About me

I’m Nathaniel (Nate) O’Connell, PhD, an Associate Professor in the Department of Biostatistics and Data Science at Wake Forest University School of Medicine. My research focuses on the intersection of machine learning and statistical inference, especially uncertainty quantification for prediction models and estimates of their performance.

Much of my methodological work is motivated by questions from biomedical collaborations and real-world issues I encounter routinely in my applied work. Over years of applied research, I have repeatedly encountered problems for which I could not find satisfactory answers. In collaborations involving machine learning and prediction, many of these questions concern uncertainty: how much confidence we can place in predictions and performance estimates, and how the choices made during an analysis affect the conclusions we draw.

My recent work on cross-validation and random forests brings together years of thinking about these problems and working toward answers. Going forward, I plan to devote more of my research to developing statistical methodology and theory, with my applied collaborations continuing to shape the problems I pursue. I co-lead SOUP Lab with Jaime Speiser, focusing on prediction and statistical inference.

I earned my PhD in Biostatistics at the Medical University of South Carolina, where my dissertation focused on eliciting patient toxicity scores and applying them to Bayesian dose-finding trials. Adaptive trial design remains an active interest. My current work extends that foundation to designs that use repeated patient outcomes.

I also serve as Director of Biostatistics for the ARISE Cancer Consortium, a multisite pediatric oncology consortium across North and South Carolina. I lead biostatistics for studies across the consortium. My broader collaborative work spans cancer, cardiovascular disease, pediatrics, neurology, emergency medicine, and electronic health record research, and has contributed to more than 70 coauthored papers.

My professional service includes grant review for the NIH, Department of Defense, and ASCO’s Conquer Cancer Foundation, reviewing for more than 20 journals, and membership on the Nature Medicine Statistical Advisory Panel in Neurology & Psychiatry.

On Out-of-Bag Thoughts, I write about technical ideas and other things I explore out of curiosity.

Beyond the research

Outside work, I enjoy time with my wife, our two kids, and our dog, Enzo. I used to long distances run competitively and still enjoy it, though I’ve gotten more into weightlifting in recent years. Related, I’ve developed my own weightlifting app, Platebook, that uses Bayesian models to track and estimate strength, fatigue, and recovery dynamics, along with a “smart” lifting program generator that builds a recovery optimal weekly workout plan within a user’s parameters. I’ll share the app (for free) and its methods here when they’re ready.

I grew up playing baseball as an Atlanta Braves fan, and I still watch 100+ games a year still today. I’m also a Philadelphia 76ers fan (there’s a story for how I’m an Atlanta baseball fan but Philly basketball fan). I also love college football. A lot of my statistical side projects start with a sports question I want to figure out simply because I’m a fan. There doesn’t need to be a paper at the end of it. My blog, Out-of-Bag Thoughts, will have room for those projects, technical posts, and whatever else I feel like writing about.

I’m also a car enthusiast. I love watching F1 racing (Ferrari is my team) and take fantasy F1 probably too seriously. I’ve developed some statistical tools I use for my F1 fantasy game, and I’ve been ranked as high as 30th in the world out of 2 million+ globally.

A few other things about me: I’m a big fan of Fast & Furious movie franchise, regardless of how cheesy they became. O.A.R. has been my favorite band since 2001, and has been my most played artist on Spotify every year since Spotify has been around. And I can ride a unicycle.