MIA: Nathaniel Diamant, No such thing as unlabeled: Self supervised learning on medical data (2021)

Models, Inference and Algorithms Broad Institute of MIT and Harvard October 13, 2021 Nathaniel Diamant Broad Institute No such thing as unlabeled: Selfsupervised learning on medical data In medical datasets, the most important labels are often the rarest. For example, while responsible for more than 450, 000 deaths a year in the United States alone, sudden cardiac death (SCD) will likely only show up in a few hundred health records in a hospital dataset of a hundred thousand patients. Furthermore, a binary label, like SCD, carries little information about the intricacies of the outcome. In contrast to the rarity and opacity of the labels, the relationships of data within a medical dataset are often plentiful and rich. Selfsupervised learning (SSL) is an approach to training deep learning models that ideally matches the characteristics of medical datasets. We propose Patient Contrastive Learning, an SSL approach which exploits a fundamental relationship in medical datasets