A press conference for SaMD certification was held in Meeting Room 13 on the 2nd floor of the NSTC Science Building, presided over by Minister Wu. Teams that obtained Software as a Medical Device (SaMD) permits during the project period (Asia University, Chang Gung Memorial Hospital, and National Cheng Kung University) were specially invited to attend. The Asia University team consists of Asia University (AU) and China Medical University Hospital (CMUH). Project Title: Validation of a Clinical Nephrology Care Platform Architected with Medical Big Data, Artificial Intelligence, and Blockchain Technology via Multi-center Randomized Clinical Trials.
This project is implemented through: (1. Blockchain Architecture) Establishing a blockchain technology platform integrating AI applications to provide secure and private data exchange services, as well as managing a comprehensive renal medical database platform across medical institutions and databases. (2. Data) Constructing a comprehensive renal medical database for "Strong Kidney Chain" intelligent applications, covering hemodialysis electronic medical records, medical imaging and pathology reports, genetic information, and multimedia data such as gait, foot imaging, and skin imaging. (3. Dialysis Outpatient Care) Fully collecting, analyzing, and predicting real-time hemoglobin levels, vascular calcification degrees, and various personal quality indicators and complication risks (such as predicting falls and survival via gait). The project develops AI technology suitable for dialysis clinics and guides patients in self-exercise management to improve physical fitness, optimize disease progression, and extend life expectancy. (4. Precision Genomic Medicine) Analyzing kidney-related diseases through Genome-Wide Association Studies (GWAS) based on patient genetic variations. By linking the rich genetic database of CMUH with its internal medical database, multi-dimensional and complete renal genetic and medical data are formed. AI analysis is utilized to develop diagnostic and treatment support systems, strengthening Taiwan's integration of genomic and clinical data for kidney-related diseases, which aids in the prevention, diagnosis, and treatment of renal diseases domestically.
The project has published 23 international journal papers and 7 conference papers, passed 5 patents, and had one technical tool approved by the FDA and TFDA: Blockchain Distributed Federated Learning Platform: A federated learning platform supporting XGBoost AI models has been developed and tested at CMUH, Asia University Hospital (AUH), and AU. Intelligent Cardiothoracic Ratio (iCTR) Estimation System: Automatically estimates the cardiothoracic ratio from chest X-ray images and provides alerts as a prognostic indicator for end-stage renal disease, coronary heart disease, etc. Model development and validation are complete, and SaMD marketing authorizations have been obtained from the US FDA and Taiwan TFDA; it is currently licensed to Longjia Intelligence for commercial promotion. Intelligent Hemoglobin Scope (Hb Scope): The model was developed using dialysis catheter images from CMUH and three branch hospitals, achieving an accuracy of 0.93. In June this year, it entered the largest dialysis treatment system in Abu Dhabi for clinical trials, with US FDA approval to be sought next. A Taiwan patent has been obtained, and a US patent is pending. Intelligent Gait Analyzer: Currently recruiting cases at CMUH and AUH; a Taiwan patent has been obtained, and a US patent is pending. Early Chronic Kidney Disease Risk Assessment: Based on genotype data from over 240,000 Asians, 7 previously unpublished chronic kidney disease genetic loci were discovered.
Certified Product: Intelligent Cardiothoracic Ratio (iCTR) Estimation System
Highlight Description: The cardiothoracic ratio (CTR) is calculated by measuring the width of the heart and the width of the thoracic cage from common chest X-ray images. It serves as a prognostic indicator for many conditions, such as coronary artery disease (CAD), end-stage kidney disease (ESKD), and aging-related diseases. Generally, a CTR > 0.5 indicates cardiomegaly, requiring immediate alerts and follow-up treatment. However, existing imaging systems cannot detect this automatically, relying on physicians to manually measure the heart and thoracic widths and calculate the ratio, which is time-consuming, subject to individual variation, lacks real-time alerts, and cannot systematically scan a patient's past CTR records. Therefore, the CMUH team developed the iCTR system, which can systematically and automatically calculate the current CTR and the historical CTR progression, providing physicians with the status and progression of a patient's cardiomegaly.
This technology involves a three-step process: image quality confirmation, heart and lung contouring, and CTR calculation and presentation, allowing it to be applied in clinical settings where data quality may be poor. This technology has obtained a Taiwan patent (I746249), with US and Chinese patents pending. It has received US FDA 510(k) clearance for software as a medical device (K212624) and TFDA approval in Taiwan (007443). Research has proven that the CTR estimated by this technology is significantly correlated with future end-stage renal disease, cardiovascular death, and overall mortality in chronic kidney disease patients. The CMUH team published these findings in the 2023 journal Communications Medicine (https://www.nature.com/articles/s43856-023-00241-9). Current clinical application status: This technology was co-developed with Longjia Intelligence, and subsequent application planning has been entrusted to them. It is currently integrated into the CMUH system for testing and is not yet in actual clinical use. Future development plans: Future expansion may include the insurance industry, medical institutions of all levels, health screening centers, medical AI companies, and pharmaceutical companies. It can also be integrated with other clinical data to serve as a risk assessment and screening tool for the prevention and treatment of heart-related diseases.
