Brian Williamson

Postdoctoral Research Fellow

Fred Hutch

About

Pronouns: he/him/his

I am a postdoctoral fellow in the Vaccine and Infectious Disease Division at Fred Hutch, where I work with Ying Huang.

I completed my doctoral studies in Biostatistics at the University of Washington under the guidance of Marco Carone and Noah Simon.

I am interested in problems of high-dimensional data, data science, and statistical inference. In particular, I enjoy working on developing new methodology for addressing scientific questions using large and complex datasets, and appropriately quantifying uncertainty in the resulting estimates. I also enjoy working on problems involving how best to tell a story using data. I have been primarily motivated by applications in public health and medicine, but I am always open to collaborations on interesting problems in any area of research.

Interests

Biostatistics
Semiparametric inference
High-dimensional data
Infectious diseases
Biomarkers

Education

PhD in Biostatistics, 2019

University of Washington, Seattle
MS in Biostatistics, 2017

University of Washington, Seattle
BA in Mathematics, 2014

Pomona College

Experience

Postdoctoral Research Fellow

Fred Hutch

Jan 2020 – Present Washington

Methodological research on issues arising in cancer biomarker panel development and vaccines against infectious diseases.

Pre-doctoral Instructor

Department of Biostatistics, University of Washington

Mar 2018 – Jun 2018 Washington

Taught a regression methods course for undergraduates.

Graduate Research Assistant

Fred Hutch

Sep 2014 – Dec 2019 Washington

Statistical methods research on issues arising in HIV/AIDS prevention clinical trials and analysis of clinical trials data.

Projects

Microbial Abundance

Variable Importance

Collaborative Science

Statistical Methods for Clinical Trials

Broadly Neutralizing Antibodies

Featured Publications

David Benkeser(?), Brian Williamson(?), Craig Magaret, Sohail Nizam, Peter Gilbert

July 2020 bioRxiv

Super LeArner Prediction of NAb Panels (SLAPNAP): a containerized tool for predicting combination monoclonal broadly neutralizing antibody sensitivity

PDF Code Project

Brian Williamson, Jean Feng

June 2020 arXiv

Efficient nonparametric statistical inference on population feature importance using Shapley values

Jean Feng and I tied together developments in the Shapley value literature in both game theory and machine learning. We propose a computationally efficient procedure that solves the issue that the classical Shapley values require estimating $2^p$ regression functions (in a problem with $p$ variables).

PDF Code Project

Brian Williamson, Peter Gilbert, Noah Simon, Marco Carone

April 2020 arXiv

A unified approach for inference on algorithm-agnostic variable importance

In this work, we extended our approach from the R-squared variable importance measure to handle nonparametric extensions of many interesting measures, including the difference in classification accuracy, the difference in area under the receiver operating characteristic curve, and the difference in deviance.

PDF Code Project

Scott Neidich, Youyi Fong, Shuying Li, Daniel Geraghty, Brian Williamson, et al.

October 2019 The Journal of Clinical Investigation collaborative-science