The Hong Kong-based Center for Artificial Intelligence and Robotics (CAIR) under the Chinese Academy of Sciences (CAS), one of China’s national research institutes, has developed a large language model-based chatbot to help doctors make medical diagnoses and treatments.
HOW IT WORKS
The chatbot, CARES Copilot 1.0, is built on Meta’s Llama 2 LLM. It is trained using 100 GPUs from Huawei and Nvidia and can process various data including images, text, speech, video, MRI, CT scans and ultrasound.
Fed with millions of data, including educational materials, expert guidelines and medical literature, CARES can retrieve information “in a matter of seconds,” with up to 95% accuracy, CAIR claims. Equipped with a 100K context window, CARES can also extract information from more than 3,000 pages of surgical materials.
Based on internal testing, it can support functions such as surgical stage identification, segmentation of instruments and anatomical structures, detection and counting of instruments, and generation of high-resolution MRI images.
The assistant AI tool, which can be integrated with medical devices, is now continuously optimized in seven hospitals in Beijing.
THE BIG TREND
China is playing catch-up with global innovators in developing AI technologies by funding local initiatives. CAIR’s latest AI development is one of those projects co-funded by the Hong Kong government’s InnoHK research program. In August last year, the Chinese government approved the rollout of the country’s first homegrown genAI service, the TaiChu model, also by CAS.
ON THE RECORD
“(LLMs), tailor-made for neuroscience, can exceed traditional educational needs and enable applications in clinical settings, operating rooms and research institutes in combination with surgery, image guidance and robotics. These models directly support frontline medical personnel in emergency situations by monitoring, providing alerts, preventing risky procedures and pushing the boundaries of neuroscience. The combination of LLMs and surgical navigation can provide surgeons with real-time anatomical positioning information, increasing surgical safety,” Ming Feng, professor and chief physician at the Department of Neurosurgery, Beijing Union Medical College Hospital, commented on the release of CARES.