Reducing health disparities in clinical decision support with machine learning

We are developing algorithms that can adapt to changing hospital environments in real time and make predictions that are equally accurate among patient subpopulations. We investigate these algorithms for clinical decision support in emergency medicine and other fields.

Code

  • PMCBoost: Proportional Multicalibration Boosting
  • Interfair: Intersectional Fairness using FOMO

    • Press

    🎉 La Cava and Lett’s fair ML tool, Interfair, won first place in the 2023 NIH Challenge, Bias Detection Tools for Clinical Decision Making.

    Related Posts

    Selected Papers

    Intersectional and Marginal Debiasing in Prediction Models for Emergency Admissions
    Elle Lett, Shakiba Shahbandegan, Yuval Barak-Corren, Andrew M. Fine, William G. La Cava (2025)
    JAMA Network Open
    doi.org  |  PubMed  |  pdf  |  code
    Deciphering the influence of demographic factors on the treatment of pediatric patients in the emergency department
    Helena Coggan, Anne Bischops, Pradip Chaudhari, Yuval Barak-Corren, Andrew M. Fine, Ben Y. Reis, Jaya Aysola, William G. La Cava (2025)
    Pacific Symposium on Biocomputing (PSB)
    psb.stanford.edu  |  arXiv  |  pdf  |  code  |  blog
    The Benefit of the Doubt Phenomenon in Emergency Triage Assignment Disparities
    Blanca Romero Mila, Helena Coggan, Andrew M. Fine, Yuval Barak-Corren, Ben Y. Reis, Jaya Aysola, Pradip Chaudhari, William G. La Cava (2026)
    Preprint
    medrxiv.org  |  pdf  |  code
    Equitable Survival Prediction: A Fairness-Aware Survival Modeling (FASM) Approach
    Mingxuan Liu, Yilin Ning, Haoyuan Wang, Chuan Hong, Matthew Engelhard, Danielle S. Bitterman, William G. La Cava, Nan Liu (2025)
    Preprint
    arxiv.org  |  pdf
    Fair admission risk prediction with proportional multicalibration
    William G. La Cava, Elle Lett, Guangya Wan (2023)
    Proceedings of Machine Learning Research (PMLR) · Conference on Health, Inference, and Learning (CHIL)
    proceedings.mlr.press  |  pdf  |  code  |  Best Paper Award