The use of diagnostic, prognostic, or more generally, decision support tools in healthcare, built from machine learning algorithms and artificial intelligence (AI), requires special attention. My research focuses on the crucial aspect of the robustness of machine learning (ML) models in healthcare. The objective of my thesis is to clearly define the various concepts of robustness that can be used to evaluate the robustness of an ML model in healthcare and identify those that should be prioritized to limit potential degradation of a model once deployed.
Theses
Conceptualizing and assessing the robustness of healthcare algorithms
Alan Balendran
Promotion : 2022
Supervisor.s : Raphaël Porcher
