A deep learning system named DeepSLE has demonstrated impressive capabilities in the detection of systemic lupus erythematosus (SLE) using retinal images, achieving a sensitivity of 98%, which surpasses that of both primary care physicians and rheumatology specialists. The AI was trained on a comprehensive dataset of retinal images from over 91,000 participants across various ethnic backgrounds. Its robustness across demographic groups suggests significant potential for broad clinical use, particularly in routine eye examinations for early lupus detection. The findings raise the possibility of integrating digital retinal photography into primary care for better lupus screening.
Source: Cell Reports Medicine