New AI Tool Has Identified Anemia via Eye Images
Researchers developed an AI model that uses images of the eye to detect anemia with high precision.
Updated on Sept. 19, 2026 in Heart Disease
Live Poll
Do you believe automated screening tools will improve healthcare access in your community?
Scientists have developed HemaViT, a deep learning framework capable of identifying anemia through non-invasive conjunctival imaging. This technology reached a 95.8% accuracy rate during recent testing on over 1,300 images.
Why it matters
Timely detection of anemia is essential for preventing severe health complications, and non-invasive methods could significantly improve diagnostic accessibility. This study marks a step toward utilizing image-based AI to assist in clinical screening.
In a study using 1,320 conjunctival images, the HemaViT model achieved 95.8% accuracy and an AUC-ROC of 0.98. These results outperformed standard architectures like ResNet50, DenseNet121, and EfficientNet-B0.
The players
HemaViT
A transformer-based deep learning framework designed for non-invasive anemia detection via conjunctival imaging.
The details
The HemaViT framework employs Dual-Attention PSPNet to isolate the conjunctiva from images and a Feature Pyramid Network to extract visual characteristics. It further integrates Vision Transformers for global context, while the Improved Waterwheel Plant Algorithm optimizes the model's settings for high-precision detection.
Timeline
September 19, 2026: The research detailing the HemaViT model was published.
Health Landscape
This research builds upon the Eyes Defy Anemia dataset, which has become a benchmark for image-based diagnostic development. The model represents a push toward non-invasive, AI-driven diagnostics that could eventually supplement traditional blood testing for common conditions.
While this tool is currently in the research phase, it highlights the potential for future non-invasive screening options. Patients concerned about symptoms of anemia should continue to discuss blood work and diagnostic testing with their primary care physician.
The takeaway
AI models are increasingly capable of screening for systemic health issues through non-invasive imaging analysis. If you have concerns about fatigue or symptoms of anemia, consult your doctor about standard laboratory testing while these new digital tools undergo further clinical validation.
What happens next
Researchers plan to conduct future validations of the HemaViT framework using larger and more diverse clinical datasets.
Further reading
Learn more about the latest innovations in cardiovascular diagnostics by visiting our Heart Disease section.
Source note: This article includes information reported by Nature.
Live Poll
Do you believe automated screening tools will improve healthcare access in your community?
