Statistical Foundations for Biomarker Research

This text provides an overview of the key issues in designing and analyzing observational biomarker studies. The failure to properly design and analyze such studies has led to falsely optimistic expectations regarding biomarker utility. The challenge in presenting relevant material is that the topics, although already published individually, are spread over a wide range of articles and textbooks in epidemiology, statistics and clinical research. This text pulls together all major concepts and presents the material in a non-mathematical fashion.

Introduction and Overarching Topics. Broad Objectives for Observational Biomarker Studies. Cross-Sectional Analyses. Study Designs for, and Correlative Analyses of Longitudinal Data. Prediction Models. Further Measures of Classification and Incremental Improvement in Accuracy. Classification Bias, Data Splitting and Validation. Further Topics in Planning and Funding Observational Biomarker Studies. Conclusions.

Douglas P. Landsittel, PhD, is a professor of medicine, biostatistics and clinical and Translational science and director of the Section on Biomarkers and Prediction Modeling at the Center for Research on Health Care (CRHC) Data Center, Division of General Internal Medicine, University of Pittsburgh, USA.