Evaluating the eye Irritation potential of a chemical is a necessary procedure in the risk assessment. We are exposed to various eye irritants in cosmetics, ocular pharmaceuticals, household products, manufacturing industries and etc. Draize test was used a standard protocol to evaluate the eye irritant potential of a chemical. However, It has several limitations such as highly cost and time-comsuming and cruelty of using rabbit’s tissue. Therefore, the QSAR model for evaluating eye irritation can be a solution to these problems.

The data used in the development of QSAR model is MMAS(Modified Maximum Average Score) data according to OECD Guideline 405, and vapor pressure data, collected from QSAR Toolbox 4.2, PhysProp database, eChemPortal database and several literatures. All data consist of 60% of ther training set for the model learning by selecting the most diverse structure set, and 40% of external sets.

In this study, a QSAR prediction model was constructed by using multiple linear regression(MLR) and support vector machine(SVM) methods and all models were verified using y-scrambling and external validation. Through these studies, we have analyzed the substructures(structure alerts) that can cause eye irritation. This predictive model is expected to be useful for development of cosmetic materials and drug screening research.

※ Acknowledgement