Back in March 2017, John Halamka published “The Potential for Blockchain to Transform Electronic Health Records” with two MIT faculty in the Harvard Business Review. He suggested blockchain could serve as a a fourth model of medical data interoperability after push, pull and view. Push is sending data from one provider to another. Pull is querying information from another provider. View allows one provider to view data inside another provider’s record. All three approaches lack a formal consent process or standardized audit trail. Blockchain creates an append-only, time-stamped chain of content using a public-private key infrastructure. The model allows for
- medical records to be maintained within provider or health system electronic medical record systems without another intermediary (e.g., electronic medical record bank),
- electronic processes built on top of the time-stamped programmable ledger that could scale across different electronic medical record systems or different implementations of the same electronic medical record system, and
- a standard audit trail across the healthcare landscape.
Within the HBR article, John refers to a MIT project to facilitate medication reconciliation. In the project, the blockchain signature is intended to serve as the verification that the information is truly the latest information around the patient’s medication list. Patients could either manage their own records or delegate that information to a family member, caregiver or a service organization unaffiliated with the health systems or providers currently managing one portion of the patient’s medical records.
Tony Little and Soleh Ayubi outline some blockchain realities in healthcare in a Medium post. I believe the important “truths” for this post include:
1. Healthcare blockchains need tokens – “Without careful analysis of how to offer economic rewards participating entities (and/or punish bad network behaviour) you won’t have all the protocols in place to realize blockchain’s full potential.”
2. Consortia are super-complex, and necessary – The technology may be the easiest part of making blockchain work in healthcare. Parties with low levels of trust need to develop rules of engagement to help do more advanced coordinated processes that leverage blockchain’s structure than what they do currently without blockchain.
3. Blockchain does not solve interoperability by itself – blockchain does not address patient identity or reconcile care gaps. Blockchain can facilitate interactions, but the technology can only enable what healthcare players have already agreed to perform. To quote Mike Jacobs at Optum, blockchain is only four parts: a distributed database, an append-only structure, smart contracts and incentives.
So what existing or emerging technologies or standards might be deployed/widely adopted in the next three-to-five years to support higher-quality healthcare transactions? I can envision the following:
- Member/Patient identity verified with a DirectTrust address. The address could be linked with a patient portal aggregator to facilitate members/patients logging into a single website instead of different portals for each healthcare provider. The aggregator should also be an “intelligent endpoint” to trigger downstream actions through synchronous or asynchronous triggers. (Disclaimer: I’m a board member for DirectTrust.org. Although the architecture could be used to support NIST‘s Identity Assurance Level 3, Authentication Assurance Level 3 and Federation Assurance Level 3, not all DirectTrust members are operating at the highest assurance level for identity, authentication and federation.)
- Blockchain linked to the DirectTrust address to serve as the member’s/patient’s Record Locator Service for encounter- or document-level consent permissions. Records remain where they were created (usually in a provider’s office or within a hospital or health system electronic medical record system). This approach minimizes what needs to be stored in the blockchain, helping the technology scale as more users adopt it.
- SMART on FHIR for real-time queries (pull, view) to each medical record source system based on a specific trigger (e.g., patient/member registers at a healthcare facility, decision support or prior authorization [CDS Hooks ])
- Direct messaging (push) to manage asynchronous interactions between healthcare entities (e.g., updating the member’s/patient’s blockchain when after a healthcare interaction)
- Parsing unstructured data
- Natural language processing and machine learning to extract concepts from free-text documents
- An ontology (to be determined) that can help organize the member’s/patient’s health information in ways that support decision making for diagnosis, treatment or surveillance