Coursera Spring Sale
40% Off Coursera Plus Annual!
Grab it
Attend this 49-minute conference talk examining fairness in foundation models for histopathology applications. Explore how covariate bias affects the performance and reliability of foundation models when applied to medical imaging, specifically focusing on histopathological analysis. Learn about a novel research approach that uses identical glass slides digitized with different scanners to create a controlled environment for studying bias in data distributions. Discover the concept of "representation shift" and understand how scanner-dependent variability in feature representations can impact the generalization capabilities of foundation models. Examine quantitative metrics including mean squared error, Kullback-Leibler divergence, and a clustering-based Calinski-Harabasz index used to assess bias effects. Gain insights into the critical implications of device-dependent performance variations for the safe and equitable deployment of AI systems in clinical settings. The presentation is delivered by Abubakr Shafique, a machine learning scientist with expertise in medical image analysis and a Ph.D. from the University of Waterloo, as part of Stanford University's MedAI Group Exchange Sessions that foster critical examination of AI applications in medicine.
