Artificial intelligence framework for multi-pathology risk assessment from retinal fundus images: deep learning approach to 15-disease screening - Takeaways - MDSpire

Artificial intelligence framework for multi-pathology risk assessment from retinal fundus images: deep learning approach to 15-disease screening

  • By

  • Robert Vasilev

  • Andrey Savchenko

  • Pavel Blinov

  • Tadej Svetina

  • Stepan Kudin

  • Nikolay Romanenko

  • Yuliya Sarana

  • Gleb Khizhnyak

  • Andrey Demchinsky

  • Taisia Shcheglova

  • May 25, 2026

  • 0 min

Share

  • 1

    An AI framework was developed for simultaneous risk assessment of 15 distinct diseases from retinal fundus images.

  • 2

    The system employs a hybrid deep learning architecture to address class imbalance in clinical datasets.

  • 3

    Internal testing achieved ROC AUC scores between 0.9524 and 0.9971, indicating strong performance across all conditions.

  • 4

    Preliminary evaluation in a real-world setting showed an overall accuracy of 64.7% in a cohort of 68 cases.

  • 5

    The study highlights the need for larger, multicenter studies to validate the system's effectiveness in clinical settings.

Original Source(s)

Artificial intelligence framework for multi-pathology risk assessment from retinal fundus images: deep learning approach to 15-disease screening

  • by Robert Vasilev, Andrey Savchenko, Pavel Blinov, Tadej Svetina, Stepan Kudin, Nikolay Romanenko, Yuliya Sarana, Gleb Khizhnyak, Andrey Demchinsky, Taisia Shcheglova

  • May 25, 2026

Related Content

Research on refractive power calculation formula of intraocular lens based on the principle of double thick lens imaging

Beyond Just Steps

Dr. Cymbor encourages a shift in patient conversations to include vigorous aerobic exercise in the morning or late evening, according to recent research.

Thirty Years Together in a Career Defined by Change

Progress begins when we accept that the only certainty is uncertainty.