A privacy-preserving solution for using AI in cardiovascular care

The use of AI in health care is challenging because sensitive patient data is scattered across different systems, and its use requires strong privacy protection. The new concept developed in the international Secur-e-Health project combines secure data processing, careful consent practices and privacy-preserving AI tools to support both disease prevention and the monitoring of patient care.

The Secur-e-Health project developed a concept that enables AI to be used in the treatment of cardiovascular diseases without compromising the protection of sensitive patient data.

“The study shows that AI tools can be built for health care in a way that respects patients’ privacy at every stage of care,” says Mika Hilvo, research team leader at VTT and national coordinator of the project. “Our work brings together the secure use of data, clinical needs and modern AI methods in a way that can support better care in the future.”

“This new approach can help strengthen trust between patients, data controllers, health care providers and researchers. The study shows that sensitive health data can be used collaboratively, securely and in a privacy-preserving way without organizations losing control of their data,” says Gaurang Sharma, the lead author of the publication and research scientist at VTT.

The publication on the solution, “End-to-End Architecture for Secure Cardiovascular Disease Risk Assessment and Clinical Care” was presented at the Nordic Conference on Digital Health and Wireless Solutions in Oulu.

The project created a privacy-preserving architecture for patient care. It combines secure data processing, careful consent practices and privacy-preserving AI tools. This can support both early risk assessment and the monitoring of care for patients with cardiovascular diseases.

The solution covers the entire care pathway, from the prevention of serious heart problems (primary prevention) to supporting the care of diagnosed patients (secondary prevention).

In the prevention-focused part of the work, the research team tested methods for training AI models using health data stored in different locations, without the data needing to be transferred to a single centralized location. This enables collaboration between organizations while helping to keep patient data better protected.

The results showed that models trained using privacy-preserving federated learning can perform as well as models trained using traditional machine-learning methods.

For patients requiring continuous care, the researchers created a secure process for obtaining treatment consent, collecting ECG monitoring data, combining data from different systems and reviewing patient information to support clinicians. The system was designed so that patient data can be combined securely without revealing identity information more widely than necessary.

The solution addresses one of the key challenges in modern health care: how to make use of AI when medical data is distributed across several different systems and must be handled with great care. By combining secure data use, patient consent and AI-based support tools, the researchers created a foundation that can help develop the digital health services of the future.

Who’s behind this story?

Sadie Harley

BSc Life Sciences & Ecology. Microbiology lab background with pharmaceutical news experience in oil, gas, and renewable industries.

Full profile →

Advertisements

Robert Egan

Bachelor’s in mathematical biology, Master’s in creative writing. Well-traveled with unique perspectives on science and language.

Full profile →