AuroraMed v0.2.0 is a synthetic-data pilot. See exactly what's built →

Identity, registration and scheduling · C0551

How does AuroraMed find duplicate patients, and can the matching weights be tuned?

In the pilot (v0.2.0, synthetic data)

Short answer

How would it be tested? The specification says a test set of 1,000 seeded duplicates reports recall and precision; algorithm version is recorded with every decision. That is the check for REQ-0204: the system must support deterministic and probabilistic patient matching with configurable weights and phonetic/fuzzy name matching, and must present candidates with scores. The pilot covers part of this. The rest is designed, not built. What v0.2.0 does today: Deterministic + weighted probabilistic (Soundex/Jaro-Winkler style) matching with live duplicate warning; weights not yet per-tenant configurable.

This exists in the synthetic-data pilot only.

Specification item
REQ-0204
Specification priority
P0 (of P0 to P3)
Tiers
Small, Medium, Large

Status as of September 30, 2026. Version 0.2.0, synthetic data only. See what's built today. Not legal, medical or security advice.

Related answers

See it in context: Patient identity and master patient index · Search the help center · Ask a question

See the synthetic-data demo first.Request a demo