Hot off the press! SIEDS 2021 - edas.info/p28115#top work by our @uvadatascience MSDS students Marissa, Tyler, Surbhi, Clair & @UVA Radiology PGY3 Dio + @yashsharma0906 @sauravsen111 — using #MachineLearning to classify #crohnsdisease MREs 1/ 
Our challenge was working with a sparsely labeled dataset with no annotations - we build 3 separate models & showed that a patient level model using multi-frame input (like humans do when we scroll through an MRE) had the highest accuracy! 2/
@JKurowskiMD @satishev @therealjonadill - would love your thoughts! @IBD_FloMD next steps for us will be to see if we can predict subsequent B2 (stricturing) / B3 (penetrating) disease using baseline MRs from patients who were B1 inflammatory at diagnosis!
• • •
Missing some Tweet in this thread? You can try to
force a refresh
