Chinese and International Researchers Develop AI Model to Enhance Eye Care Accuracy

Beijing, Sept. 5 (Xinhua) — An international research collaboration led by Chinese scientists, published in Nature Medicine, has revealed that “EyeFM,” a foundational artificial intelligence (AI) model, holds the potential to revolutionize global eye health care.

The EyeFM model was developed by Tsinghua University, Shanghai Jiao Tong University, and international research partners. It was trained on 14.5 million eye images along with clinical texts from multi-ethnic datasets worldwide.

A total of 44 ophthalmologists from primary care and specialized clinics in China, India, Malaysia, Denmark, Equatorial Guinea, and the United States helped evaluate the performance of EyeFM. The results reinforced the value of the AI model as a clinical assistant. Moreover, “doctor feedback” was incorporated to optimize the model for use in both resource-limited clinics and highly specialized hospitals.

Earlier AI tools of a similar kind tended to learn from only a single type of data, making them less capable of processing diverse inputs the way human doctors can. They also often relied solely on historical records and were rarely pre-tested across different treatment scenarios and healthcare settings. Additionally, very few comparable AI tools had been applied in randomized trials, and studies on doctor-AI collaboration were limited.

In a double-blind, single-institution trial involving 668 high-risk patients in China, 16 ophthalmologists were randomly assigned to use either EyeFM or standard care during retinal disease screenings. Preliminary analysis showed that EyeFM raised diagnostic accuracy to 92.2%, compared with 75.4% in the control group.

This study provides high-level evidence that large-scale medical AI models can support both primary and specialized care, offering a replicable approach that transforms AI into an effective clinical tool.

Source:

www.xinhuathai.com