Taiwan's valuable healthcare data is set to be utilized for AI training and application for the first time. Yesterday, the Ministry of Health and Welfare’s National Health Insurance Administration (NHIA), Google Cloud, and Google Health announced the launch of "AI-on-DM," the nation's first AI-on-Diabetes Management project. This initiative will integrate long-term data on type 2 diabetes from the NHIA with Google Health’s medical large language model, utilizing AI platforms to predict and manage complication risks for over 2 million type 2 diabetes patients in Taiwan years in advance.
The "AI-on-DM" project, which stands for "Artificial Intelligence on Diabetes Management," will leverage Google Cloud's advanced AI technology, cloud computing, and data analysis capabilities to support the NHIA’s "Healthy Taiwan" national initiative. This project aims to expand in the future to include the management of patients with hypertension, hyperglycemia, hypercholesterolemia, and cancer.
Google Cloud has long been developing machine learning and AI technologies, with a specific focus on diabetes-related image recognition technology. Lee Kun-Yuan, Managing Director of Google Cloud North Asia, stated that this collaboration with a national institution marks a first for Google Cloud Health. The goal is to highlight to governments worldwide the importance of incorporating AI into healthcare to improve care and medical strategies. The outcomes of this collaboration could influence how governments formulate and implement national healthcare strategies, especially in addressing the growing prevalence of preventable diseases.
NHIA Director-General Shi Chong-Liang noted that this collaboration is a milestone for Taiwan and is part of the national strategy to develop precision medicine. Initially, the focus will be on type 2 diabetes, one of Taiwan's most common and costliest chronic diseases. Diabetes affects over 13% of the adult population in Taiwan, with one in four Taiwanese suffering from hypertension, placing a significant and rapidly growing burden on the healthcare system and society.
Shi explained that the NHIA will first organize the data and input it into Google's MedLM, a foundational model fine-tuned for the healthcare industry. This data will be used to build, train, and test a customized generative AI large language model on the Vertex AI platform. The National Health Research Institutes (NHRI) will provide domain expertise to establish the initial model, aiming to predict complication risks for type 2 diabetes patients years in advance.
Shi emphasized that this collaboration includes developing a new digital care model, allowing disease management across Taiwan to leverage Google’s AI technology, thereby enhancing the ability to detect disease risks.
Resource (mandarin)
