American Statistical Association New York City Metropolitan Area Chapter The Levin Lecture Series Department of Biostatistics Colloquium Columbia University

Statistical methods for the evaluation of the accuracy of diagnostic tests usually assume that the true disease status is binary. However, this assumption may not be realistic in practice, when disease status is defined by dichotomizing continuous or ordinal categorical measures. In this presentation, we will discuss the analysis of studies in which both the diagnostic test and the reference standard are reported as continuous measures. Using a semi-parametric model and isotonic regression or monotone smooth splines, we will derive estimates of sensitivity, specificity and the ROC curve and examine the effect of a reference standard threshold on the performance of a diagnostic test.

Constantine Gatsonis, Ph.D., is Henry Ledyard Goddard University Professor of Biostatistics and Chair of the Department of Biostatistics at Brown. He joined the faculty 1995 and became the founding Director of the Center for Statistical Sciences. Dr. Gatsonis is a leading authority on the design and analysis of clinical trials of diagnostic and screening modalities and has extensive involvement in methodological research in medical technology assessment and in health services and outcomes research. He is Group Statistician of the American College of Radiology Imaging Network (ACRIN), a NCI funded collaborative group conducting multi-center studies of diagnostic imaging and image-guided therapy for cancer. In his ACRIN work, Dr Gatsonis is the chief statistician of the Digital Mammography Imaging Screening Trial (a national study comparing digital to film mammography) and is also the chief statistician for ACRIN's arm of the National Lung Screening Trial (NLST). Dr Gatsonis was the lead statistician of the International Breast MRI Consortium and of the Radiologic Diagnostic Oncology Group (RDOG). Dr. Gatsonis is the founding editor-in-chief of Health Services and Outcomes Research Methodology and serves as a deputy editor of Academic Radiology and a member of the editorial board of Clinical Trials. Previous editorial experience includes membership of the editorial board of Statistics in Medicine, and Medical Decision Making. Dr Gatsonis was elected fellow of the American Statistical Association and the Association for Health Services Research.