QSAR-Models, Validation, and IIC-Paradox for Drug Toxicity

Abstract

Three kinds of drug toxicities are examined in this modeling analysis. These are: (i) toxicity of psychotropic drugs; (ii) cardiac toxicity; and (iii) drug carcinogenicity. Predictive models for the toxicity data are built up by the Monte Carlo technique. The simplified molecular input-line entry system (SMILES) is used for the representation of the molecular structure. Quantitative structure – activity relationships (QSAR) developed here are mathematical functions of corresponding SMILES. The index of ideality of correlation was tested as a tool to improve predictive potential of these models.