Loading Events

« All Events

  • This event has passed.

Exposure-response modeling for binary and time-to-event data using R and Stan

November 3 @ 8:00 am - November 4 @ 4:00 pm

Event Navigation

Cost: Industry: $1,200 / Academia+Government: $800 / Student: $400

Join us for a 2-day workshop following ACoP13!

Instructors

  • Jim Rogers, PhD
  • Andrew Tredennick, PhD
  • Matthew Weins, MA
  • Ramon Garcia, PhD

Learning Objectives

  • Describe confounded exposure-response
  • Analyze binary data (exploratory analysis and modeling using R and Stan)
  • Analyze time-to-event data (exploratory analysis, semi-parametric and parametric modeling using R and Stan)
  • Interpret TTE models with continuously time-varying hazard

Topic Outline

Day 1

  • Study designs and confounded exposure-response
  • General theory / background
  • Binary data
  • Models for binary data
  • Bayesian models for binary data
    – Lunch Break –
  • Time-to-event (TTE) data
    • What makes it different?
      • Describing a distribution w/o censoring (density, CDF)
      • Describing a distribution w/o censoring (hazard, CDF, survival function)
      • Non-parametric estimation of S(t), H(t) and h(t)
      • Study design implications: number of events vs number of subjects
    • Visualizing TTE data vs predictors: K-M plot (session 1)
      • How to interpret it in general?
      • Considerations for exposure metrics in TTE analyses
      • Utility and pitfalls of exposure quartiles
      • Hands-on example: visualizing TTE endpoint vs treatment or exposure
  • TTE semi-parametric modeling
  • Introduction to parametric survival analysis

Day 2

  • TTE parametric modeling (Bayesian)
  • Additional topics in TTE modeling

To register, visit the ACoP meeting website.


REGISTER