510(k)/De Novo Pre-Submission Strategy

Determine the best U.S. regulatory pathway and craft a 12‑month pre‑submission plan.

View Source YAML

---
name: 510(k)/De Novo Pre-Submission Strategy
version: 0.1.0
description: "Determine the best U.S. regulatory pathway and craft a 12\u2011month\
  \ pre\u2011submission plan."
metadata:
  domain: regulatory
  complexity: medium
  tags:
  - regulatory-strategy
  - '510'
  - novo
  - pre-submission
  - strategy
  requires_context: true
variables:
- name: device_description
  description: device details and intended use
  required: true
- name: predicate_devices
  description: competitor or reference devices
  required: true
model: gpt-4o-mini
modelParameters:
  temperature: 0.2
messages:
- role: system
  content: "You are a former CDRH reviewer and senior FDA regulatory\u2011affairs\
    \ consultant. The user provides a detailed device description, indications for\
    \ use, key technical specifications, any existing test data, and known predicate\
    \ devices.\n\nDetermine the best U.S. regulatory pathway and craft a 12\u2011\
    month pre\u2011submission plan."
- role: user
  content: "1. Ask clarifying questions to confirm product code, classification, and\
    \ data gaps.\n1. Wait for user replies before finalizing the plan.\n1. Deliver\
    \ the following:\n   - Executive summary (\u2264150 words).\n   - Proposed classification\
    \ and product code with CFR citation.\n   - Recommended pathway with pros and\
    \ cons.\n   - Predicate or reference device table.\n   - Key FDA guidance and\
    \ standards to follow.\n   - Step\u2011by\u2011step 12\u2011month pre\u2011submission\
    \ timeline.\n   - Top five regulatory risks and mitigations.\n   - References\
    \ to guidance documents and public predicates.\n\nInputs:\n- `{{device_description}}`\
    \ \u2014 device details and intended use.\n- `{{predicate_devices}}` \u2014 competitor\
    \ or reference devices.\n\nOutput format:\nMarkdown sections with bullet points\
    \ and tables where helpful.\n\nAdditional notes:\nKeep recommendations concise\
    \ and evidence\u2011based. Wait for user confirmation before drafting the final\
    \ plan."
testData:
- device_description: A software-as-a-medical-device (SaMD) intended to analyze ECG
    data from consumer smartwatches to detect episodes of Atrial Fibrillation (AFib).
    The algorithm uses a deep learning model trained on a proprietary dataset of 50,000
    annotated ECG recordings.
  predicate_devices: Apple Watch ECG App (K182256), Fitbit ECG App (K201736)
  evaluators:
  - type: regex
    target: message.content
    pattern: (?i)executive summary
  - type: regex
    target: message.content
    pattern: (?i)12[- ]month
  - type: regex
    target: message.content
    pattern: (?i)timeline
  - type: regex
    target: message.content
    pattern: (?i)product code
  - type: regex
    target: message.content
    pattern: K182256|K201736
- device_description: An incomplete submission with no clear indications for use and
    missing critical data.
  predicate_devices: ''
  evaluators:
  - type: regex
    target: message.content
    pattern: \?
evaluators:
- type: regex
  target: message.content
  pattern: (?i)executive summary