Glaucoma can gradually damage the optic nerve and lead to permanent vision loss, but spotting risk isn’t just about examining a single eye image. Important clues may be spread across different types of information, from the appearance of the optic disc to a patient’s medical history, medications, and previous eye exams. This patented technology addresses this challenge by enabling AI to analyze multiple data sources. It starts by evaluating colour fundus photographs, extracting anatomical features relevant to glaucoma. Instead of relying solely on image predictions, it also constructs a knowledge graph from electronic health records. This allows data such as medical history, lifestyle, medications, and exam results to inform the system’s decision. The goal isn’t just providing AI with more data, but representing different evidence types in a way that contributes to the same prediction. An eye image shows visible structures; medical records provide relationships and context that aren’t visually apparent. This strategy reflects a broader trend in the MUST patent portfolio: when evidence is scattered across various formats, AI can integrate these different pieces for a more complete understanding.
When an eye image is only part of the story
Patent number: CN 118711790 A
Inventor(s): K. Zhang, Y. Gao, and Z. Zou
Citation: K. Zhang, Y. Gao, and Z. Zou, “青光眼辅助预测方法及系统 [Glaucoma-assisted prediction method and system],” China Patent Application CN 118711790 A, Sep. 27, 2024.
