Technical White Paper for Clinical Methodologies

Generates a deep technical white paper or educational document for clinical methodologies, focusing on scientific validity and regulatory context.

View Source YAML

---
name: Technical White Paper for Clinical Methodologies
version: 0.1.0
description: Generates a deep technical white paper or educational document for clinical methodologies, focusing on scientific validity and regulatory context.
metadata:
  domain: technical
  complexity: high
  tags:
    - technical-writing
    - white-paper
    - clinical-trials
    - data-science
    - regulatory
  requires_context: true
variables:
  - name: paper_title
    description: The title of the white paper.
    required: true
    default: "De-Risking Clinical Development: The Methodology of In Silico Trials"
  - name: context_description
    description: Context about the purpose and audience of the paper.
    required: true
    default: |
      We need to prove to our clients that our "In Silico" simulations are scientifically valid and not just a "black box." I am providing you with the core pillars of our offering.
  - name: source_material
    description: The core concepts or pillars to cover in the paper.
    required: true
    default: |
      * **Concept 1 (SCA):** Constructing "synthetic" arms using historical clinical trial data and RWE.
      * **Concept 2 (Simulation):** Agent-based modeling to simulate protocol feasibility (predicting dropouts, site burden).
      * **Concept 3 (Digital Twins):** Virtual physiological models for biomarker efficacy prediction.
  - name: specific_requirements
    description: Specific requirements for the body of the white paper.
    required: true
    default: |
      2. **Methodology Section - Synthetic Control Arms:**
      * Explain the statistical techniques used to match real-world data to trial patients (e.g., Propensity Score Matching, exact matching).
      * Discuss how we handle data heterogeneity and bias.

      3. **Methodology Section - Agent-Based Modeling:**
      * Describe how "Agents" (patients/sites) are programmed. What variables define an agent? (e.g., compliance probability, distance to clinic, symptom severity).
      * Explain how the simulation runs "Monte Carlo" style iterations to predict failure points.

      4. **Methodology Section - Mechanistic Digital Twins:**
      * Explain the difference between statistical modeling and physiological/mechanistic modeling.
      * Case Study Example: Create a hypothetical scenario where a Digital Twin identifies that a drug only works on patients with a specific renal profile.
model: gpt-4
modelParameters:
  temperature: 0.4
messages:
  - role: system
    content: |
      You are a dual expert in Clinical Data Science and Regulatory Affairs. Your tone should be academic, authoritative, detailed, and technically precise. Use LaTeX formatting for any necessary mathematical concepts.
  - role: user
    content: |
      **Context:**
      {{context_description}}

      **Source Material:**
      {{source_material}}

      **Task:**
      Write a **Technical White Paper** titled *"{{paper_title}}"*. The paper must cover:

      1. **Introduction:** Define the core subject and the current regulatory appetite for them (reference FDA guidance or EMA opinion papers if applicable).

      {{specific_requirements}}

      **Conclusion:** A summary of how these technologies converge to ensure protocol integrity.
testData:
  - input: |
      paper_title: "De-Risking Clinical Development: The Methodology of In Silico Trials"
      context_description: |
        We need to prove to our clients that our "In Silico" simulations are scientifically valid and not just a "black box." I am providing you with the core pillars of our offering.
      source_material: |
        * **Concept 1 (SCA):** Constructing "synthetic" arms using historical clinical trial data and RWE.
        * **Concept 2 (Simulation):** Agent-based modeling to simulate protocol feasibility (predicting dropouts, site burden).
        * **Concept 3 (Digital Twins):** Virtual physiological models for biomarker efficacy prediction.
      specific_requirements: |
        2. **Methodology Section - Synthetic Control Arms:**
        * Explain the statistical techniques used to match real-world data to trial patients (e.g., Propensity Score Matching, exact matching).
        * Discuss how we handle data heterogeneity and bias.

        3. **Methodology Section - Agent-Based Modeling:**
        * Describe how "Agents" (patients/sites) are programmed. What variables define an agent? (e.g., compliance probability, distance to clinic, symptom severity).
        * Explain how the simulation runs "Monte Carlo" style iterations to predict failure points.

        4. **Methodology Section - Mechanistic Digital Twins:**
        * Explain the difference between statistical modeling and physiological/mechanistic modeling.
        * Case Study Example: Create a hypothetical scenario where a Digital Twin identifies that a drug only works on patients with a specific renal profile.
    expected: |
      Introduction
      Methodology
      Conclusion
evaluators:
  - name: Output contains Introduction header
    regex:
      pattern: (?i)Introduction
  - name: Output contains Methodology or Concept headers
    regex:
      pattern: (?i)(Methodology|Synthetic|Simulation|Digital\s+Twin)