Every Scalpal patient is a FinchNode subject. FinchNode’s demo API (https://api.finchnode.com/demo/v1, no key) serves a fixed set of synthetic scenarios. None of them has an acute surgical problem, so every case is the real chart plus an authored acute presentation, kept separate and labeled as authored.

The demo patients

The rest (rate limited, consent revoked, partial consent, a source unavailable) exist to test error handling. The partial-consent chart has no name, so the headset hides it. Run npx tsx $T chart <subject> to see exactly what Scalpal builds for a subject.

What Scalpal reads

Demographics (age at the chart’s dataAsOf), live medications, conditions, allergies, the latest eGFR, creatinine, hemoglobin, A1c and platelets, consent, and the latest vitals for the coach’s VR baseline. Documents, claims, immunizations and notes are ignored. Labs match on LOINC codes; medications, conditions and allergies match on names, so a brand name like “Eliquis” won’t raise the bleeding flag.

Risk flags

A risk only becomes a Time-Out chip if the procedure pins it to a step, and an interview must let the learner elicit every one of them.

New patients

  • A subject FinchNode adds works fully: snapshot its record and scenario into the package’s fixture/ folder.
  • An invented subject works offline only. The live service lists only what FinchNode serves, and the headset’s editor checks count patients exactly.
  • Sandbox u_ subjects reuse the case of the demo scenario they mirror; don’t author against them.