The program will be available shortly. Please check back later.
Presenter: Andreas Kleppe, Institute for Cancer Genetics and Informatics, Oslo University Hospital, and Department of Informatics, University of Oslo
Many deep learning studies are not designed to provide unbiased estimation of the system's performance in the intended application. Reports of overoptimistic estimates and opportunities may inflate the expectation of what is currently possible, misguide resource allocation, and hamper the progression of the field. In this talk, we will look into how the performance of a deep learning system in an intended application could be estimated more reliably than what is currently common practice, even if restricted to using retrospective data. To exemplify how some choices of the learning setup may influence the generalisability of the system, results will be presented from experiments where the goal is to predict whether a patient will eventually die from a cancer or survive following surgery. The presentation builds upon the publications:
The Lancet: Deep learning for prediction of colorectal cancer outcome: a discovery and validation study.
Nature: Designing deep learning studies in cancer diagnostics.
In compliance with GDPR consent requirements, presentations given in a Visual Intelligence context maybe recorded with the consent of the speaker. All recordings are edited to remove all faces, names and voices of other participants. Questions and comments by the audience will hence be removed and will not appear in the recording. With the freely given consent from the speaker, recorded presentation may be posted on the Visual Intelligence YouTube channel.