Online Integrated Testing Strategy (ITS) for quantitative skin sensitization

Online Integrated Testing Strategy (ITS) for quantitative skin sensitization potency assessment using Bayesian networks and accounting for bioavailability


Barry Hardy


Douglas Connect


Ahmed Abdelaziz, Daniel Bachler, Maja Brajnik, Thomas Exner, Barry Hardy


Jaworska, J. S.; Natsch, A.; Ryan, C.; Strickland, J.; Ashikaga, T.; Miyazawa, M. Bayesian Integrated Testing Strategy (ITS) for Skin Sensitization Potency Assessment: A Decision Support System for Quantitative Weight of Evidence and Adaptive Testing Strategy. Arch. Toxicol. 2015, 89, 2355–2383.


The public health problem of skin sensitization is attributed to high direct and indirect costs. Many in vitro tests are available for evaluating the hazard of chemical skin sensitization aiming to replace and reduce the in vivo animal test (Local Lymph Node Assay - LLNA). Strategies for integrating such tests, together with in silico calculations of chemical bioavailability, have been proposed in the literature. The integrated testing strategy for skin sensitization (ITS-3) [1] is one such approach. It utilizes Bayesian networks for predicting skin sensitization hazard and guides the testing process. In this work, we reproduce the results from the ITS-3 article and develop a web-application (DC SkinSens) that guides the user through the application of the ITS-3 strategy to assess chemical potency of skin sensitization potency (LLNA pEC3). Open and free tools were favoured during the course of this work to ensure reproducibility and encourage adoption by regulators and industry. The Bayesian network was built and trained using the statistical package R. The overall network accuracy for 4-category classification (non-, weak, moderate or strong sensitizers) was 80% while the accuracy of the most confident predictions was 98%. The web application is freely accessible online on