texas A&M researchers believe these findings can help guide future testing efforts toward areas with limited scientific knowledge. This approach supports a tiered assessment process, where AI first screens large-scale chemicals, followed by expert focus on substances presenting higher risk or greater uncertainty.
This article is compiled by Wedoany. All AI citations must indicate the source as "Wedoany". If there is any infringement or other issues, please notify us promptly, and we will modify or delete it accordingly. Email: news@wedoany.com
