We’re excited to share our latest open-access publication in Computers in Biology and Medicine:
“MyDigiTwin: A privacy-preserving framework for personalized cardiovascular risk prediction and scenario exploration”.
In this work, they (first author Héctor Cadavid (eScience Center) with co-authors, in particular, Hyunho and Esther from the BIGR and Daniel from the Radiology & Nuclear Medicine / Epidemiology department) present MyDigiTwin, a framework that connects personal health environments (PHEs) with privacy-preserving predictive modeling to let individuals explore personalized cardiovascular (CVD) risk scenarios—without sharing raw health data.
MyDigiTwin connects Personal Health Environments with privacy-preserving federated learning to deliver individualized cardiovascular risk predictions without sharing raw data. A reusable FHIR/ZIB harmonization pipeline standardizes inputs and enables interactive “what-if” scenario exploration for lifestyle and clinical changes. In a Lifelines proof-of-concept using vantage6, federated models improved discrimination over local training while maintaining calibration, showing scalability for real-world use.
This article is open access under CC BY in Computers in Biology and Medicine (Available online October 9, 2025; DOI: 10.1016/j.compbiomed.2025.111180).