In the first two parts of this series, I laid out where blockchain + AI actually delivers value, and the scalability tricks that make it possible. In this final part, I want to ground it all in a real, live national project: South Korea’s medical mydata initiative, 건강정보 고속도로 (Health Information Highway, also called MyHealthWay).
This is an interesting case because it sits exactly where blockchain could matter — patient data sovereignty, sensitive consent, multi-party trust — but the actual system uses no blockchain at all. Understanding why is more illuminating than any abstract argument.
Korea’s MyHealthWay is run by the Ministry of Health and Welfare through the Korea Health Information Service. It’s a centralized data intermediation platform that lets patients view their own medical records across institutions and consent to share them with third parties — insurance companies, research institutions, health apps.
By 2025, all 47 of Korea’s top-tier general hospitals are connected, with total participating institutions targeted at 1,263 (including local hospitals and clinics).
The architecture is straightforward:
It works. And it works well enough that the question becomes: why would you add blockchain?
Despite the centralized architecture working, three weaknesses are worth examining.
1. Single point of trust. The government is the trusted hub. If the central servers are breached, all the consent and access records are vulnerable. If a future administration changes data policy, patients have limited recourse. This is structural, not a criticism of current operations.
2. Audit trail integrity. Today, a patient who wants to know “who accessed my data and when” has to trust the government’s log. If there’s ever a dispute about whether consent was given, the government’s records are the authoritative source — and the government is also one of the parties.
3. Cross-border applicability. A Korean patient seeing a doctor in Bangkok or Boston can’t use MyHealthWay — the trust authority is national.
For each of these, blockchain offers a structurally cleaner answer.
Consent ledger. Every time a patient grants or revokes consent, the hash of that consent record gets written to a chain. The actual medical data never goes on chain. But the fact of consent — its scope, timestamp, and revocation history — becomes tamper-proof and independently verifiable. If a dispute arises later, the consent record is mathematically unforgeable, regardless of who controls the storage.
Decentralized Identity (DID). Currently, identifying yourself at a Korean hospital means presenting your national ID or going through phone-based authentication. A DID-based approach lets you hold your medical identity in your own wallet and present only the specific attributes needed: “I am over 40” or “I am insured under this policy,” verified cryptographically without revealing your national ID number. The European Union’s Digital Identity Wallet mandate, going live in 2026, follows this model.
Research data attribution. MyHealthWay explicitly contemplates patients “donating” data to research, drug development, and treatment innovation. Blockchain enables a richer model: patients can contribute anonymized data to research projects via smart contracts that automatically distribute compensation based on how the resulting drug, study, or AI model uses the data.
There is a hard problem here that blockchain advocates often skim past: the right to erasure.
Korea’s Personal Information Protection Act (개인정보보호법) Article 36 grants individuals the right to delete their personal data on request. The EU’s GDPR has the same “right to be forgotten.” These laws encode an important value — your data shouldn’t follow you forever against your will.
Blockchain’s defining property is immutability. Once recorded, data cannot be deleted.
This is a direct collision. And the standard workaround — putting only hashes on chain, not personal data — works legally and technically, but only if you’re disciplined. The moment a developer puts even a name or email address on chain “to make things easier,” the entire system runs afoul of the law.
This is one major reason Korean medical institutions have been cautious about blockchain adoption: not because the technology doesn’t work, but because the legal architecture isn’t yet settled.
A homegrown company has been working in this space for nearly a decade: MediBloc. Their MediPass service connects hospitals and insurance companies for automated insurance claim submissions, deployed across Seoul National University Hospital, Samsung Medical Center, Severance Hospital, and others. They’re gradually transitioning the system to DID-based identity verification.
What’s notable about MediBloc’s approach is that it doesn’t try to replace MyHealthWay. It runs alongside, focusing on the trust-and-verification layer (signing, identity, claim authenticity) while leaving the data collection and standardization to the government platform.
This is probably the realistic future shape: government-run data infrastructure plus a private-sector blockchain layer for portable identity and verifiable consent. Not “blockchain replaces government” but “blockchain complements government on the parts where centralized trust is awkward.”
Will blockchain become the backbone of Korea’s medical mydata system? Almost certainly not. The government infrastructure is already in place, it’s faster and simpler, and switching costs are enormous.
But will blockchain quietly become embedded in specific corners? I think yes. Watch these areas:
The pattern is clear: blockchain wins where the trust problem is genuinely multi-party or cross-jurisdictional. It loses where a single trusted operator works well.
Across these three articles, the same theme keeps surfacing. Blockchain + AI is real, but it’s not universal. Its strongest applications are narrow, specific, and often unglamorous.
The interesting work isn’t in finding new things to put on blockchain. It’s in honestly diagnosing where the trust problem is, and asking whether decentralizing actually solves it better than the alternatives.
For Korea’s medical mydata system, that lens suggests blockchain is a complement, not a replacement. For content provenance, blockchain is a backup to PKI, not a substitute. For AI training data attribution, blockchain plus zero-knowledge proofs may genuinely be the only workable model.
The technology is here. The remaining work is institutional, legal, and frankly diplomatic — getting different parties to agree on what shared infrastructure should look like. That’s slower and less exciting than launching a token. But it’s where the real value is built.
This concludes The Blockchain + AI Reality Check series. If you’re a Python developer thinking about how to actually work in this space, our companion series “From Python to Web3” walks through portfolio projects, language choices, and a six-month learning roadmap.
Written by ChainLab — Chain Within Your Life. We build blockchain infrastructure for the real world. chainlab.dev
Data on Chain — Part 3 of 3 · ChainLab The first two parts were…
Data on Chain — Part 2 of 3 · ChainLab In Part 1 we settled…
Data on Chain — Part 1 of 3 · ChainLab Here is a tempting idea.…
Korea's tech is world-class, so why isn't AI payment running smoothly here yet? The answer…
You've probably used "safe payment" on a secondhand marketplace. AI payments need this too. The…
If last part's x402 was a vending machine, these two approaches are closer to a…