Anonymize or synthesize? Privacy-preserving methods for heart failure score analytics

Area of Science:

• Medical Informatics
• Health Data Privacy
• Cardiology Research

Background:

• Data availability is crucial for medical research but hindered by patient privacy concerns.
• Anonymization and data synthesis are key strategies to overcome privacy restrictions in health records.

Purpose of the Study:

• To investigate the effectiveness of anonymization and data synthesis for protecting patient privacy.
• To evaluate data utility and privacy risks using real-world cardiology health records.

Main Methods:

• Implemented anonymization (ARX) and data synthesis (ASyH) techniques on structured cardiology data.
• Assessed data utility by calculating heart failure risk scores on protected datasets.
• Conducted privacy risk assessments, including re-identification risk analysis.

Main Results:

• Anonymization and synthesis methods showed minimal deviations in heart failure risk scores compared to original data.
• Privacy risk analysis revealed only minor residual risks for common privacy threats.
• Both individual and combined approaches effectively protected patient privacy.

Conclusions:

• Anonymization and data synthesis successfully protect privacy while preserving data utility for heart failure risk assessment.
• Combining anonymization with data synthesis offers enhanced privacy protection with minimal impact on data utility.
• Generated datasets are available to the scientific community under a use and access agreement.