結果依賴採樣之多維度存活資料在不同相關矩陣中的設計最優化和篩選準則

In this grant, our goal is to derive estimations under various working correlation structures for multivariate failure times sampled from a general multivariate-outcome-dependent sampling (MODS) design under the framework of the semiparametric accelerated failure time (AFT) model and try to answer a much needed call for cost-efficient studies. We developed the optimal allocation of the supplemental samples under varying MODS designs while easing the commonly encountered budget constraints in the real studies.