Pooya Mobadersany and Dr. Jeff Goldstein have published a paper in Lab Investigation describing a convolutional network that learns to aggregate fields across whole slide images for prediction of slide-level labels. This network, called GestAltNet, was used to predict gestational age from images of placenta tissues, and demonstrated a clear advantage over alternatives in predicting gestational age from whole-slide images. GestAltNet was able to reduce prediction error from 1.5 weeks to 1.1 weeks. This method can assist pathologists in identifying discordance between gestational age estimated from placenta histology and other methods which can indicate conditions like gestational diabetes or preeclampsia. Furthermore this approach can reduce annotation effort during model development and provide insights into patterns and regions that have diagnostic or prognostic value during inference.