Doctoral Thesis: Rigorously Tested & Reliable Machine Learning for Health
32-G575
Michael Oberst
Abstract:
How do we make machine learning as rigorously tested and reliable as any medication or diagnostic test? Machine learning (ML) has the potential to improve decision-making in healthcare, from predicting treatment effectiveness to diagnosing disease. However, standard retrospective evaluations can give a misleading sense for how well models will perform in practice. Evaluation of ML-derived treatment policies can be biased when using observational data, and predictive models that perform well in one hospital may perform poorly in another.
In this thesis, I introduce new tools to proactively assess and improve the reliability of machine learning in healthcare. A central theme will be the application of external knowledge, including review of patient records, incorporation of limited clinical trial data, and interpretable stress tests. Throughout, I will discuss how evaluation can directly inform model design.
Details
- Date: Thursday, May 25
- Time: 10:00 am - 11:30 am
- Category: Thesis Defense
- Location: 32-G575
Additional Location Details:
Thesis Committee: Professors David Sontag (supervisor), Jonas Peters, Tommi Jaakkola
(zoom link available upon request)