Skip to main content
AI clinical knowledge

Approved knowledge, in natural language

Lumen allows clinicians to ask questions about approved institutional protocols, procedures, diagnostic criteria, and reporting standards — and get an answer grounded in a visible, approved source.

What it can discuss

Grounded answers, not generated policy

The system retrieves and explains approved clinical knowledge. It distinguishes facility-specific criteria from system defaults and keeps the approved source material visible — it never replaces clinical judgment.

ProtocolsProceduresDiagnostic criteriaMeasurement techniqueReporting standardsAccreditation requirementsWorkflow policy

Synthetic examples · not real patient data

“What criteria are we using for a >70% internal carotid artery stenosis?”

This facility classifies a >70% ICA stenosis using a modified velocity criteria set: PSV greater than 230 cm/s, with ICA/CCA ratio and plaque estimate as supporting criteria.Source: Facility protocol — Vascular Lab Criteria, v3 (overrides system default)Retrieved verbatim from approved documentation — not generated.

“What's our protocol for a renal artery study in a patient with a prior stent?”

The in-stent renal artery protocol specifies the required sampling sites, the expected velocity ranges after stent placement, and the thresholds used to flag in-stent restenosis.Source: Facility protocol — Renal Artery Studies, Post-Intervention sectionRetrieved verbatim from approved documentation — not generated.

“What's our protocol for contrast-enhanced carotid ultrasound?”

No approved document at this facility covers contrast-enhanced carotid ultrasound. I can't answer this — there is nothing in the approved material to answer it from. Your medical director can approve a protocol, and it will be retrievable here once they do.Searched 2 approved documents — no matchNothing was generated to fill the gap.
Provenance

Where the knowledge comes from

The difference between this and a general-purpose chatbot is not the model. It's that every answer is traceable to a document your facility approved, and that the document is the thing you control.

Your documents

The corpus is material your facility has reviewed and approved. Nothing is pre-loaded from the open internet.

Versioned

“Vascular Lab Criteria, v3” is a real version, not a label. Approve v4 and every answer moves with it.

Yours overrides ours

Where a facility document and a system default disagree, the facility wins — and the answer says which one applied.

Documentation, not patients

The knowledge layer retrieves from approved documents. It is not connected to the patient record: no exam values, measurements, or images are part of what it reads.

“AI should make clinical knowledge easier to access. It should never quietly become the authority.”

Ask Lumen a real facility protocol question

Bring one of your own facility's approved protocols to a demo and see how it's retrieved.