Machine Learning of Toxicological Big Data for Read-Across Structural Activity Relationships (RASAR) to Develop Chemical Hazard Data
Industry is currently facing a substantial challenge. There is an increasing need to develop chemical hazard data for both new and existing chemicals, while at the same time there are current and proposed regulatory mandates as well as ethical drivers to reduce, refine, and replace (the “3 R’s”) the use of animal test methods with alternative non-animal methodologies. This creates a challenge for innovators, formulators, and companies who strive to introduce a new chemical or fill data gaps on existing chemicals while adhering to the principles of the 3 R’s.
In silico, or computational toxicology software offers an alternative method for predicting chemical hazards. Traditional Quantitative Structure Activity Relationship (QSAR) methodology employs simple analog identification and predicting hazard data from chemical analogs with known hazard data. This webinar will cover the UL Cheminformatics Tool Kit, a novel in silico approach combining big data with advanced machine learning incorporating data fusion. The approach integrates novel models called RASARs (Read-across Structure Activity Relationship) with data fusion techniques that extend this concept by considering all available property data on both the target and analogs rather than only the modeled hazard. This increases balanced accuracies of predictions by 10% on average across assessed hazard endpoints. Applications of predicted data to inform research and development, fill data gaps, and foster green chemistry will be illustrated.
Presenter(s): Stacie Abraham
Stacie Abraham, Senior Regulatory Specialist, UL Product Supply Chain Intelligence, has 33 years of combined chemistry and diverse industry regulatory experience. Stacie is an expert in global chemical control regulations including the U.S. Toxic Substances Control Act (TSCA), European REACH, and others. Her expertise also includes the Globally Harmonized System of Classification and Labelling (GHS), toxicology data analysis, applications of computational toxicology data, and regulatory compliance consulting and program design. Stacie holds a B.S. degree in Chemistry, Summa Cum Laude, from the University of Pittsburgh.