A collaboration between Johns Hopkins University and the science safety company UL has led to an important discovery: computers are superior to animal models in detecting toxic chemicals.
Not only is the software more effective than animal testing in predicting chemical toxicity, it is also quicker and less expensive than traditional animal tests.
Over 100,000 chemicals are found in consumer products and medications, and for most, little is known about their toxicity. To overcome this gap in knowledge, researchers at Johns Hopkins created a database of toxicity studies and collaborated with UL to expand the database to include over 70 million chemical structures with tens of thousands of additional biological and animal data points.
The software developed by UL can use this existing information to predict the toxicity of untested chemicals by quickly finding chemicals with similar properties from the database, as they may have similar effects on human health and the environment. Using this approach, toxicity results were reproducible in up to 84% of cases, and the software was able to identify 92% of toxic substances, outperforming animal testing in both cases.
According to UL President of Supply Chain and Sustainability Carlos Correia, “Though we have to wait for a formal validation, this finding signals the beginning of the end to animal testing, starting with these very common acute effects.”
Dr. Thomas Hartung, Chair for Evidence-based Toxicology at Johns Hopkins University, also sees the extraordinary value of the technology.
“Nothing is as strong as an idea whose time has come,” notes Hartung. “This promises to make half of animal testing for safety superfluous.”
Some U.S. agencies have already started using the software. We look forward to the potential this technology will have on reducing animal use for the detection of toxic chemicals.
Source: “Computers outperform lab rats in detecting toxic chemicals,” EurekAlert, February 16, 2018