Retfound

Retfound improves hypertension detection in silico (AUC 0.705 vs 0.634 (p<0.0001)).

In silicofine-tuned on 100 images, SEED dataset

Performance and label efficiency of traditional deep-learning models and a retina-specific foundation model for ocular and systemic disease detection: a retrospective comparative study · The Lancet. Digital health · 28 Aug 2026

Retfound improves systemic disease detection in silico.

In silicofine-tuned on smaller datasets (≤400 images)

Performance and label efficiency of traditional deep-learning models and a retina-specific foundation model for ocular and systemic disease detection: a retrospective comparative study · The Lancet. Digital health · 28 Aug 2026

Retfound has no effect on ocular disease detection in silico (AUC 0.938-0.966 vs 0.914-0.965).

In silicofine-tuned on full datasets

Performance and label efficiency of traditional deep-learning models and a retina-specific foundation model for ocular and systemic disease detection: a retrospective comparative study · The Lancet. Digital health · 28 Aug 2026