AI for Health Healthon 2026

Ibsar Al-Ayn إبصار العين

AI for early detection of diabetic retinopathy — a portable, smartphone-based retinal camera paired with an explainable AI platform.

90%Accuracy on held-out test images
4Deep-learning models in the ensemble
5ICDR-based severity levels
AR/ENBilingual clinical reports
Overview

Ibsar Al-Ayn is an AI-enabled portable screening solution for the early detection of diabetic retinopathy, a major cause of preventable blindness. It combines a smartphone-based retinal camera with an explainable AI platform to bring screening closer to patients in Saudi Arabia.

The kit has two parts. A handheld retinal camera uses smartphone optics with a red preview light, a white flash and a Bluetooth trigger to capture fundus images. These images are sent to an AI web platform that first checks image quality, then grades diabetic retinopathy severity on the five-level ICDR scale using an ensemble of four deep-learning models. Grad-CAM heatmaps highlight the retinal regions behind each prediction, and the platform produces Arabic and English PDF reports with referral tracking to specialists. The system achieved 90% accuracy on held-out test images.

Key features

  • Retinal image quality assessment
  • Five-level ICDR grading
  • Ensemble of four deep-learning models
  • Explainable AI with Grad-CAM
  • Arabic & English reports
  • PDF reports & referral tracking

What's next

The next goal is real-world validation: building the physical prototype, validating the models on Saudi retinal datasets, and piloting the solution in a primary-care setting.

Back to portfolio