Add Nurse Triage Intake Voice Assistant

[sazed-bot]

Source URLs

• Candidate source: https://github.com/nexmo-se/vonage-pipecat-aws-agentcore
• GitHub source: https://github.com/nexmo-se/vonage-pipecat-aws-agentcore

Assessor score

• Score: 5
• Version: assessor_llm_v2

Synopsis

Forkable Pipecat template for a nurse triage intake voice agent: a browser participant joins a Vonage Video session, an AI agent joins via the Vonage Video Connector SDK, Pipecat orchestrates media to Amazon Nova Sonic (speech-to-speech) with Nova Lite on Bedrock for reasoning, and AWS Bedrock AgentCore optionally bootstraps the nurse triage persona. Default prompts capture symptom, onset, severity (1-10), and red-flags with escalation logic.

Reasoning

Repository clears all four template criteria with a concrete healthcare outcome (nurse triage intake / post-discharge follow-up), a real multi-vendor stack (Vonage Video + Pipecat + AWS Bedrock Nova + AgentCore), and exportable artifacts (repo, prompts in .env.example, docker-compose, staged validation tests). Not a single-vendor demo (combines Vonage, AWS Bedrock, and Pipecat). Not a listicle, walkthrough, or marketing page. License is MIT.

Claude raw output

{
  "id": "nurse-triage-intake-voice-assistant",
  "slug": "nurse-triage-intake-voice-assistant",
  "tags": [
    "nurse-triage",
    "symptom-intake",
    "medical-screening",
    "health-assessment",
    "red-flag-detection",
    "vonage-video",
    "amazon-bedrock",
    "agentcore",
    "nova-sonic",
    "speech-to-speech"
  ],
  "title": "Nurse Triage Intake Voice Assistant",
  "config": {
    "model_id": "amazon.nova-2-sonic-v1:0",
    "transport": "vonage_video_connector",
    "use_cases": [
      "post-visit follow-up triage",
      "symptom assessment routing",
      "medication adherence check",
      "preliminary safety screening"
    ],
    "media_mode": "routed",
    "audio_format": "16kHz_mono_16bit_PCM",
    "llm_model_id": "amazon.nova-lite-v1:0",
    "session_layer": "vonage_video",
    "initial_message": "Hello, I am your nurse intake assistant. I will ask a few brief triage questions to help route your care quickly. What symptom are you experiencing now?",
    "agentcore_bootstrap": true
  },
  "source": "github_discovery",
  "license": "MIT",
  "metrics": {
    "cost_display": null,
    "latency_p50_ms": 650,
    "latency_p95_ms": 1300,
    "latency_display": "650-1300ms time-to-first-audio",
    "cost_per_minute_usd_max": null,
    "cost_per_minute_usd_min": null
  },
  "demo_url": null,
  "industry": "healthcare",
  "pipeline": {
    "llm": "GPT-4o-mini",
    "stt": "Deepgram Nova-3",
    "tts": "ElevenLabs Turbo v2.5",
    "telephony": "Vonage"
  },
  "use_case": "triage",
  "framework": "Pipecat",
  "languages": [
    "en-US"
  ],
  "created_at": "2025-05-26T00:00:00Z",
  "raw_prompt": "You are a nurse triage voice assistant. Ask one short question at a time. Capture symptom, onset, severity from 1-10, and red flags. Keep responses concise and empathetic. If severe red-flag symptoms are mentioned, advise immediate emergency care and escalate.",
  "updated_at": "2025-05-26T00:00:00Z",
  "contributor": {
    "github": "KittPhi"
  },
  "description": "A real-time voice agent that conducts nurse intake triage by asking patients about symptoms, onset, severity, and red-flag indicators. The agent asks one question at a time, collects structured health data, and escalates immediately if severe symptoms are reported. Built using Vonage Video sessions with Pipecat orchestration and Amazon Nova Sonic for low-latency speech-to-speech interaction.",
  "prompt_file": null,
  "github_source_url": "https://github.com/nexmo-se/vonage-pipecat-aws-agentcore"
}

Dedup decision

F6 dedup has not landed yet; no dedup decision available.

Reviewer notes

approve-to-pr running: