Objectives:

Project Leads: Martijn Schuemie, PhD,Marc Suchard, MD, PhD

The plan:

• Jan 2015: final XPRESS pipeline ready
• March 2015: 10 models built
• May 2015: Test portability of 10 models
• July 2015: Run the XPRESS vs. “queries” comparison at a non-stanford site.
• Oct 2015: Decide the approach to use in OHDSI

Levels of membership/commitment:

• Prioritizing the list of phenotypes to build models for
• Apply a trained model at your site and assess performance (i.e. examine portability)
• Run the XPRESS pipeline, and build models for a set of phenotypes.
• Get in the guts of the pipeline, mess with feature engineering, etc.
• Try different approaches (models for episodic events)

The commitments (all tentative, please think about finalizing):

George Hripsack (+ Barnett Chiu, Sunny Chang):

• Apply a trained model from other sites.
• Run XPRESS for new phenotypes.

Nicholas Tatonetti (+ Alexandre Yahi)

• Prioritizing phenotypes
• Applying a trained model from other sites

Jon Duke (+ need name)

• Apply a trained model from other sites
• Run XPRES to build new models

[thinking about availability]

• Tiffany Callahan (UC Denver, biostats)
• Daniella Meeker (Cheryl Chiu)

Start Date: January 2015.

Repository: TBD

WG Minutes: Need to find a good place

WG Agendas: Need to find a good place