Back to docsRecipe

Patient intake summarizer

De-identified clinical intake pipeline. Transcribe a patient conversation, extract structured SOAP notes, and flag high-risk indicators — all without storing PHI.

⚠️

PHI-safe

Names, dates, and identifiers are stripped before the model ever sees the transcript.

📄

SOAP output

Subjective, Objective, Assessment, and Plan sections generated automatically.

🚨

Risk flags

Suicidal ideation, abuse indicators, and medication conflicts surfaced inline.

Quick start

  1. Upload an audio file or paste a raw transcript into the Meridian console.
  2. The pipeline runs de-identification, then routes the sanitized text to the clinical model.
  3. Review the structured SOAP note and any risk flags in the dashboard.

This recipe is designed for triage and documentation support. It is not a diagnostic device. Always validate output with a licensed clinician.