A research team from the UC Berkeley School of Public Health has won a $100,000 prize from the National Institutes of Health for building a clinical prediction model to forecast a patient’s risk of developing long Covid-19.
The competition team, led by postdoctoral scholar Zachary Butzin-Dozier, included members from the Center for Targeted Machine Learning and Causal Inference, where Alan Hubbard is a Principal Investigator. The team took third place in the NIH Long COVID Computational Challenge (L3C), a prestigious competition that drew submissions from 35 teams.
The primary objective of the competition, known as L3C, was to support creative data-driven solutions that meaningfully advance the current understanding of the risks of developing Post-Acute Sequelae of SARS-CoV-2. (PASC), known as long Covid.
The Berkeley Public Health team built a clinical prediction model that combined many smaller prediction models (this combined model is known as an ensemble or a Super Learner). The model used various aspects of a patient’s health such as their cardiovascular health, respiratory health, history of hospital use, and age, to predict the patient’s risk for developing long Covid.
“The great thing about this competition is that it gave us a great chance to jump into an applied analysis,” said Butzin-Dozier, a postdoctoral scholar who managed the project. “In this case, NIH had identified that research question for us, and gave us the data, so we could jump right into the analysis.”
Butzin-Dozier said the members of the team plan next to evaluate the relationship between the timing of the Covid vaccines and long Covid among people who have had breakthrough infections.
“There’s good data that vaccination protects you against long Covid,” he said.
Other members of the winning research group are: Yunwen Ji, Haodong Li, Jeremy Coyle, Junming (Seraphina) Shi, Rachael V. Philips, Andrew Mertens, Romain Pirracchio, Mark van der Laan, and John M. Colford, Jr. The students will share the $100,000 prize.
The top prize went to a joint project between the University of Chicago and the Argonne National Laboratory, and Geisinger AI Lab took second place. The University of Wisconsin-Madison, the University of Pennsylvania, and Ruvos were awarded honorable mentions.