Abstract
As medical data migrates to the cloud, privacy-preserving feature matching becomes essential for systems like personal health records and diagnostic related groups (DRG). Public Key Encryption with Equality Test (PKEET) allows authorized testers to compare encrypted data without decryption. However, standard PKEET is vulnerable to Offline Message Recovery Attacks (OMRA) due to the limited terminology in medical fields. This paper proposes Public-Key Authenticated Encryption with Designated Equality Test (PKAE-DET), a new framework that resists inside adversary attacks without requiring complex multi-server setups. We demonstrate its security under standard mathematical assumptions and its practical efficiency in medical diagnostic scenarios.
Introduction: The Vulnerability of Encrypted Medical Data
The transition of medical records to cloud environments necessitates methods to match patient features—such as checking if two patients share a diagnosis—without compromising sensitive information. Public Key Encryption with Equality Test (PKEET) was developed for this purpose, allowing a cloud "tester" to determine if two ciphertexts represent the same plaintext.
However, medical data often uses a constrained vocabulary. This allows a malicious database administrator to perform an Offline Message Recovery Attack (OMRA). By encrypting every possible medical term (e.g., "Diabetes", "Flu") using a patient's public key, the attacker can use the equality test to identify the underlying data through brute force.
Proposed Framework: PKAE-DET
To address these vulnerabilities, we introduce Public Key Authenticated Encryption with Designated Equality Test (PKAE-DET). This framework shifts the security burden to the encryption and authorization phases through three core constraints:
- Authentication: Senders (e.g., doctors) must use their secret keys to encrypt data. This prevents adversaries from generating valid "guess" ciphertexts without the sender's credentials.
- Designated Tester: The encryption process is tied to the public key of a specific cloud provider, ensuring only that designated server can perform comparisons.
- Granular Authorization: Receivers must issue a specific "trapdoor" token to authorize a test between a particular sender and target.
Technically, the scheme utilizes asymmetric bilinear pairings and is proven secure against chosen-ciphertext attacks (IND-CCA) under the Bilinear Diffie-Hellman (BDH) assumption.
Evaluation and Real-World Application
The performance of PKAE-DET was evaluated using the Java Pairing Based Cryptography (JPBC) library on standard hardware. We specifically applied the framework to the Diagnostic Related Groups (DRG) classification system used by the Centers for Medicare & Medicaid Services (CMS).
In the DRG process, patients are categorized for insurance purposes based on symptoms and criteria. PKAE-DET allows a cloud server to perform this categorization by matching encrypted symptoms against encrypted criteria without ever learning the actual medical content.
Experimental results indicate that the "Test" algorithm—the most frequent operation—scales linearly. While pairing-based cryptography introduces more overhead than standard AES, performing 800 tests in approximately 35 seconds is a viable trade-off for the high security required in healthcare.
Conclusion
PKAE-DET provides a robust solution for secure medical data matching by resisting insider attacks in a single-server environment. While the computational cost is higher than unauthenticated schemes, it offers the necessary protection for sensitive patient information. Future work will focus on optimizing performance for low-power IoT devices and exploring more flexible authorization patterns, such as range-based queries.
Summary updated