Show simple item record

dc.contributor.authorTan, Xiaoyan
dc.description.abstractConsidering traditional animal-based testing approaches are expensive and time-consuming, the application of in vitro testing to predict patterns of toxicity was highlighted, ToxCast program was launched by EPA in 2007 with a large number of toxicology information collected from widely dispersed sources. Using the publicly available database, this analysis was performed to develop predictive toxicology models to discriminate between chemicals with reproductive toxicity and those without through examining the association between in vitro assays and in vivo lowest effect level. Three models were built to predict LEL concentration of chemicals based on in vitro measurements, which includes Support vector machine, logistic model, classification and regression tree, and cross validation was used to test the accuracy of models. Except for logistic model failed to give a clear result, SVM model and CART model showed a good prediction with the index of the area under a ROC curve.
dc.subjectToxCast, SVM, logistic regression, CART
dc.titleAnalysis of toxcast phase ii data-predicting in vivo toxicities from in vitro data using optimal models
dc.description.departmentPublic Health
dc.description.advisorHanwen Huang
dc.description.committeeHanwen Huang
dc.description.committeeStephen Rathbun
dc.description.committeeKeivin Dobbin

Files in this item


There are no files associated with this item.

This item appears in the following Collection(s)

Show simple item record