Causal Discovery from Biomedical Data – June 8-11, 2015, Carnegie Mellon University, Pittsburgh, PA
Location: Baker Hall, Giant Eagle Auditorium, A51, CMU

Course Director: Richard Scheines, PhD

Day 1 Morning Presentation

Day 1 Afternoon Presentation

Day 2 Morning Presentation

Day 2 Afternoon Presentation

June 8June 9June 10June 11

Causal Graphical Models – Richard Scheines with Joe Ramsey

  • Introduction
    • Overview of causal graphical models
    • Loading Tetrad
    • Causal graphs/interventions
  • Building Models
    • Parametric models: Bayes nets, Structural Equation Models (SEM), etc.
    • Instantiated models
  • Estimation
    • Estimation
    • Inference
    • Updating
  • Hands-on Real-Data Examples

Dinner on your own, though CCD personnel will be on hand to give advice and rides.

Search – Richard Scheines with Joe Ramsey

  • D-Separation
  • Model Equivalence
  • Search Basics
    • PC
    • GES
    • GES-IDA
    • GES Speed-up
  • Hands-on Real-Data Examples
  • Latent Variable Model Search
    • FCI
    • SEM search

Break early for Poster Session and Dinner (6-8 pm)

Biomedical Causal Discovery Overview

  • Overview of biomedical problems as case studies
  • Using Background Knowledge
  • Time Series
  • Handling Feedback
  • Unit of Analysis (single unit, neurons, etc.)

Dinner on your own, though CCD personnel will be on hand to give advice and rides.

Biomedical Causal Discovery Case Studies

Attendees choose the biomedical case studies of most interest to them

  • Morning case studies
  • Group Discussion
    • Overview by CCD Systems Architecture & Algorithm groups
    • Question & answer period
    • Interactive discussion of course material
    • Attendees discuss the types of datasets they need to model – help guide CCD algorithm development
    • CCD user outreach activities