Skip to content

Instantly share code, notes, and snippets.

@erikr
Created February 23, 2026 14:18
Show Gist options
  • Select an option

  • Save erikr/08958d4e6af070829ac8c620328ad472 to your computer and use it in GitHub Desktop.

Select an option

Save erikr/08958d4e6af070829ac8c620328ad472 to your computer and use it in GitHub Desktop.

Write a rigorous Scoping Review on the application of Foundation Models (FMs) in chest medicine based on the PCC framework below.

Scope:

  • Models: Large Language Models (LLMs) and Vision Transformers (ViTs/multimodal). Exclude legacy CNNs/Deep Learning unless used as baselines.
  • Data: EHR (unstructured text) and Chest CT imaging.

PCC Framework (Inclusion Criteria):

  1. Population:
  • Interstitial Lung Disease (ILD): Fibrosis quantification and mortality risk.
  • COPD and Asthma: Phenotyping and exacerbation prediction.
  • Immunosuppressed Patients: Distinguishing Pneumonitis vs. Infection (e.g., HSCT/chemo).
  • Oncology Patients: Drug-Induced Pneumotoxicity (DI-ILD) from immunotherapy (ICI) or ADCs (e.g., Trastuzumab deruxtecan/Enhertu).
  1. Concept: Application of FMs for prognostication, outcome prediction, deep phenotyping, and risk stratification.
  2. Context: Clinical analysis of EHR text and CT imaging.

Required Structure:

  1. Title: Formal academic title.
  2. Abstract: Structured summary (Background, Objectives, Results, Conclusion).
  3. Introduction: Define FMs vs. traditional AI. Explain the rationale for mapping FM capabilities to complex respiratory tasks.
  4. Methodology: Simulate a search strategy focusing on model architecture and specific clinical concepts.
  5. Thematic Synthesis (Results):
  • Theme 1: NLP/LLMs in Obstructive Disease (Phenotyping/risk from notes).
  • Theme 2: Vision Transformers in ILD (Severity/progression prediction).
  • Theme 3: Complex Diagnostics (Pneumonitis vs. Infection in immunosuppressed hosts).
  • Theme 4: Toxicology/Oncology (AI surveillance of Enhertu/ICI toxicity).
  1. Discussion: Summarize evidence, highlight gaps (validation/multimodal integration), and address technical limitations (interpretability, hallucinations).
  2. Conclusion: Assessment of clinical readiness.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment