Improved angelization technique against background knowledge attack for 1:M microdata.

Area of Science:

• Health Informatics
• Data Privacy
• Computer Science

Background:

• Sharing Electronic Health Records (EHRs) across organizations offers benefits for medical treatment and research but raises significant privacy concerns.
• Traditional privacy models often assume one record per individual (1:1 datasets), which is insufficient for realistic scenarios with multiple records per individual (1:M datasets).
• Existing privacy models like θ-Sensitive k-Anonymity, (p, l)-angelization, and (k, l)-diversity demonstrate high utility loss and inadequate privacy for 1:M datasets.

Purpose of the Study:

• To address the limitations of current privacy models in handling 1:M datasets.
• To propose a novel algorithm that balances enhanced privacy with data utility for 1:M datasets.
• To evaluate the effectiveness of the proposed algorithm against existing methods.

Main Methods:

• The study analyzes the inapplicability and high utility loss of existing privacy models (θ-Sensitive k-Anonymity, (p, l)-angelization, (k, l)-diversity) for 1:M datasets.
• A new algorithm, (θ*, k)-utility, is proposed to improve privacy and utility preservation for anonymized 1:M datasets.
• Experiments were conducted using a real-world dataset to compare the proposed approach with existing methods.

Main Results:

• The proposed (θ*, k)-utility algorithm demonstrates superior performance in preserving both privacy and data utility for 1:M datasets compared to existing models.
• Existing models show significant utility loss and privacy vulnerabilities when applied to 1:M datasets.
• Experimental results validate the effectiveness of the new algorithm on real-world data.

Conclusions:

• The (θ*, k)-utility algorithm offers a robust solution for privacy-preserving data sharing of 1:M EHR datasets.
• The findings highlight the inadequacy of conventional privacy models for complex, multi-record datasets.
• This research contributes to the development of more secure and practical data sharing practices in healthcare.