M1 Thesis — Blockchain
 for Healthcare Security

Academic Context

Master 1 MIAGE — Université Paris Nanterre
Faculty supervisor: Prof. Marta Rukoz, Full Professor
Master’s director: Prof. François Delbot, Senior Lecturer
Industry supervisor: M. Rémi Masson, Manager — Audioptic Trade Services (Optic 2000)
Defended on 12 June 2025 · Grade: 18 / 20

Research Question

The digitisation of healthcare systems generates an unprecedented volume of sensitive medical data. These data must simultaneously satisfy three conflicting requirements:

• Availability: a physician must be able to access a patient record urgently, from any facility
• Confidentiality: health data are among the most sensitive under the GDPR (and HIPAA in the United States)
• Auditability: every access must be recorded and provable for regulatory compliance

Traditional centralised systems cannot satisfy all three properties without introducing a single point of failure or trust. The blockchain — a distributed, immutable ledger — appears as a promising avenue, provided the right consensus protocols are chosen and appropriate cryptographic mechanisms are added.

Method — Systematic Literature Review (SLR)

This thesis is a Systematic Literature Review (SLR) conducted following the PICOC methodology over the 2022–2025 period. The final corpus comprises 28 peer-reviewed articles from sources including IEEE Xplore, ACM Digital Library, and PubMed.

Research Questions:

Main Consensus Mechanisms Studied

The corpus analysis distinguishes five consensus families, each presenting distinct trade-offs:

Proof-of-Work (PoW) — well-known and robust, but energy consumption exceeds that of PBFT by two orders of magnitude in a comparable permissioned network. Unsuitable for healthcare.

Proof-of-Stake (PoS) — more energy-efficient, but security rests on staked capital, which is not relevant in a hospital consortium network without a native economic token.

PBFT and BFT variants — deterministic Byzantine fault tolerance, immediate finality (no forks), low energy consumption. The network tolerates up to f malicious nodes if n ≥ 3f + 1. Particularly suited to healthcare consortia where participants are known and registered. Throughput can vary by up to three-fold depending on the BFT variant.

Raft — classical non-BFT consensus, tolerating crashes in honest nodes but not Byzantine behaviour. Useful in tightly controlled closed environments.

Proof-of-Authority (PoA) — each validator is identified by a certificate (X.509). Simple, performant, suited to permissioned environments where identity is audited.

RQ1 Conclusion: in a permissioned healthcare network, BFT and PoA algorithms offer the best balance of security, performance, and energy efficiency.

Technical & Regulatory Challenges

Immutability ↔ Right to Erasure (GDPR)

The immutable nature of blocks directly conflicts with Article 17 of the GDPR (right to erasure). Two solutions emerge from the corpus:

• Chameleon hashes: hash functions allowing controlled rewriting of a block by a designated authority (e.g., the patient), without invalidating the chain
• Off-chain storage: raw medical data (imaging, reports) is encrypted and stored on IPFS or a secure cloud; only the content hash (CID) is anchored on-chain. Deleting the off-chain data satisfies the right to erasure without modifying the blockchain

Zero-Knowledge Proofs (ZKP)

ZK-Proofs allow proving compliance with a constraint without revealing the underlying data — for example: proving a patient gave consent for access on a given date, without revealing their identity or the accessed data. This addresses both GDPR confidentiality requirements and regulatory auditability.

Governance and Compliance by Design

The corpus highlights the need for a “compliance-by-design” approach (which we conceptualise as Dev-Sec-Law-Ops): integrating technical, legal, and clinical experts from the design phase, rather than treating compliance as a layer added after the fact. This entails:

• Traceability and auditability embedded in the protocol
• Clear access-rights governance (consent smart contracts)
• Cryptographically provable access revocation

Recommendations and Perspectives

The thesis concludes with recommendations for the transition to industrial prototypes:

1. Prefer PBFT or PoA for hospital consortium networks
2. Adopt a hybrid architecture: on-chain smart contracts + encrypted IPFS off-chain storage
3. Integrate ZKP for consent and access proofs
4. Plan for chameleon hashes from the design phase to handle the GDPR right to erasure
5. Conduct external audits and real-environment tests before any large-scale deployment
6. Involve lawyers, clinicians, and engineers from the scoping phase

Summary