Blockchain-Based Bidirectional Updates on Fine-Grained Medical Data

Chunmiao Li, Yang Cao, Zhenjiang Hu, Masatoshi Yoshikawa

Abstract

This paper proposes a decentralized architecture for sharing medical data that enables stakeholders—including doctors, patients, and researchers—to access and update specific authorized segments of a medical record. By integrating blockchain-based smart contracts for fine-grained access control with bidirectional transformations (BX) for data synchronization, the system ensures that updates to discrete "data slices" are accurately propagated to the full medical records maintained locally by each peer.

Introduction

Sharing electronic medical records (EMRs) is critical for collaborative healthcare and medical research. However, existing systems face significant challenges regarding privacy and data consistency. Traditional EMR systems often treat records as monolithic entities; granting a researcher access to a specific attribute, such as medication mechanism, may inadvertently expose sensitive clinical notes or personal identifiers. Furthermore, maintaining synchronization across fragmented data views is technically complex. If a physician updates a dosage in their local view, propagating that change to a patient's view or a researcher's dataset typically requires manual intervention or centralized oversight, both of which introduce risks.

Proposed Architecture

The proposed system adopts a "local-first" approach where comprehensive medical records are stored in users' local databases, while the blockchain maintains only metadata and permission structures. This design avoids the privacy risks and performance bottlenecks associated with storing sensitive data directly on a public ledger.

High-level system architecture showing the interaction between client-side users, server-side applications, and the blockchain.

The architecture relies on two primary components:

  1. Fine-Grained Access Control via Smart Contracts: Permissions are managed at the attribute level rather than the file level. For instance, a doctor may have authorization to update "Dosage," while a patient might only be permitted to modify "Clinical Data."
  2. Bidirectional Transformations (BX): Utilizing "asymmetric lenses," the system implements get (forward) and put (backward) operations. The get operation extracts a specific data slice (a "view") from the master record (the "source"). The put operation merges modifications from a view back into the source while ensuring global consistency.

Data Synchronization Workflow

The system manages data integrity through a structured process for Create, Update, and Delete (CRUD) operations. When a user attempts to modify data, the request is handled as a blockchain transaction, ensuring an immutable and auditable history.

The 7-step workflow for executing Create, Update, and Delete operations within the decentralized system.

As illustrated in the workflow, the process involves:

  1. Local execution and transaction submission to a smart contract.
  2. Permission verification by the smart contract.
  3. Notification of sharing peers upon authorization.
  4. Fetching updates and executing BX programs to synchronize local master records.

This mechanism prevents update conflicts by leveraging the chronological ordering of the blockchain; only one transaction for a specific data element can be validated per block.

Discussion and Future Work

While the proposed architecture addresses key privacy and consistency issues, several challenges remain. The throughput of public blockchains like Ethereum may limit the speed of non-urgent updates. Additionally, the model currently lacks an explicit incentive mechanism for nodes to process medical transactions. Future work will involve implementing a prototype and conducting experiments with de-identified real-world patient data to evaluate computational overhead and performance on various hardware platforms.

Conclusion

By combining the immutability of blockchain with the synchronization capabilities of bidirectional transformations, this research provides a robust framework for fine-grained medical data sharing. The "local-first" architecture prioritizes patient privacy while ensuring that all authorized stakeholders maintain a consistent and up-to-date view of relevant medical information.

Summary updated

BibTeX
@inproceedings{li2019blockchain,
  author = {Chunmiao Li and Yang Cao and Zhenjiang Hu and Masatoshi Yoshikawa},
  title = {{Blockchain-Based Bidirectional Updates on Fine-Grained Medical Data}},
  booktitle = {Workshops @ ICDE},
  year = {2019},
  url = {https://openreview.net/forum?id=m1lxDjcOO-mI},
}
Figure