George Austin​

Overview

Titles

Ph.D. Student 

Email
gia2105@cumc.columbia.edu

Lab

Tal Korem

George is a PhD Student working in Tal Korem’s lab. His research involves developing machine learning methods to enable novel biological discoveries, with a focus on supporting microbiome applications to clinical settings. He has developed multiple models to correct for technical challenges in microbiome research, such as contamination and processing bias, and has applied these techniques to enable novel insights into the early pathogenesis of preeclampsia, and robust microbial signals within tumors.

Prior to joining Columbia’s PhD program, he worked as a machine learning scientist at UnitedHealth Group, where he worked on challenges such as colorectal cancer screening, multi-drug interactions, and automated prior authorization approvals.

Education History

MS, Computer Science, Columbia University, 2022

BS, Applied Mathematics, Columbia University, 2019

Publications

Austin GI, Korem T. Compositional transformations can reasonably introduce phenotype-associated values into sparse features. mSystems. 2025 May 20;10(5):e0002125.

Austin GI, Brown Kav A, ElNaggar S, Park H, Biermann J, Uhlemann AC, Pe'er I, Korem T. Processing-bias correction with DEBIAS-M improves cross-study generalization of microbiome-based prediction models. Nat Microbiol. 2025 Apr;10(4):897-911.

Austin GI, Pe'er I, Korem T. Distributional bias compromises leave-one-out cross-validation. ArXiv [Preprint]. 2025 Mar 24:arXiv:2406.01652v2. 

Kindschuh WF, Austin GI, Meydan Y, Park H, Urban JA, Watters E, Pollak S, Saade GR, Chung J, Mercer BM, Grobman WA, Haas DM, Silver RM, Serrano M, Buck GA, McNeil R, Nandakumar R, Reddy U, Wapner RJ, Kav AB, Uhlemann AC, Korem T. Early prediction of preeclampsia using the first trimester vaginal microbiome. bioRxiv [Preprint]. 2024 Dec 2:2024.12.01.626267. 

Austin GI, Korem T. Planning and Analyzing a Low-Biomass Microbiome Study: A Data Analysis Perspective. J Infect Dis. 2024 Aug 27:jiae378.

Austin GI, Park H, Meydan Y, Seeram D, Sezin T, Lou YC, Firek BA, Morowitz MJ, Banfield JF, Christiano AM, Pe'er I, Uhlemann AC, Shenhav L, Korem T. Contamination source modeling with SCRuB improves cancer phenotype prediction from microbiome data. Nat Biotechnol. 2023 Dec;41(12):1820-1828.

Austin G, Kowalkowski H, Guo Y, Miller-Wilson LA, DaCosta Byfield S, Verma P, Housman L, Berke E. Patterns of initial colorectal cancer screenings after turning 50 years old and follow-up rates of colonoscopy after positive stool-based testing among the average-risk population. Curr Med Res Opin. 2023 Jan;39(1):47-61.