All Times EDT
Keywords: decision tree classfication, cardic safety
The Comprehensive in vitro proarrhythmia assay initiative (CiPA) is aiming to incorporate the mechanistic understanding into a new regulatory paradigm for better predicting drugs’ proarrhythmic risk and reduce the need for the thorough QT (TQT) study. The CiPA steering Committee proposed using 28 selected drugs with fairly known mechanisms and torsade de pointes (TdP) risks in human for model and metric selection and optimization.
However, there are still challenges for using in vitro assays. For example, the human induced pluripotent stem cell-derived cardiomyocytes (hiPSC-CMs) has been studied as an prediction of risk level and showed a reasonable but not high accuracy. Categorizing of drug TdP risks were based on marginal categorizations of individual measurements collected in a CiPA study. As the later stage with increased knowledge and experiences about the developing assay, statistical classification model like regression tree could be used to improve the process. This poster will present the uses of tree based categorization approaches, compare different models, and make recommendations.