A Hong Kong AI model proves more accurate than doctors in diagnosing eye conditions

The open-source VisionFM model from the Chinese University of Hong Kong is comparable to doctors at diagnosing ophthalmic diseases, paper finds

Reading Time:2 minutes

A new AI model from the Chinese University of Hong Kong has proven effective in identifying ophthalmic diseases. Photo: Shutterstock

Published: 9:00am, 22 Dec 2024

Researchers at the Chinese University of Hong Kong (CUHK) said they have developed a foundational artificial intelligence (AI) model that could help address ophthalmic clinical tasks including disease screening and diagnosis, as generative AI plays a greater role in medical research.

VisionFM is a new foundational model that is effective in diagnosing and predicting multiple ophthalmic diseases, which could lead to the adoption of more clinical applications with additional data, they said in a study published last month in the NEJM AI journal.

The model demonstrated performance that is comparable to or better than intermediate-level ophthalmologists in diagnosing 12 ocular diseases, according to the paper. It also outperformed the ophthalmic field’s first foundational model, RETFound, in predicting glaucoma progression, the researchers added.

The development of VisionFM comes as more medical researchers and professionals explore how generative AI can aid the healthcare industry.

The Centre for Artificial Intelligence and Robotics (CAIR), a Hong Kong-based research centre under the Chinese Academy of Sciences, the country’s national research institute, earlier this year launched an AI model named CARES Copilot, with the aim of helping surgeons in tasks including surgical planning and generating diagnostic reports.

CAIR’s model was based on Meta Platforms’ Llama 2 open source large language model, and it has been used by hospitals including the First Affiliated Hospital of Sun Yat-sen University in Guangzhou and the Prince of Wales Hospital in Hong Kong, the centre said last month.