Department of Biostatistics Seminar/Workshop Series

How To Do Bad Biomarker Research

Frank E. Harrell, Jr., Ph.D

Professor and Chairman, Department of Biostatistics

Vanderbilt University School of Medicine

This talk will cover some of the bad statistical practices that have crept into biomarker research, including improper use of continuous variables, the use of improper predictive accuracy indexes, and setting the bar too low for demonstrating that biomarker information is new. Some special problems in dealing with higher-dimensional candidate biomarkers will be briefly discussed. Kidney disease poses special problems in biomarker research because commonly used definitions of kidney dysfunction/injury are based on categorizations that violate several statistical and clinical principles. This will also be discussed.