2026-06-16 フィンランド技術研究センター(VTT)
＜関連情報＞
- https://www.vttresearch.com/en/news-and-ideas/privacy-preserving-solution-using-ai-cardiovascular-care
- https://link.springer.com/chapter/10.1007/978-3-032-28829-5_24
安全な心血管疾患リスク評価と臨床ケアのためのエンドツーエンドアーキテクチャ End-to-End Architecture for Secure Cardiovascular Disease Risk Assessment and Clinical Care
Gaurang Sharma,Juha Pajula,Tuomas Granlund,Petri Alhainen,Tommi Kiljander,Ornela Bardhi,Jaakko Lähteenmäki,Aada Illikainen,Antti Väänänen,Noora Lipsonen,Sari Kaari,Ville Salaspuro & Mika Hilvo
Digital Health and Wireless Solutions: Connected Digital Health: Digital Twins, Wearables, Wireless Systems, and Secure Architectures Published:16 June 2026
DOI:https://doi.org/10.1007/978-3-032-28829-5_24
Abstract
Cardiovascular disease (CVD) risk assessment comprises multiple data-intensive processes, including patient consent management, data acquisition, electronic health record (EHR) integration, and predictive model development, each introducing distinct privacy risks. This study presents a privacy-first, end-to-end system architecture for CVD risk management spanning both primary and secondary prevention. For primary prevention, we developed predictive models without sharing data via real-world federated learning (FL) infrastructure, supported by a secure platform for server deployment and controlled data access. For secondary prevention, we developed privacy-aware workflows for consent management, electrocardiogram (ECG) monitoring, and integration of electronic health records (EHRs). Furthermore, we leveraged the integrated data to build secure clinical decision-support tools that mitigate hallucinations and adapt dynamically to updates in patient records. Together, these components form a generalizable, privacy-preserving architecture for AI-driven cardiovascular care that applies to other data-intensive clinical domains.
