Blockchains in Healthcare

Healthcare Data and Databases AIMed 2016

E. Kevin Hall, M.D.

Yale University

Healthcare Today

We must reinvent healthcare to be preventative.
Current state of art: Take point measurement (particularly if symptomatic) and compare to some clinical study (often from some time ago) on a "representative population".
If outside a threshold, some decision is made.
This is what sensitivity and specificity are based on.

A flawed idea.
Doesn't fit with the idea that we all have our "unique" genetic code.
And while genomics is important, we know twins vary in phenotypes: more encoded into our physiologic states than simple genomics.

Digital: No. Accessible: No

Digital: Yes. Accessible: Still No

Our next problem.

The promise of digital health data is more than this.

It's greater than isolated medical records on some hard disk.

The promise is accessibility and interoperability. Our data should work for us; our patients' data should work for them.

And no preventable disease should lead to death.

Patients' records should be consistent and available across institutional boundaries.
Access should be strictly determined by the patient him/herself.
When shared, all parties should be able to understand the data with a goal of increasing healthcare quality.

	Accessible	Interoperable	Canonical	Patient "Owns"
EHRs	+ (Locally)	+ (Locally)	+ (Locally)	-
USB Drives	+	+/-	--	+
"Independent" Cloud	+/-	+/-	+/-	-

A shared, distributed, immutable ledger for transactional data that establishes accountability and transparency.

On Distributed Communications, Paul Baran 1964

A Blockchain is a distributed transaction ledger.
Composed of "Blocks" which are time-similar collections of transactions.
Blocks are immutable, integrity verified - each block contains a cryptographic hash.
Hash of a block in the chain is dependent on the prior block. Chain is immutable.

A blockchain does not depend on a trusted central authority like a bank or hospital EHR.
Instead the blockchain is distributed to all nodes which participate in the network.
Because no central authority validates the blockchain, a network consensus must be reached.

Blockchains promise to free our data from single data silos.
In an environment of tailored personal health data, blockchains create a single canonical dataset dependent on time, but independent of third parties.
With single, longitudinal time series of very personal data, can reduce personal false positives.

Patients are seen at multiple institutions. In their points of view, their records are a sequential series of events irrespective of location.
Furthermore: later events can be based on earlier ones. A bad PFT might lead to prescription of an inhaler - and its prescription is dependent on the first event.

First, data is cryptographically secure.
Much like in banks, access is awarded appropriately and as needed.
But more: The chain is immutable, later blocks depend on the accuracy of those earlier.
In sync, distributed data: there is not any single point of failure more important than any other.

Interoperability with FHIR Profiles

Miner election is distributed evenly.

We have not achieved the true value of digital health data.
There is currently no single canonical data ledger.
Data is hidden from owners inside EHRs; owners do no effectively own their own health data.
Data owners are not the arbiters of data access.

It is likely that blockchains can answer these limitations.