The research team at The Chinese University of Hong Kong (CUHK) has successfully developed a new foundational artificial intelligence model called “VisionFM,” which focuses on the diagnosis and prediction of ophthalmic diseases. It has shown exceptional performance, particularly in disease screening and predicting the progression of glaucoma. Related research was published last month in the journal NEJM AI.
Image source note: Image generated by AI, licensed from service provider Midjourney
The capabilities of VisionFM are impressive. Research indicates that this model's accuracy in diagnosing 12 ophthalmic diseases is comparable to that of intermediate ophthalmologists, and in some tasks, it even performs better, particularly in predicting the progression of glaucoma, surpassing the previous foundational model in the field, RETFound. The research team believes that as the volume of training data increases, the application scenarios for VisionFM will further expand, potentially playing a key role in more ophthalmic clinical tasks in the future.
The launch of this groundbreaking technology coincides with the accelerated application of generative artificial intelligence in the medical field. An increasing number of researchers and medical experts are exploring how AI technology can enhance healthcare services, from surgical planning to diagnostic support, with generative AI reshaping medical practices.
For instance, earlier this year, the Hong Kong Centre for Artificial Intelligence and Robotics (CAIR) launched an AI model based on the open-source Meta Platforms model Llama2, which has been implemented at Sun Yat-sen University First Affiliated Hospital and Prince of Wales Hospital in Hong Kong to assist in surgical planning and generating diagnostic reports.
With the emergence of advanced technologies like VisionFM, the potential of artificial intelligence in the medical field is gradually being realized. This not only helps improve diagnostic accuracy but also provides new pathways for optimizing healthcare services and enhancing patient experiences.