AI conversations in healthcare tend to gravitate toward the big possibilities: predicting disease, assisting clinical decisions or automating complex workflows. However, most of AI’s immediate value may be happening behind the scenes, helping interoperability teams solve the complex, time-consuming work required to keep healthcare data moving.
CliniSync, Ohio’s statewide health information exchange (HIE), is putting that idea into practice. As part of a broader modernization effort, the organization is using Rhapsody Axon ™ to help improve the quality and consistency of data flowing through a network that processes roughly 10 million messages each day.
During a recent Civitas Networks for Health webinar, the CliniSync team shared what that looks like in practice and what they are learning about where AI can make a meaningful difference in interoperability.
Moving Beyond “Getting the Data In”
CliniSync supports more than 150 hospitals, making consistency across incoming data feeds a significant undertaking. Historically, each interface was highly customized, requiring custom work on both CliniSync’s side and its data processor’s side.
As CliniSync modernizes its infrastructure, the organization is working toward a more standardized approach. Incoming data flows through the Rhapsody integration engine, where it is normalized before moving into the clinical data repository. Patient demographics can then be queried against the Rhapsody enterprise master person index (EMPI) to establish an enterprise identifier and connect records belonging to the same person.
Axon at Work at CliniSync
- 150+ hospital feeds moving toward a more standardized data model
- Massive time savings in designing and building integration workflows
- Less repetitive development to conform incoming data to CliniSync’s required schema
- No patient messages sent through an external LLM, keeping integration professionals in control
The challenge was how to do that efficiently at scale. Moving data feeds from more than 150 hospitals into the new environment meant conforming large volumes of existing data to a more rigorous schema. Tackling those variations interface by interface would require significant time and specialized integration expertise.
That’s where AI entered the picture.
Using AI to Tackle a Real Integration Challenge
Rather than sending protected health information to an LLM, the CliniSync team provided Axon with the requirements for how the data needed to be structured.
Working iteratively with Axon, the team developed an approach that checks messages against the required schema. When a message doesn’t conform, the logic can deconstruct it and reorganize the necessary segments and data into the expected structure.
CliniSync’s approach has resulted in a “massive” time savings by supporting the design and build of the integration workflow, rather than processing patient messages through an external LLM. Integration professionals remain in control of the work while using AI to help translate complex requirements into working logic.
For an HIE processing millions of messages every day, that ability to reduce repetitive development work can have an impact well beyond a single interface.
A New Model for Integration Work
CliniSync’s experience also points to a larger challenge facing healthcare IT teams: the volume and complexity of integration work continue to grow. Healthcare organizations are adding new applications, modernizing legacy infrastructure and preparing their data environments for increasingly sophisticated analytics and AI use cases. At the same time, the specialized expertise required to build and maintain healthcare integrations can be difficult to scale.
AI gives experienced integration professionals another way to approach that work. Embedded directly within Rhapsody and Corepoint, Axon is designed to help teams interpret requirements, navigate complex healthcare standards and support integration work across FHIR, HL7v2, X12 and API-based workflows. Teams can use it across the integration lifecycle, from building and optimizing integrations to troubleshooting and maintenance.
That doesn’t remove integration expertise from the equation. It gives that expertise more leverage. Instead of spending as much time interpreting specifications, searching documentation or repeatedly solving similar transformation problems, integration professionals can apply more of their expertise to the work that requires judgment and experience.
For CliniSync, that means applying AI to a very practical challenge: standardizing data across a large and complex HIE environment while reducing the custom work required to get there. At roughly 10 million messages a day, improvements in how that work gets done can quickly add up.
And that may be one of the most meaningful near-term opportunities for AI in healthcare interoperability: helping skilled teams spend less time on repetitive integration work and more time moving data where it needs to go.
