lona2010 - Fotolia
From the advent of big data, healthcare has been one of its most promising domains. Healthcare generates oceans of data day in and day out, and much of it is ripe for exploitation by machine learning.
It's not all that easy, of course. Most healthcare data, from electronic health and medical records to patient registries and histories, grew up organically -- from custom in-house systems -- and is, thus, largely standards-free. That makes it very difficult to bring data of any particular kind under one roof.
If you can't bring all the data to you, you have to go to the data. And that introduces the even bigger problem: security. Both the access to and analysis of the data must be HIPAA-compliant. And that problem looms largest over the one kind of healthcare data that is, by far, the richest source of insights and business value: claims.
Healthcare claims are, of course, deeply integrated with other healthcare data. A patient's claims history lives alongside her treatment history; it reflects her demographic and socioeconomic profile, and it articulates patterns of recurring health and illness where chronic conditions exist.
But even without these relationships, databases of healthcare claims data containing the histories of many patients are of enormous value: They can be mined for correlations between illnesses and a vast range of behaviors, seemingly unrelated conditions and demographic factors to identify early warning triggers that can be used to bolster preventative care management. You just need access to as much of this data as possible.
The problem, then, is accessing a number of claims sources in a way that's secure enough to satisfy all stakeholders. And one security technology rises above all others to meet this challenge: blockchain.
The initial promise of blockchain
Over the past decade, blockchain's features have become well-known, if not well-understood. Physically, a blockchain is a distributed network of servers, peers that maintain immutable ledgers of transactions, persisting high-trust transactions between parties transparently.
Access to data secured in a blockchain is under its owner's absolute control, and smart contracts can be embedded to control what information is shared, with whom and under what circumstances. Data protected by a blockchain is hack-proof, and its authenticity can be verified. These features match up exceedingly well with the demands of healthcare data governance.
Blockchain's application potential in healthcare is broad:
- It can be used to unify patient health records into single, longitudinal entities.
- It can improve EHR and EMR interoperability.
- Its smart contracts can improve supply chain management by monitoring supply-demand cycles.
- It can be used to improve tracking in clinical trials.
- And it can be a solution to the claims analysis problem.
During its lifecycle, a single healthcare claim could be touched by as many as 300 different people, according to Siva Kannan, vice president of engineering at Gem, a Capital One partner company doing blockchain work in medical information tracking. The claim is endlessly passed from system to system, vulnerable in any number of ways, HIPAA precautions aside.
The ideal is to secure existing repositories with blockchain, implementing its access controls and embedding smart contracts to fine-tune the use of healthcare claims data for the purpose of big data analytics, ensuring patient anonymity and limiting the data items used.
This is an attractive scenario, but there are issues right out of the gate. The biggest is a general blockchain weakness -- not specific to healthcare: It scales very poorly. Because of the computational burden of its hashing scheme -- to keep the chain secure -- and the necessity of copying each full ledger to every network node, it can only handle a limited number of transactions, which is currently less than a dozen per second.
The expense of implementing and administrating a blockchain is another general weakness. Even a modest network must include a considerable number of servers to service widely disparate big data sources. And there is considerable complexity involved in building the blockchain initially.
A healthcare-specific issue is the lack of standards in healthcare claims formatting and processing. Succeeding in establishing a blockchain that can securely access the data necessary for claims mining doesn't solve this issue; it's just another barrier hiding behind the others.
There are others -- lack of universal use cases, lack of regulatory guidelines, legacy infrastructure barriers and a shortage of expertise -- but most of those are self-correcting with widespread adoption. The challenge is in handling those that aren't.
Several promising trends and possibilities should be considered.
The first is the increasing potential of analytics and machine learning to shift the healthcare industry from treatment to prevention. This is the most tectonic shift upcoming in getting the system under control, and predictive systems derived from healthcare claims data analysis is a big ripple in that shift. Put another way, there's a great deal of motivation to make this work, and the ROI justifies the investment.
Another is the fact that the smart contract mechanism in blockchain solves, in part, the standards issue: Yes, claims differ in format between companies, but smart contracts must be created and embedded in blockchain transactions anyway. It makes sense to negotiate the data disparities at that point, even if the solution is inelegant.
And a huge step forward is blockchain as a service (BaaS). The major cloud players -- Microsoft Azure, IBM Cloud, Oracle Blockchain and Amazon Web Services -- are investing in BaaS products that will alleviate much of the pain of implementation and administration described above.
That leaves scalability, the most difficult barrier. A number of incremental solutions are already in play -- Google plasma, sharding, proof-of-stake -- but the most promising for claims mining is off-chain state channels. These are a means of connecting a blockchain user with the protected data source that establishes direct contact between them after the blockchain has verified the legitimacy of the exchange. The concern that the direct connection is not as secure as the chain itself is mitigated by the ability of smart contracts to revert the exchange back to the blockchain if arbitration is required.
The future of both claims mining for preventative healthcare solutions and blockchain as the way to make it happen are promising. The learning curve is daunting, but the payoff is not only substantial in ROI -- it changes the game, opening the door to a new way of thinking about healthcare. That door is within reach.