Engagement snapshot
- Mandate
- Reduce reconciliation load without compromising patient result integrity.
- Timeline
- 12 weeks across core lab workflows.
- Team shape
- Lab operations lead, interface manager, 2 automation engineers, and 3 senior lab coordinators.
The problem
Orders, specimens, and results often mismatched across EHR feeds, LIS records, and courier updates, forcing manual reconciliation and delayed reporting.
What we built
Built a reconciliation engine that matched orders across systems, highlighted ambiguity classes, and opened work queues only when human judgment was needed.
Operating context
The network had multiple interfaces and courier touchpoints, which made mismatches inevitable. The real issue was that resolution work was too manual and too opaque, leading to preventable reporting delays.
Key constraints
- Patient and specimen integrity could not be compromised for the sake of throughput.
- System identifiers were not always consistent across the order lifecycle.
- Lab staff needed the workflow to surface ambiguity clearly rather than overconfidently auto-match edge cases.
What we built
Cross-system matching engine
Compared orders, specimens, and result events across EHR, LIS, and logistics feeds using deterministic and probabilistic matching.
Ambiguity classification
Separated clean matches from risky cases so humans only reviewed files where true judgment was needed.
Operational status layer
Made specimen and order state visible across the workflow so delayed or broken paths were easier to spot.
Delivery path
Reconciliation pattern audit
Studied historical mismatch types and identified where automation could safely handle resolution.
Controlled lab pilot
Launched in a subset of high-volume tests to validate match quality and reviewer queue design.
Network deployment
Expanded with monitoring on mismatch class frequency and delay reduction.
Why it mattered
The workflow removed a large amount of manual reconciliation effort without pushing risky cases through blindly. The biggest gain was reducing delay-causing ambiguity and making unresolved cases visible sooner.
Implementation notes
- Healthcare reconciliation systems need explicit uncertainty handling, not hidden confidence assumptions.
- Operational visibility can be as valuable as raw automation rate.
- Interface inconsistencies are manageable once the reconciliation model is designed around them directly.