Take off at dawn. Balloon Fiesta, NM 2018

Meet up with an Architect


The momentum seems real. A Decentralized Distributed Healthcare Ecosystem(DDHE) can be a reality. I started looking closer into blockchain and IPFS. And OMG, there’s so much info out there! So many startups and open source tool projects too. Healthcare organizations are now prototyping blockchain based projects and there is no shortage of literature, podcasts and Youtube videos praising the good that will come from a decentralized internet (See below references).

Its quite dizzying actually, not unlike gawking up at hundreds of unique giant balloons swirling in the wind, decorating the New Mexico dawn… Which coincidently, I took my kids to the Balloon Fiesta in New Mexico. It was amazing and I highly recommend it. But I digress…

While there, I met up with a good friend and colleague, Andrew Padilla, who like me, recently became an independent healthcare IT consultant. He’s collaborated with me as a senior architect on a few cloud and healthcare enterprise infrastructure projects. I consider him an IT thought leader with a pragmatic approach to problem solving and designing infrastructure. We met at Java Joe’s for breakfast, good java (the liquid kind), cool bohemian vibes, live New Mexican folk song balladeers, hangin at the “Breaking Bad” film location where Tuco’s hideout was and… discussing emerging tech… How dope is that!?! But again I digress…

From our past projects integrating enterprise systems and developing cloud platforms for AI, we experienced the known fact that cognitive analytics development requires lots of trusted quality data for training models, developing logic and testing to ground truth and validate an AI implementation. As pervasive as AI is becoming, for healthcare, the value of patient data is skyrocketing (see this). But if patient data is as abundant as it is reported to be, why is it so difficult to get or use? Some of the new healthcare startups seem to be focused on collecting patient data directly from patients and storing it in a new paradigm (IPFS and/or blockchain), which is great but what about all the data currently accumulating and stored in hospital’s legacy systems around the world? It seems obvious that hospital patient data is as useful and necessary as data directly obtained from patients for completeness, but likely because it’s so difficult to obtain, its not currently being addressed with the emerging tech by new startups.

We have experienced the difficulty in collecting and aggregating multi-siloed and multi-modal data to create longitudinal patient records for use by analytics and AI. And yes, it ain’t easy... Suffice it to say it requires a multi-disciplined iterative approach to resolve. I want to understand how this process can be improved with the new emerging tech paradigm.

Over green chile huevos rancheros and sipping americanos, Andrew and I shared and discussed what we learned from playing with Ethereum dev/test tools (Truffle & Ganache) and IPFS (including IPLD & IPNS). In discussing how we might implement certain healthcare IT use cases, it became evident that blockchain alone wouldn’t be enough. Though blockchain provides the immutable and secure distributed ledger and programmable smart contracts, it’s design is not meant for storing large amounts of data which is necessary for many healthcare use cases (ex. medical images or large amounts of structured and unstructured patient records). For decentralized and distributed storage with flexible accessibility to massive amounts of data, IPFS provides a content based addressing file system using hashing, which makes it immutable, permanent and can be secured by encrypting the file content. The pairing of blockchain with IPFS seems natural and shows great promise for developing the type of decentralized applications (dapp) that would constitute a complete solution for collecting, transforming, aggregating, storing and managing patient data from legacy system silos to be processed by and/or for analytics and AI. (Read an example of this from Core Health)

However as mentioned earlier, getting quality data is a struggle. A prerequisite for properly obtaining data from disparate silos, is having useful information about the data as well… Metadata! Things like the version of the format, the source system, the method of connectivity and/or interfacing and the business process or semantics that the data supports. Typically with system integration projects, this information would come in the form of an interface specification document or worse yet by word of mouth from an SME who luckily still works at the hospital. It seems dapp contracts could be programmed to reference a specific dataset to process as well as references to a metadata definition for proper parsing and mapping of the data to a desirable target model. We also realized in discussion, a metadata language could enable data governance, another necessary component for a successful DDHE, by ensuring the data meets certain quality levels (accuracy, completeness, regulatory compliance etc…). As well as enabling provenance, because knowing when and where data has been is essential for troubleshooting system flow failures. Coincidentally Andrew wrote an enlightening article describing, metadata as a language for the Internet, in Medium. Read it here.

Another “aha” moment during our breakfast was identifying additional features of dapp programmable smart contracts potential to store a reference to an ‘analytics deployment manifest’ (in IPFS) which would in turn reference a collection of containers to deploy to where data lives. This would solve the issue of moving massive amounts of PHI data outside the four walls of the hospital for processing (can you say HIPAA?).

Dapp smart contracts could also trigger other combinations of dapps, opening up a slew of possibilities in how functionality could be put together and deployed to build more complex capability. Since each dapp programmatically enables its own contractual and service level agreements, there is granular control built into the infrastructure that insures fair value and payment for each service. And it insures that all participants of the ecosystem, whether consumer or provider, will do what they say they will do. This inherent trust factor in the infrastructure of a DDHE is what potentially can make many healthcare interactions and use cases successful in the new paradigm, where in the old paradigm, they have failed due to lack of incentive or have produced income inequalities that result in distrust… but more on this later in future blogs.

Summarizing from a 3,000 ft balloon pilot perspective, here are some building blocks for a DDHE:

  • An existing open blockchain ecosystem that can enable and support smart contracts. We’re learning Ethereum tools to prototype dapps.
  • A distributed and decentralized filesystem that has quick access to immutable content. IPFS is the obvious choice for this and we are prototyping with IPLD and IPNS as well.
  • An open and extensible metadata standard for interoperability, data governance and provenance
  • Container, microservice and/or serverless architecture for deployment and operation of analytic applications.

I’m sure there is more, but this seems enough for now to explore, research designs and prototyping applications for a DDHE. Identifying a healthcare use case to walk through or prototype to understand where we are at with this emerging technology is important. In the next blog I’ll describe a use case that we will attempt to prototype as our first Ethereum dapp working with IPFS and utilizing a metadata definition. I’ll also log any progress we make researching the Truffles toolset for Ethereum dev/test, IPFS and whatever else grabs our attention.

How useful is all this? Who knows but for me… Its like taking off in a balloon for the first time at dawn not knowing where you may end up by the end of the day, but not really worrying because you know wherever you land you’ll likely be better off than where you were when you first started. So maybe, just enjoy the journey… Stay tuned.

References:

Saraj Raval (A brilliant, entertaining programmer, motivator and educator):

Podcasts:

New healthcare companies building on blockchain:

Articles: