A game theoretic framework for analyzing re-identification risk.

Area of Science:

• Computer Science
• Information Security
• Game Theory

Background:

• Organizations aim to share de-identified data for insights but fear re-identification.
• Existing research focuses on attack methods, not the likelihood of re-identification.
• A gap exists in practical frameworks for balancing data utility and privacy risks.

Purpose of the Study:

• To introduce a game theoretic framework for balancing data re-identification risk and data sharing value.
• To assess the feasibility of achieving zero re-identification risk while maximizing data utility.
• To compare the proposed framework's effectiveness against existing de-identification policies.

Main Methods:

• Developed a game theoretic model assuming recipients act only when gains outweigh costs.
• Applied the framework to a case study using grant funding as data value and privacy fines as re-identification gain.
• Conducted a sensitivity analysis to test the robustness of findings.

Main Results:

• Achieved zero re-identification risk, enabling data sharing comparable to solutions with minimal risk.
• The zero-risk solution facilitates significantly greater data sharing than the Health Insurance Portability and Accountability Act (HIPAA) de-identification policy.
• Findings remain robust despite substantial changes in potential gains and losses for involved parties.

Conclusions:

• A game theoretic framework can effectively balance data sharing value with re-identification risk.
• Pragmatic data sharing policies can be developed using this risk-benefit analysis approach.
• The proposed method offers a superior alternative to current de-identification standards for maximizing data utility while ensuring privacy.