Every year around 10 millions of animals are used to assess compound’s safety, with a total cost estimated to 3 billions euro. Despite animal testing is mandatory by OECD to guarantee consumer safety, toxicologists are working on the development of alternative testing methods to solve this ethic and economic concern. Thomas Hartung and his team from Johns Hopkins University in Baltimore, Maryland, are working on alternative toxicity methods to improve public health. In a recent paper published in Toxicological Sciences, Hartung’s team explained how toxicological data set from the European Chemical Agency could be used to predict compounds toxicity using machine-learning. This method name RASARs for “read-across structure activity relationship” uses binary fingerprints and Jaccard distance to define chemical similarity and predict compounds toxicity. This method showed balanced accuracies in the 80%-95% range across 9 health hazards. This new efficient approach in toxicity testing draws attention to the new possibility of big data to replace animal testing in the future.

Link to the article : Machine Learning of Toxicological Big Data Enables Read-Across Structure Activity Relationships (RASAR) Outperforming Animal Test Reproducibility


Leave a Reply