Prediction Models

Home > Prediction Models


Because quality of predictive models depend on the underlying quality (reliability and relevance) and the completeness of the data on which the model is built, a nano-specific data gap filling method and a PChem score-based screening scheme have been proposed and evaluated for data quality and completeness of the S2NANO database.

The relationships between the physicochemical properties of nanomaterials and their various cytotoxicity endpoints are determined by performing S2NANO data collection and pre-processing procedure using Random Forest, Backpropagation, Support Vector Machine, Resilient Backpropagation, and Quasi-QSAR algorithms through in silico approach.