AI models for tasks such as pathology and dermatology struggle to generalize to new patient groups or hospitals that they were not trained on; learning more robust features from unlabeled data could prevent overfitting to the training distribution and thereby increase fairness.
References
Rajpurkar, P. et al. Nat. Med. 28, 31–38 (2022).
Kinyanjui, N. M. et al. In Medical Image Computing and Computer Assisted Intervention – MICCAI 2020 (eds. Martel, A. L. et al.) Vol. 12266, 320–329 (2020).
Quiñonero-Candela, J. et al. (eds.) Dataset Shift in Machine Learning (MIT Press, 2022).
Sagawa S. et al. In Int. Conference on Learning Representations https://go.nature.com/3wRZsW6 (2022).
Vaidya, A. et al. Nat. Med. https://doi.org/10.1038/s41591-024-02885-z (2024).
Ktena, I. et al. Nat. Med. https://doi.org/10.1038/s41591-024-02838-6 (2024).
Chen, R. J. et al. Preprint at https://doi.org/10.48550/arXiv.2308.15474 (2023).
Balachandar, S., Garg, N. & Pierson, E. In Int. Conference on Learning Representations https://go.nature.com/4a7Pqyg (2024).
Gao, I. et al. In Proc. 40th International Conference on Machine Learning 202, 10800–1083, (PMLR, 2023).
Pfohl, S. R. et al. J. Biomed. Inform. 113, 103621 (2021).
The All of Us Research Program Investigators. N. Engl. J. Med. 381, 668–676 (2019).
Mullainathan, S. & Obermeyer, Z. Nat. Med. 28, 897–899 (2022).
Movva, R. et al. In Proc. 8th Machine Learning for Healthcare Conference 219, 443–472 (PMLR, 2023).
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Competing interests
The authors declare no competing interests.
Rights and permissions
About this article
Cite this article
Movva, R., Koh, P.W. & Pierson, E. Using unlabeled data to enhance fairness of medical AI. Nat Med 30, 944–945 (2024). https://doi.org/10.1038/s41591-024-02892-0
Published:
Issue Date:
DOI: https://doi.org/10.1038/s41591-024-02892-0
- Springer Nature America, Inc.