First of all, thanks to all participating in the break out session during the Researchers Meeting last week!
As discussed we would like to start having regular TCs on Patient Level Prediction so we can move this important topic forward and together do some very interesting research.
If you like to take part in the Patient Level Prediction Workgroup please let us know by replying to this item in the forum or by sending us an email. Based on the location of the contributors we will try to find a good timeslot for our TCs.
Some interesting topics we would like to address:
How to obtain a good evaluation framework?
Extension of the methods library with other algorithms.
Inclusion of temporal information in the models?
Can we better use patient similarity in model development.
and many more…
We will create a working group page once the list of contributors is final.
Thanks Peter for taking the lead on this. I’d be delighted to participate
in the workgroup, looking forward to seeing us take patient level
prediction all the way through from methods research to open source
development to clinical applications to improve patient care. I think if
we can all agree to a common evaluation framework and share code for
implementing feature construction and model fitting, then we can
collectively accelerate the science of prediction quite quickly. I’m very
excited to see this get going in earnest. Thanks. Patrick
Lisa Schilling, MD, MSPH
Professor of Medicine
Department of Medicine, Division of General Internal Medicine
University of Colorado, School of Medicine
Mailing Address:
Division of General Internal Medicine
University of Colorado School of Medicine
8th Floor, Academic Office 1-Office 8219
Mailstop B180
12631 E. 17th Ave
Aurora, CO 80045