The Hong Kong University of Science and Technology recently announced the development of four major language models for healthcare.
WHAT IT’S ABOUT
The LLM-based tools were developed from HKUST SuperPOD, the university’s AI supercomputing facility, and include the following:
-
MOM: An AI model for identifying breast cancer pathologies in MRI images.
-
mSTAR: A tool for pathology assistants that allows entire slide images to be modeled directly.
-
MedMr: A multimodal language model for chatbots.
-
XAIM: An explainable AI model that provides visual and textual explanations of the AI-generated analytics.
In a statement to Healthcare IT newsthe HKUST research team further shared the details and findings of the development of these models.
Their AI model for breast cancer has achieved a diagnostic accuracy of 87% in multi-center testing. MOME can also predict patients’ responses to neoadjuvant chemotherapy.
mSTAR, the team said, addresses “a broad range of pathological clinical problems, including seven types of clinical diagnostic and prognostic applications such as cancer subtyping and staging, metastasis detection, molecular prediction, survival analysis and report generation.”
The medical chatbot MedMR, which answers questions, generates medical reports and makes initial diagnoses based on medical images, has achieved 93% accuracy in identifying tumors and non-tumors using Patch Camelyon’s publicly available PCam200 dataset .
Furthermore, they shared that XAIM achieved 98.67% accuracy in diagnosing skin lesions on the PH² dataset from Portugal.
The research team is now preparing to test these models at various hospitals in the city. “We are in discussions with several hospitals in Hong Kong about possible trials and implementations.”
“Before deploying the AI models in clinical practices, we will also need to collaborate with hospitals for large-scale, multi-center model validation to ensure their generalizability and reliability,” she added.
THE BIG TREND
Early this year, the Center for Artificial Intelligence and Robotics in Hong Kong, a research center under one of China’s national research institutes, the Chinese Academy of Sciences, also introduced a doctor chatbot called CONCERN. The chatbot is built on Meta’s Llama 2 LLM and is being tested in seven undisclosed hospitals in Beijing.
Other Asian healthcare systems have been working on similar generative AI projects. In October the Singapore’s Ministry of Health has announced a new investment to support a national project that will roll out genAI across the public health system by the end of 2025. Singapore-based Docquity is also helping community health centers in West Java, Indonesia, adopt genAI virtual assistant TehAIto support health professionals in diagnosing tuberculosis, growth retardation and high blood pressure.