Key Takeaways
- A new AI system from Tel Aviv University and Sheba Medical Center detects anemia via video scans of eye blood vessels, achieving 82.8% accuracy.
- This non-invasive method could eliminate the need for traditional blood draws, enhancing accessibility, particularly in remote areas.
- The technology demonstrates promise for future handheld devices for quick health screenings in clinics and at home.
Innovative AI Technology for Anemia Detection
Researchers from Tel Aviv University (TAU) and Sheba Medical Center have created an artificial intelligence-based system capable of detecting anemia and estimating vital blood markers through a brief video scan of eye blood vessels. This novel approach could potentially eliminate the need for invasive blood draws, making health screening faster, more accessible, and non-invasive.
In a proof-of-concept study involving 224 participants, the new technology achieved an impressive 82.8% accuracy in detecting anemia by analyzing blood-flow patterns in the conjunctiva of the eye. The findings were published on April 8, 2026, in the scientific journal npj | Digital Medicine. The study was led by Tamir Denis, a master’s graduate from TAU, in collaboration with esteemed researchers from both TAU and Sheba Medical Center.
Blood tests are among the most widely performed medical procedures globally, yet they often rely on invasive sampling and complex laboratory analysis. Past efforts to find alternative, non-invasive methods have not shown significant correlations. This new technology targets the eye, a rich source of vascular information. Anemia affects around 30% of the global population, making this breakthrough particularly significant for improving healthcare accessibility in underserved regions.
The system, named Video-to-Vessels, leverages high-magnification recordings (50x) of tiny blood vessels in the conjunctiva and converts this data into a compact digital representation of vascular structure and blood-flow characteristics. These insights are processed through an AI system trained to correlate blood-flow patterns with key markers like hemoglobin (Hb) levels and red blood cell (RBC) counts.
During the study, participants underwent both standard blood tests and imaging via a 10-second video captured from both eyes. The correlation between the predictive system and laboratory results was robust for both hemoglobin levels and RBC counts, highlighting the potential of this approach. Notably, the system excelled in detecting subtle differences in extremely thin blood vessels, which provide critical information for predicting hemoglobin levels. In these narrow vessels, blood cells flow in single file, making it easier to analyze their movement and detect variations associated with hemoglobin concentration.
Another critical discovery was the impact of video processing on the system’s accuracy. Stabilizing eye movements and reducing digital noise significantly enhanced performance; without these improvements, predictive correlations dropped by 38% for hemoglobin and 19% for red blood cell counts.
Although the current study serves as a proof of concept, researchers believe that further expansive and diverse studies are essential before clinical implementation. They envision the possibility of developing this technology into a compact handheld device for first-line screenings in clinical and community settings, potentially even at home.
This innovative approach represents a significant advancement in the quest for non-invasive health diagnostics, aligning with the ongoing efforts to enhance healthcare delivery and accessibility for individuals worldwide.
The content above is a summary. For more details, see the source article.