SeminarsOptimal Composite Markers for Time-Dependent ROC
224
reads
To significantly improve the accuracy of evaluation on the coronary artery disease (CAD) related vital status over time, our research interest mainly focuses on finding a combination of several baseline plasma numbers with the highest time-dependent receiver operating characteristic (ROC) curve. In this study, we consider a time-varying coefficient generalized linear model (GLM) with an unknown link function for the CAD-related death time. Based on censored survival data, a class of nonparametric estimation methods is proposed for the optimal linear combination of plasma biomarkers under the random censoring and conditionally independent censoring mechanisms.Moreover, the consistency and asymptotic normality of the estimators are derived. The finite sample properties of the proposed estimators are further examined through simulations. Finally, an application of our methods to the British Columbia Vital Statistics data is presented.
reads
Hung Hung
2008-02-19
13:30:00 - 15:00:00
405 , Mathematics Research Center Building (ori. New Math. Bldg.)
To significantly improve the accuracy of evaluation on the coronary artery disease (CAD) related vital status over time, our research interest mainly focuses on finding a combination of several baseline plasma numbers with the highest time-dependent receiver operating characteristic (ROC) curve. In this study, we consider a time-varying coefficient generalized linear model (GLM) with an unknown link function for the CAD-related death time. Based on censored survival data, a class of nonparametric estimation methods is proposed for the optimal linear combination of plasma biomarkers under the random censoring and conditionally independent censoring mechanisms.Moreover, the consistency and asymptotic normality of the estimators are derived. The finite sample properties of the proposed estimators are further examined through simulations. Finally, an application of our methods to the British Columbia Vital Statistics data is presented.