Tibshirani Gives Brogan Lecture on UniLasso

Robert Tibshirani and Donna Brogan

Robert Tibshirani, professor of biomedical data science and statistics at Stanford University, presented the 2025 Donna J. Brogan Lecture in Biostatistics on April 10.

Tibshirani has made seminal contributions to statistical methodology, including the Lasso, generalized additive models, and significance analysis of microarrays. He has also coauthored several texts, including The Elements of Statistical Learning. His work has been widely recognized with honors, including the Committee of Presidents of Statistical Societies Presidents’ and Distinguished Achievement and Lectureship awards, the Statistical Society of Canada Gold Medal, and election to the National Academy of Sciences and Royal Society of the UK.

Tibshirani’s lecture was titled “UniLasso: A Novel Statistical Method for Sparse Regression.” In his presentation, he introduced ‘UniLasso,’ a two-stage approach that preserves the signs of the univariate coefficients and leverages their magnitude. Through comprehensive simulations and applications to a variety of biomedical problems, he demonstrated that UniLasso outperforms Lasso in terms of sparsity and model interpretability. He also established asymptotic support recovery and mean-squared error consistency under a novel set of conditions, distinct from the traditional irrepresentability conditions associated with the Lasso.

This annual lecture series honors Donna J. Brogan, a former faculty member at Emory University and chair of the Rollins School of Public Health Department of Biostatistics.