BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:All models are wrong and yours are useless: making clinical predic
 tion models impactful for patients - Florian Markowetz\, Cancer Research U
 K Cambridge Research Institute
DTSTART:20251007T181500Z
DTEND:20251007T203000Z
UID:TALK222574@talks.cam.ac.uk
CONTACT:Peter Watson
DESCRIPTION:Most published clinical prediction models are never used in cl
 inical practice which leads to a huge gap between academic research and cl
 inical implementation. Here\, I propose a checklist to enable academic res
 earchers to be proactive partners in improving clinical practice and to de
 sign models in ways that ultimately benefit patients. Over the years I hav
 e come to see academic papers not as ends in themselves\, but as the begin
 ning of the journey to clinical implementation\, and I am frustrated with 
 how little of my own work ever had clinical impact. I argue that you shoul
 d outline the road to implementation whilst designing your prediction tool
  by thinking about what medical decisions they are making and how these to
 ols can be used in routine practice. I will illustrate these ideas by disc
 ussing AI models from our work to find a minimally invasive alternative to
  endoscopy and looking at ways of assessing breast cancer survival rates a
 fter surgery that can be used in the clinic.
LOCATION:MRC Cognition and Brain Sciences Unit\, Chaucer Road\, Cambridge
END:VEVENT
END:VCALENDAR