1. DeepSLE achieved 98% sensitivity detecting SLE from retinal images.
2. It outperformed primary care and rheumatology specialists.
3.  Trained on 254,246 retinal images from 91,598 participants.
4. Detection robust across demographics.
5. Detected lupus retinopathy with 100% sensitivity.
6. Highlights potential for opportunistic screening in routine eye exams.
7. Limitations due to datasets mainly from China and the UK.

Source: Cell Reports Medicine