- Healthcare
- Systems Modernization
- Discovery Workshop
- Cybersecurity
Off brittle HL7 feeds, without a weekend of downtime
A provider network integrating with labs, imaging and partner systems over point to point HL7 v2 feeds pays a custom mapping for every new partner and an outage every time one changes. Phased migration to FHIR ends that, and clinical operations never stop.
Representative engagement. This describes a pattern we build rather than one named client: the situation that produces it, how we approach it, and the range of outcomes that kind of work lands in. Figures are stated as ranges or targets, never as a measured result for a specific customer. Our named client work is on the work index.
- Industry
- Healthcare
- cutover pattern, old and new paths running in parallel
- Zero downtime
- shared terminology layer instead of one per partner
- 1 mapping
The problem
Each integration is its own dialect. A partner changes a segment, a feed breaks quietly, and somebody notices when a result does not arrive. Adding a partner means writing another mapping, so the cost of the next integration is roughly the cost of the last one, forever. The obvious fix is a modern interoperability layer. The obstacle is that none of this can stop. Clinical systems do not get a maintenance weekend, results cannot queue up unread, and a cutover that goes wrong is a patient safety incident rather than an incident report. So the constraint shapes the plan: whatever replaces the current feeds has to run beside them, prove itself on real traffic, and take over one partner at a time.
What we built
A FHIR facade in front of the existing systems, introduced one interface at a time. Each partner is moved individually, with the old feed still running and both paths compared on live traffic until the new one is boring. Shared terminology and identity mapping sit in one place instead of being re-solved per integration, which is what turns the eleventh partner into a configuration task rather than a project. Feeds are monitored actively, so a broken interface pages somebody rather than being noticed by its absence. Security and access control are part of the migration rather than a follow up phase: scoped access per partner, full audit logging, and a documented answer for who can read what, which is the question a compliance review opens with.
What changed
Partner integrations stop being bespoke. The first few cost what they always did, and each one after that costs less, which is the actual return on an interoperability layer. Every interface moves behind a switch with the previous path still live, so the cutover pattern is a reversible step rather than an event. That is the property clinical leadership cares about, and it is why the work is sequenced this way rather than done in one release.
Built with
- Go
- FHIR
- HL7 v2
- PostgreSQL
- Kubernetes