Governed AI · FHIR R4 · Full Provenance

From raw lab results
to a patient message.

Clinicians lose hours a day to lab results: reviewing them, spotting what matters, and explaining them to patients.

Validor ingests raw analyzer output, validates and normalizes it into FHIR R4, then runs a governed AI layer to surface priority findings, plot trends, and draft the patient message.

The AI augments the data, it never gates or alters it. Nothing reaches a patient without a clinician's approval.

Priority findings

Not every result needs a flag. These do.

History under every flag

Plotted trend for every abnormal value.

Patient-ready drafts

Results explained in plain language. Approved by clinician before sending.

See Validor in action
A single run through the full pipeline, replayed end to end: ingestion, validation, FHIR R4 persistence, then AI annotation and a patient message draft. Every stage emits timestamped audit trail events.
Clinical scenario

A patient on long-term statin therapy had three lab visits (BMP, LFT, lipid panel) in the past year. The patient's most recent lab report was just released.

Clicking 'Run Demo' simulates the service receiving new lab results from a lab analyzer and runs the full pipeline with live API calls.

From raw lab data to AI enrichment
Every file travels one deterministic path, then it's handed to the governed AI layer for triage and message drafting. No partial writes, no message sent without a clinician's approval.
Validor data flowA deterministic pipeline parses, validates, normalizes and persists every lab result. De-identified data crosses into a governed AI layer, where retrieval of clinical guidelines and patient history feeds AI annotation and a patient message draft that needs clinician approval. An append-only, timestamped audit log records events from every stage of both lanes.DETERMINISTIC PIPELINE: FROM RAW CSVParseCSV rowsValidateRule checksNormalizeFHIR mapPersistReports, obsAUDIT LOGappend-onlytimestampedRecords eventsfor every stageof the pipeline,end-to-endde-identified dataGOVERNED AI LAYERRetrievalRAG guidelines+ history queryAI annotationPriority + trendMessage draftNeeds approvalRAG guidelinesHistory querySchema-verifiedClinician sign-off