Every year, millions of individuals across the United States make planned (and sometimes unplanned) visits to their primary care physician, or urgent care, or a specialist to treat their medical conditions, using health insurance to help cover the costs of treatments and medicines. Many or most of these visits will entail some consumer payment, since most health insurance subscribers must pay an out-of-pocket maximum before insurance benefits entirely take effect.
This, in turn, requires accurate tracking and reporting of deductibles and out-of-pocket payments—known in the healthcare industry as “accumulators”—for all parties, including patients, providers, and payers. It’s especially important for consumers, who often contact healthcare providers to challenge or better understand charges they’ve been billed for. This is especially the case when they thought they had reached an out-of-pocket maximum or deductible.
A major healthcare organization in the United States faced an enormous challenge in providing a standard view of accumulators across both medical and pharmacy claims processing systems. Data synchronization was at least 24-hours out of sync among various systems, meaning that at times an individual’s deductible or out-of-pocket limit may have been met but not recognized in real time. This led to inaccurate charges and time-consuming reconciliation work.
The organization turned to the experts at Luxoft, an AWS Partner Network (APN) Advanced Consulting Partner and AWS IoT Competency Partner, to understand how it may be able to use blockchain technology to solve a decades-old data reconciliation problem and drive better outcomes for all parties involved. As a global IT service provider and consulting firm, Luxoft helps companies in highly regulated industries meet business challenges and drive innovation using advanced technologies.
The healthcare provider uses many different systems to process various medical and pharmacy claims. Each system is separately controlled by various payers, medical benefits and pharmacy benefit. After spending time with the organization to understand its business model and pain points, Luxoft identified the potential for blockchain, a distributed ledger that cryptographically links and secures records to one another via timestamps and other attributes, to unify the disparate systems managed within the healthcare organization.
“The payers systems aren’t unified,” says Todd Gehrke, blockchain solutions architect at Luxoft. “Claims are processed and billing is calculated using accumulator values that are synchronized using batch processes. We devised a method for the payer’s legacy systems to tie into a blockchain to share accumulator totals. We were targeting a use case that provided significant value with minimal disruption.”
By connecting systems in near real time using a private blockchain, Luxoft determined that the organization could share and update real-time accumulators in a secure, reliable, and auditable way.
Within weeks, Luxoft kicked off a proof of concept (PoC) with the healthcare organization. The team at Luxoft focused the scope of its PoC on self-insured entities engaging with the healthcare organization to put together healthcare plan options for employees.
To execute the PoC within a 4- to 6-week timeframe and build the technology stack for the blockchain solution, Luxoft made many working assumptions based on the organization’s business processes and restrictions.
“These assumptions helped us develop functional requirements for the blockchain solution,” says Gehrke, highlighting the role that data isolation played in choosing a technology stack. “Confidential transactions were paramount to what the organization was trying to do. We leveraged blockchain technology to give visibility and drive consistency while ensuring that nobody other than the entities that are covering the subscriber can see the data. That was key.”
The team chose to use The Linux Foundation’s Hyperledger Fabric running on AWS as the underlying distributed ledger technology stack because it allows for the creation of channels that provide a private network between two or more interested parties.
“We designed the solution so there would be a unique channel for each group of payers,” says Gehrke. “Channels also give us clean separation of accumulator values across plan years. Through scale testing, we discovered that performance is impacted when the number of channels exceeds 300. So to keep the proliferation of channels to a minimum we implemented plan year logic in the smart contracts.”
Defining Hyperledger channels based on payer combinations for the healthcare organization's PoC.
“Because this is a private blockchain solution, we needed to automate the deployment of a blockchain network that could meet the scale and fault-tolerant needs of the organizations involved,” says Gehrke. “We built a Hyperledger Fabric network inside of AWS using a number of AWS tools and services, such as Amazon VPC and security services. Given blockchain’s need to be distributed, we automated the build and deployment of a multi-region architecture of blockchain nodes across AWS. For the PoC, we decided that each region, or node, would represent a payer, enabling us to demonstrate isolation of private transactions within a payer grouping.
Luxoft decided to use a three-region deployment to run its load tests for the PoC. “This gave us realistic network latency benchmarks results. The results were pretty consistent, ranging from 80-150ms for reading and 200-400ms for writing transactions,” says Gehrke.
A multi-region topology on AWS designed for the healthcare organization's PoC.
Having to integrate blockchain into existing systems can become an obstacle for adoption. The challenge was especially acute for the healthcare organization. Much of what was driving the non-functional performance requirements for the PoC stemmed from legacy systems written in COBOL and running on mainframes, so integration needed to be a simple as possible. For these, Luxoft developed a RESTful interface to accelerate integration of the solution with the organization’s on-premises legacy systems.
Luxoft also had to consider the integration of blockchain into healthcare industry Business Process Management (BPM) platforms to increase efficiency. To that end, Luxoft created two adaptors. The first was built on Appian’s low-code platform, enabling users to visually model blockchain interactions. The second, fabric-jms-xmlcall, was built for the Pega platform, and is a standalone proxy service for calling fabric chaincode using XML protocol and MQTT transport. This made integration into any Java based software relatively simple.
“We had both functional and non-functional success critera. Our implementation met all the functional requirements, such as managing accumulators and data isolation. We surpassed the throughput objectives for batch processing by tuning the block size and write interval. We also reduced the window of systems being out of sync from 24 hours to 200 milliseconds,” says Gehrke.
He adds, however, that there is inherent latency introduced when using a blockchain.
“The shortcomings we experienced were due to the inherent latency of blockchain consensus and the archaic nature of the legacy systems we needed to integrate with,” Gehrke says. “We were integrating with mainframe code written in COBOL decades ago running business rules for claims adjudication systems. We faced a latency constraint. If a single transaction time was over 250 milliseconds, the system started having problems. The calls to the systems start getting backed up, and then they can’t auto scale. A better approach would be to rewrite all of the business rules for processing the claim into a smart contract, but the legacy mainframe is the hub of enterprise computing."
Following its testing, the team walked the healthcare organization through potential future-state developments and next steps it could take to test the accuracy and consistency of the blockchain solution.
A key lesson the Luxoft team learned through its testing was the power cloud-first organizations have to migrate to blockchain solutions. Complex solutions like claims process systems are always going to have off-chain processing and storage. Cloud services helps companies tolerate latency and auto scale.
The following is a blockchain claims processing architectures with off-chain processing and storage on AWS:
Luxoft sees massive potential for blockchain solutions deployed on AWS to drive a fundamental shift in how different entities and the healthcare industry interacts, and thus enabling better outcomes for individuals.
“Through our PoC, we helped the organization understand that in its end state. A solution like this isn’t simply an ‘inside the enterprise’ solution. It has the ability to empower cross-enterprise consortiums and open up new markets opportunities,” says Gehrke. “By using blockchain, organizations can now create a consortium solution to enable plans using payers from competing companies or even nonprofit organizations to work together. New opportunities for partnerships across organizations can help reduce cost, improve healthcare plan flexibility, and drive efficiencies within the different organizations.”
The following are some example architectures of a Hyperledger Fabric implementation on AWS:
A high-level two-organization Hyperledger Fabric architecture taking advantage of AWS services.
A high-level multi-organizational Hyperledger Fabric architecture taking advantage of AWS services.
Luxoft, an AWS Partner Network Advanced Consulting partners, uses new technologies and resources to help clients build promising blockchain solutions. Through its Enterprise Blockchain Accelerator (EBA), Luxoft provides organizations with R&D resources to develop proofs of concepts and accelerator platforms to incubate blockchain projects.
If you’d like to see the shared fabric-skeleton Luxoft built, the company has shared it as a tool for DevOps automation of Hyperledger Fabric on AWS to its public GitHub.