TL;DR
This paper introduces E-Tenon, a novel scheme for secure and privacy-preserving Electronic Health Record (EHR) sharing. It achieves this by storing most EHR data in an open database, using multi-level attribute-based encryption (ML-ABE) for fine-grained access control, and a multi-signature scheme for data integrity and authenticity, significantly reducing computational overhead compared to traditional encrypted database approaches.
What Problem Does This Paper Address?
The increasing adoption of Electronic Health Records (EHRs) in healthcare brings numerous benefits, but also significant challenges in securely sharing sensitive patient data. Existing solutions often rely on encrypting entire databases, which is computationally expensive and hinders efficient data analysis for research and improved healthcare practices. Furthermore, while Attribute-Based Encryption (ABE) offers fine-grained access control, naive implementations struggle with multi-level access policies and can lead to attribute duplication and high overhead. There's a critical need for a system that allows EHRs to be readily accessible for analysis while ensuring robust privacy, integrity, and authenticity, especially in an open database environment.
Methods
E-Tenon tackles these challenges by integrating two key cryptographic techniques:
- Multi-Level Attribute-Based Encryption (ML-ABE): This is an extension of Ciphertext-Policy ABE (CP-ABE). Instead of encrypting the entire EHR, E-Tenon encrypts "pointers" to EHR blocks. These pointers are encrypted with access policies defined by the patient, allowing for fine-grained, multi-level access control. This means different users with varying attributes can access different levels of detail within the EHR.
- Multi-Signature (MS-BN): To ensure data integrity and authenticity, E-Tenon employs a multi-signature scheme. This allows multiple parties (like the data owner and a service provider) to jointly sign the EHR data. This mechanism prevents any single entity from unilaterally altering the data and provides a verifiable record of who agreed to the data's content.
The core innovation lies in how these are applied to an open database (Tenon Database - TDB). Most EHR data is stored in plaintext, with only the relationships between data blocks (pointers) being encrypted. Data is also preprocessed and shuffled to further obscure relationships and enhance security.
Datasets
Not discussed in the paper. The paper focuses on a theoretical framework and simulation-based performance evaluation rather than training or testing on specific EHR datasets.
Key Insights and Ideas
- Open Database for EHRs is Feasible: The most striking insight is that EHRs can be stored in an open database without compromising privacy, by cleverly encrypting only the relationships (pointers) between data blocks and employing data shuffling. This drastically reduces the computational burden of traditional encrypted databases.
- ML-ABE for Granular Control: Using ML-ABE allows for sophisticated, hierarchical access policies. Patients can define precisely who sees what level of detail in their EHR, moving beyond simple "all or nothing" access.
- Multi-Signature for Collaborative Trust: The multi-signature scheme is crucial for establishing trust between data owners and service providers. It ensures that both parties agree on the integrity of the data before it's stored, preventing manipulation and providing non-repudiation.
- Efficiency through Selective Encryption: By encrypting only pointers and using an open database, E-Tenon significantly reduces the computational cost of encryption and decryption, making it more practical for real-world healthcare systems, especially in emergency situations.
Key Results
- Performance: E-Tenon demonstrates significantly faster signing and verification times (approx. 2.34 ms) compared to a benchmark scheme (Zhang et al. [16], approx. 43.50 ms for signing, 25.84 ms for verification). Decryption is also faster (approx. 761.28 ms for E-Tenon vs. 1249.28 ms for Zhang et al.).
- Security: The scheme is proven to be selectively CCA-1 secure against confidentiality and privacy threats, and MU-UF-CMA secure against integrity and authenticity threats, inheriting these properties from its underlying ML-ABE and MS-BN components.
- Storage/Communication: The use of an aggregated access structure and selective encryption leads to smaller ciphertext sizes and reduced communication overhead compared to schemes that encrypt entire data structures.
Limitations and Future Work
- Limitations:
- The scheme relies on the security of the underlying ML-ABE and MS-BN schemes.
- While the TDB is "honest-but-curious," it still has access to metadata and encrypted pointers, which could potentially be exploited if the security of the encryption is compromised.
- The paper acknowledges that the multi-signature process requires an extra round of communication compared to single signatures.
- Future Work:
- The authors suggest exploring the use of Fog Computing to further enhance efficiency by outsourcing EHR preprocessing and decryption tasks to edge devices.
- Investigating how to further improve efficiency while maintaining strong security.
Summary updated