Kai F. Yu

Conditional estimation following a group sequential clinical trial

Recently, an unbiased estimator Emerson Kittelson 1997 and an essentially unbiased estimator Todd, Whitehead Facey 1996 have been developed to analyze a group sequential clinical trial after it has stopped. However, for these methods, the expectation of the estimator is not close to the parameter value when conditioned on the stopping time. We propose an approach conditioned on the stopping time, which reduces the discrepancy. The result leads to a group of estimators with less discrepancy and sometimes less mean squared error as well. The new estimators are compared by simulation to five existing methods in the case of a two-armed clinical trial with normal response.

An Adaptive Approach to Designing Comparative Diagnostic Accuracy Studies

Comparative diagnostic studies usually involve comparison of the area under receiver operating characteristic curves when biomarkers are measured on a continuous or ordinal scales. In designing such studies, specification of a number of nuisance parameters is often required to compute sample sizes. When these parameters are incorrectly specified, statistical power to detect a meaningful difference in area can be substantially adversely affected. We propose an adaptive method to calculate the sample size and show these procedures to be effective in controlling error rates.

Point and Interval Estimation of Primary and Secondary Parameters in a Two-Stage Adaptive Clinical Trial

Investigated in this paper is the point estimation and confidence intervals of the treatment efficacy parameter and related secondary parameters in a two-stage adaptive trial. Based on the minimal sufficient statistics, several alternative estimators to the sample averages are proposed to reduce the bias and to improve the precision of estimation. Confidence intervals are constructed using Woodroofes pivot method. Numerical studies are conducted to evaluate the bias and mean squared error of the estimators and the coverage probability of the confidence intervals.

Two-stage procedures for selecting the best diagnostic biomarkers

Considered in the paper is the problem of selecting a diagnostic biomarker that has the highest classification rate among several candidate markers with dichotomous outcomes. The probability of correct selection depends on a number of nuisance parameters from the joint distribution of the biomarkers and thus can be substantially affected if these nuisance parameters are misspecified. A two-stage procedure is proposed to compute the needed sample size that achieves the desired level of correct selection, as so confirmed by simulation results.