Equitable AI for disease screening

Background

A significant number of human diseases disproportionately impact racial and ethnic minorities, as well as socioeconomically disadvantaged groups. This disparity largely stems from unhealthy lifestyle choices and inadequate healthcare access often associated with poverty. For example, vision loss due to glaucoma, a lead cause of irreversible blindness globally, is more severe in Blacks than in Whites and Asians. In addition, many diseases are underdiagnosed in racial and ethnic minorities and socioeconomically disadvantaged groups impacted. For instance, it has been reported that 50% of people with glaucoma do not know they have the disease, and Blacks and Hispanics are 4.4 times and 2.5 times greater odds of having undiagnosed and untreated glaucoma than non-Hispanic Whites. Disease detection with deep learning using medical imaging data is promising to provide affordable disease screening to alleviate societal disease burden and reduce health disparities between different demographic identity groups, which can be deployed in primary care and pharmacies without needing the subjects to visit the more expensive and busy specialty clinics. Potential deep learning systems for automated disease screening should promote healthcare equality prior to clinical implementation.

What We Do

Our team is dedicated to developing equitable AI models for disease screening with a special focus on ocular diseases. Our recent research has developed a series of deep-learning techniques to improve model performance equity for eye disease screening.  We have developed a fair identity normalization technique to equalize feature importance between different identity groups to improve model performance equity. Furthermore, we have developed a fair Error-bound scaling technique to improve segmentation performance fairness using the Segment Anything Model (SAM). Most recently, we developed a multimodal vision-language model termed FairCLIP to improve model performance fairness in the context of vision-language modeling by minimizing the vision-language correlation distance between different demographic groups with Sinkhorn loss. Our team is committed not only to developing novel equitable AI techniques but also to releasing public datasets that can be used to facilitate a greater community of researchers to study medical AI equity. The success of this project will greatly benefit racial and ethnic minorities and socioeconomically disadvantaged groups with more equitable disease screening. Our equitable AI models developed based on the application of eye disease screening can be generalized to be used in other disease screening tasks to make a broader impact on various medical AI applications. Check out our open-source code repositories on our Harvard Ophthalmology AI Lab GitHub account.

Selected Publications

  • Luo, Y., Tian, Y., Shi, M., Elze, T. and Wang, M., 2024. Harvard Glaucoma Fairness: A Retinal Nerve Disease Dataset for Fairness Learning and Fair Identity Normalization. IEEE Transactions on Medical Imaging.
  • Tian, Y., Shi, M., Luo, Y., Kouhana, A., Elze, T. and Wang, M., FairSeg: A Large-Scale Medical Image Segmentation Dataset for Fairness Learning Using Segment Anything Model with Fair Error-Bound Scaling. In The Twelfth International Conference on Learning Representations.
  • Luo, Y., Shi, M., Khan, M.O., Afzal, M.M., Huang, H., Yuan, S., Tian, Y., Song, L., Kouhana, A., Elze, T., Fang, Y. and Wang, M., 2024. FairCLIP: Harnessing fairness in vision-language learning. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition(pp. 12289-12301).