BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:CBL Alumni Talk: Finale Doshi-Velez - Finale Doshi-Velez\, Harvard
  University
DTSTART:20210611T150000Z
DTEND:20210611T160000Z
UID:TALK158830@talks.cam.ac.uk
CONTACT:Elre Oldewage
DESCRIPTION:The volumes of medical data being recorded are now far beyond 
 what human experts can analyze\, especially at the bedside.  At the same t
 ime\, a lot of essential information goes unrecorded\; any purely machine 
 system will quickly hit fundamental statistical limitations.  Thus\, to ma
 ke a positive impact in healthcare\, we need robust tools that can communi
 cate their limitations and assumptions to the ultimate human decision-make
 r.\n\nWhether it is building models\, optimizing treatment policies\, or v
 alidating outputs\, my lab addresses this challenge by keeping human-machi
 ne validation and integration in mind from the start.  In this talk\, I wi
 ll discuss how our pursuit of small\, inspectable models lead us to find (
 and resolve) a decade-old failing of supervised generative models\; how th
 ose models (seemingly) resulted in better strategies for managing hypotens
 ion in the ICU\; how we realized that some of their strategies were (perha
 ps) bogus and how we regained (some of) our trust via novel ways of incorp
 orating human input into statistical off-policy evaluation\; and how our r
 ecommendations for interpretability and careful validation are being heard
  throughout the world.  All through the way\, I will emphasize the many in
 teresting technical questions whose answers have potential for real impact
 s in health.
LOCATION:https://eng-cam.zoom.us/j/89542469691?pwd=bVJjdXJGd3BkaDk1SHk3a1d
 LZzlZUT09
END:VEVENT
END:VCALENDAR