Innovative Chronic Disease Care Model – Combined AI Advanced Cardiometabolic Image and Wearable Device Brand New Index to Establish and Verify Risk Prediction System of Taiwanese
Summary
At present, the cardiovascular risk prediction models used by the global medical community are based on clinical information and do not include cardiovascular imaging information.
Having developed the AI automatic full-thoracic computed tomography (CT) cardiac/aortic calcification quantitative model (HeaortaNet) and used non-contrast chest CT images (>1700 cases) from the National Health Insurance (NHI) database, we have established a cardiovascular risk perdition model based on heart/thoracic aorta calcification and peripheral fat quantification.
This project is expected to improve the aforementioned prediction model by subdividing the cardiac calcification according to the position, quantifying the calcification of the ascending aorta and the aortic arch separately, and integrating the continuous blood pressure monitoring information into the model to provide health information for Taiwanese.
Advantages
Our project plans to develop (1) omni-directional visceral/body fat quantification, (2) total vascular calcification classification and quantification, (3) wearable device signal algorithm development, and (4) new risk prediction model. We will comprehensively promote our technology implement to the AI industry in 4 aspects: database establishment/management, AI model development, clinical application, and regulatory agency (TFDA) certification. In addition, the model will transform traditional chronic disease care into monitor physiological information by wearable devices, with development of AI model that determine sub-health states, and help people prevent the disease occurrence in advance.
Applications
This project follows the previous one which was subsidized by the Ministry of Science and Technology for the construction of a large cross-module database of artificial intelligence cardiovascular images. It not only builds a large-scale multi-centered cardiovascular CT imagebank (>3,500 cases; ~10 GB), but also develops AI models and applications as below:
(1) Heart/ascending & descending aorta calcification fraction, peripheral fat, Taiwanese survival/hospitalization risk model.
(2) Implementation of AI in the diagnostic workflow in the Department of Imaging Medicine of National Taiwan University Hospital.
(3) Integration with wearable devices for the public healthcare in the community.
Keywords
Cardiometabolic Image, Artificial Intelligence, computed tomography, Wearable Device, Risk Prediction System
◎ PI
PI Tzung-Dau Wang
Professor, Dept. of Internal Medicine; Director, Division of Hospital Medicine, College of Medicine, NTU
WebSite