- Health IT infrastructure is getting in on potentially game-changing technology. Long derided for its inability to keep pace with sister industries, healthcare stands on the brink of a dynamic shift. Core capabilities leveraged by the likes of Instagram, Netflix and Apple — to name-drop a few ubiquitous cases — are starting to take root within the domain of providers, patients and payers.
Look to the U.S. west coast for an emerging healthcare example. Cal INDEX — a non-profit, statewide health information exchange backed by $80 million in private seed funding from Blue Shield of California and Anthem Blue Cross — gives providers online access to approximately 9 million health information records with three years’ worth of historical, claims and pharmacy data. The exchange, one of the largest population health management platforms in the world, may house upwards of 20 million records within the next 18 months.
The Cal INDEX system runs on a massively scalable, non-relational database called Cassandra, originally developed at Facebook in 2008 and subsequently open-sourced as a top-level project at the Apache Software Foundation in 2010.
Cassandra can handle huge amounts of data and thousands of concurrent users or operations per second, even across multiple data centers. In fact, Apple has 75,000 nodes deployed on the database to support the firm’s cloud services. Instagram uses it as the database backbone for its messaging platform, while Netflix applies the technology to ingest, analyze and provide recommendations to customers.
To get a sense of Cassandra’s potential for further build-out in healthcare, HealthITInteroperability asked integration workflow and real-time analytics expert Dave Bennett where things are headed. Bennett, currently executive vice president of product and strategy at software company Orion Health, formerly advised billionaire healthcare capitalist Patrick Soon-Shiong, MD, on software ventures ranging from genomic analysis and device integration to population health management.
“I got involved with all these distributed compute technologies because I was fascinated by how they could actually be capable of supporting real-time data stores for transactional systems. You might think in terms of an EMR system, but with the ability to do real-time analytics on the same platform. Workflows, transactional systems and analytics could all work on a single stack,” Bennett explained.
Instagram’s concept of posting a picture and message that can be seen by all followers in real time capitalizes on Cassandra’s fast-write/fast-read capabilities while fitting into a transactional model.
“If you think about the professional biometrics that we’re starting to do in healthcare, it has the same pattern,” said Bennett. “It involves fast writes to a system and then reads that might be posted to a circle of care — all of the people who are caring for a chronic patient, for example. Cassandra has the enterprise capability that we believe is going to be required in healthcare in the near future as the datasets get bigger and more diverse.”
Aside from gaining access to real-time claims information, providers will likely use the technology to help manage their value-based networks, according to Bennett. Similarly, payers will have access to provider-patient relationships in those networks and will use the data for member engagement or risk-adjustment purposes. In addition, “there are a lot of capabilities around having this liquid data in place” that will enable entities such as Cal INDEX to better manage their populations.
“We have to not only support the data sets of today, but the data sets of tomorrow,” continued Bennett. For instance, a new project at Cleveland Clinic will support genome variation data on the Cassandra platform and tie it to clinical data.
“We’re moving to a world of very precise medicine,” he concluded, “using variation, phenome, clinical and claims data to create the best care for patients. Providers are going to be looking at these technologies aggressively in the next few years. We need a system that looks like the Facebooks and the Apples of the world incorporating all these different data set requirements. We need to be able to support scale-down, scale-up capabilities in real time. As patients get more involved with their own care, they’re going to expect what they get from retailers and financial services providers today — and we need to replicate that.”
Photo credit: NASA/Terry Virts