IEEE Computer Society Newsfeed

Subscribe RSS

« Back

New Computational Method Determines Chemical Toxicity

Researchers from the University of Kansas have devised a new computational method for predicting whether a chemical compound will be toxic. Regulatory constraints are exerting pressure on the chemical industries and other firms that need to comply with safety regulations regarding the handling and disposal of dangerous chemicals. The method would enable users to predict if a compound would be toxic, even at a low dose, allowing them to make more informed decisions regarding a compound’s use, disposal, and/or recycling, especially if it can adversely affect human health. The researchers created a statistical algorithm based on a combination of Random Forest selection with Naïve Bayes statistical analysis. They tested the algorithm against some 300 chemicals for which the toxicity is known, building on information about relationships between compounds derived from the pharmaceutical industry. They successfully predicted toxicity in every two out of three chemicals tested. The researchers say their computer-based method is able to very quickly screen a large number of compounds for obvious toxicity and may eliminate animal testing. It may also allow regulators to focus on those compounds with the greatest potential toxicity. The work will be published in a forthcoming issue of the International Journal of Data Mining and Bioinformatics. (EurekAlert)(InderScience Blog) 

Trackback URL: