How to Balance AI Use and Patient Privacy with a Medical Data Warehouse System
How Hospitals Can Use AI Without Exposing Patient Data |
NSYSU is bringing together international healthcare data standards, including FHIR, to support AI use in hospitals.
Summary
A research team at NSYSU in Taiwan developed a medical data warehouse system that enables the use of AI while protecting patient data. The system is notable for combining international healthcare standards (like FHIR) with federated learning, allowing joint research and model training under each hospital's local control.
Key Players
Details
While artificial intelligence (AI) is increasingly used in diagnosis and disease prevention, concerns over highly sensitive patient data privacy remain a major hurdle. To address this, a research team at National Sun Yat-Sen University (NSYSU) developed a medical data warehouse system that does not require centralizing raw patient records or granting broad cloud access. The core of the approach integrates international healthcare data standards, such as FHIR, with cryptographic protection mechanisms and federated learning. This allows hospitals to collaborate and jointly train AI models while keeping the raw patient data within each institution's control. Crucially, this system moves beyond theory by addressing real-world operational challenges through prototype development and pilot testing, including interoperability across different systems, encrypted data operations, and cross-institution access control. Its modular and cloud-agnostic design provides flexibility for regional healthcare providers facing cost constraints or vendor lock-in concerns. This work demonstrates a practical method for advancing medical innovation using AI without compromising patient privacy, offering a solution for global healthcare systems seeking responsible adoption of artificial intelligence.
Technology Note
FHIR（Fast Healthcare Interoperability Resources）は医療データ交換の国際標準。このエントリの関連技術: FHIR
Original content copyright by respective publishers