Kean Researcher Develops AI Tool to Help Make Drug Discovery Safer and More Efficient
Kean University Associate Professor Supratik Kar, Ph.D., has helped develop an artificial intelligence (AI) tool that can predict whether potential medicines may damage the kidneys, giving researchers a new way to identify safety concerns before compounds reach laboratory or clinical testing.
Kar co-developed KidneyTox_v1.0 (KidneyTox) with colleagues from the University of Salerno in Italy. The web-based platform uses AI to predict kidney toxicity in small molecules while explaining why a molecule is predicted to be toxic or non-toxic. The project, published earlier this year in Scientific Reports, enables researchers, students and scientists to evaluate compounds at no cost.
Kidney toxicity remains one of the leading reasons promising drug candidates fail during development. Traditional testing requires costly laboratory and animal studies before drug candidates can advance to clinical trials, a process that can take years. The platform gives researchers an early screening tool to identify potential safety concerns before those expensive and time-consuming stages begin.
"Most artificial intelligence tools simply tell you whether a molecule is toxic or not," Kar said. "We wanted to go further by showing scientists which part of the chemical structure is responsible for that prediction. If a molecule appears toxic, researchers can potentially redesign it before investing years of work and significant resources."
Kar and his colleagues built KidneyTox using a curated dataset of 565 chemically diverse FDA-approved small molecules, including nearly 300 drugs known to cause kidney toxicity as well as compounds considered safe. The model combines AI with user-friendly molecular descriptors and statistical performance metrics designed to help researchers understand and evaluate its predictions.
"We're not asking scientists to blindly trust the prediction," Kar said. "The platform explains the reasoning behind each result and tells users when a molecule falls outside the chemical space the model understands. That transparency makes the predictions much more useful for research."
Researchers worldwide have already begun using KidneyTox to evaluate compounds before experimental testing. The platform builds on Kar’s previous work developing AI tools to predict liver toxicity and is another step toward his long-term goal of creating a comprehensive resource capable of evaluating multiple forms of organ toxicity.
"We want to create a comprehensive tool that can evaluate all major organ toxicities for a single molecule as well as batches of molecules with a single click," he said. "By keeping every platform open source, we hope researchers everywhere can benefit from these resources and ultimately help design safer medicines for patients."
Legal Disclaimer:
EIN Presswire provides this news content "as is" without warranty of any kind. We do not accept any responsibility or liability for the accuracy, content, images, videos, licenses, completeness, legality, or reliability of the information contained in this article. If you have any complaints or copyright issues related to this article, kindly contact the author above.
