OpenAdaptThe right-record check: it stops on look-alike patients
Before it types into a record, OpenAdapt confirms the record matches the one it was asked for, and stops when it can't tell. In our look-alike test set (same name and birth date, near-identical record numbers), it stopped 5 of 5. It let the right patient through and stopped a plainly different one. It also stops some right patients.
Updated 2026-10-18
The test cases
| Case | Record numbers | Result |
|---|---|---|
| A letter O for a zero | MG4408 and MG44O8 | Stopped |
| A lowercase l for a one | MG4118 and MG41l8 | Stopped |
| All digits, with letter O's | 100512 and 1OO512 | Stopped |
| A dash-formatted number | Look-alike characters with dashes | Stopped |
| Same name and birthday, a different clear number | MG5439 and MG7263 | Stopped |
| The right patient | A clear, matching number | Let through |
| A plainly different patient | Different name and number | Stopped |
The limits
- On pixel-only screens it stops whenever the right patient's own record number has look-alike characters: 47 to 61 in 100 correct-patient checks on our test corpora (re-measured 17 Oct 2026). Those are extra stops, not wrong entries. With a confirmed list of the system's record numbers, a rendered record-list test's stops fell from 176 to 3 in 280 clicks.
- It covers only the steps that carry a patient's identity, not every click.
- These are test cases on rendered rows, not results from a real EMR.
Related: How it works: the right record · filing reports to the right chart
Technical details
The check is verify_target_identity in openadapt-flow. Look-alike identifiers make the OCR tier abstain (halt) rather than verify; a clear mismatch returns a mismatch verdict. Local OCR, no network calls. The false-stop range is the current matcher on the v1 and v2 string corpora, re-measured 17 Oct 2026 (docs/validation/IDENTITY_EVIDENCE.md in packages/flow). The known-ID list (runtime.known_ids, openadapt-flow 2.2.0) lifts a look-alike stop only when exactly one listed record number fits both reads (benchmark/known_ids/KNOWN_IDS.md).
Sources
- OpenAdapt look-alike test set (synthetic rows), 2026: 5 of 5 look-alike patients stopped; of the 2 control cases, the right patient was let through and a plainly different patient was stopped. On pixel-only screens the check stops whenever the right patient's own record number has look-alike characters (O and 0, l and 1); on our test corpora that was 47 to 61 in 100 correct-patient checks, re-measured 17 Oct 2026 (IDENTITY_EVIDENCE.md). With a confirmed list of the system's record numbers (openadapt-flow 2.2.0), those stops on a rendered record list fell from 176 to 3 in 280 clicks whose neighboring row wasn't a look-alike twin (KNOWN_IDS.md).
See it on your process
Bring one task your team types every day. We'll tell you whether it fits, and what a test on your own system would cover.