other

some talks

A few talks I've given outside of specific papers

  1. Test-Driven Development: Making the most of LLMs in research today and planning (a bit) for the future
    Keynote, AI and Public Affairs Workshop, Indiana University Bloomington
    Some thoughts on LLMs in academic research: given the pace of development, these ideas may already feel dated, but that's precisely why such discussions matter. I also recommend reading up on the big picture/trajectory, broader risks, and benchmark progress.
  2. Sunny with a Chance of Hurricane: Decision analytic metrics for forecast evaluation
    How to design forecast evaluation metrics for policy
  3. The Surprising Power of Napkin Math: Why simple models matter (more) in 2024
    Keynote, Modeling for Policy, University of Stuttgart, Germany
    Why simple models are one of our most powerful tools for research and policy
  4. Don't Feed the Bugs: Reproducibility for health policy research
    Why researchers should work more like software engineers (much optimism about LLMs improving this)
  5. What Exactly Did I Do Today? Making the most of the PhD research years
    What I wish I knew before starting my PhD

podcasts

  1. Casual Inference Podcast — "Starting the Conversation on Models"
  2. Humans in Public Health — "Harnessing AI for Smarter Health Policy Research"
  3. JAMA Author Interviews — "Supporting Children and Families in an Era With Abortion Restrictions"
  4. Radio Health Journal — "Why We Should Experiment On Pregnant Women"
  5. JAMA Network Open Conversations — "Modeling Contact Tracing Strategies for COVID-19"

other writing

  1. A New Vision for Women's Health Research: Transformative change at the National Institutes of Health
    as a member of the National Academies of Sciences, Engineering, and Medicine Committee on the Assessment of NIH Research on Women's Health, 2025
  2. Our best evidence says acetaminophen is safe during pregnancy. Better evidence could lay the issue to rest.
    with Katherine McDaniel, Stat, 2025
  3. Keeping pregnant participants out of medical research is more dangerous than including them.
    Stat, 2025
  4. Surveillance, screening, and self-testing can help us manage COVID-19 in the long term.
    with Joseph Hogan, Elizabeth Stuart, and Joshua Salomon, Boston Globe, 2022
  5. Data from routine COVID testing can help schools stay open this year.
    with Joshua Salomon, Stat, 2021
  6. Pandemic models can be more useful. Here’s how.
    (still one of my favorites) with Joshua Salomon, Health Affairs Forefront, 2021