Dear workshop community,
We will convene for the Harvard University Applied Statistics Workshop (Gov
3009) TOMORROW on Wednesday (5/1). *Note: This will be the last workshop
meeting of the 2018/2019 academic year. *
The speaker is* Michael Hughes *(Tufts Engineering) who will be presenting
his latest work, "Discovering Disease Subtypes that Improve Treatment
Predictions: Prediction-Constrained Topic Models for Personalized
Medicine".
*Where:* CGIS Knafel Building, Room K354 (see this link
<https://map.harvard.edu/?bld=04471&level=9> for directions).
*When: *Wednesday, May 1st at 12 noon - 1:30 pm.
*Abstract: *For complex diseases like depression, choosing a successful
treatment from several possible drugs remains a trial-and-error process in
current clinical practice. By applying statistical machine learning to the
electronic health records of thousands of patients, can we discover
subtypes of disease which both improve population-wide understanding and
improve patient-specific drug recommendations? One popular approach is to
represent noisy, high-dimensional health records as mixtures of
low-dimensional subtypes via a probabilistic topic model. I will introduce
this common dimensionality reduction method and explain how off-the-shelf
topic models are misspecified for downstream prediction tasks across many
domains from text analysis to healthcare. To overcome these poor
predictions, I will introduce a new framework -- prediction-constrained
training -- which learns interpretable topic models that offer competitive
drug recommendations. I will also discuss open challenges in using machine
learning to improve clinical decision-making.
*All are welcome! Lunch is provided! *
Best,
Connor Jerzak
Applied Statistics Workshop -- Graduate Student Coordinator
An anonymous feedback form for the workshop can be found here at this link
<https://docs.google.com/forms/d/e/1FAIpQLScp4lPVBtp4Akf6K6ggmfcTUSIUHEJX89-CU8HWrQPpe9pjTw/viewform?usp=sf_link>.
Workshop listserv sign-up at this link
<https://lists.fas.harvard.edu/mailman/listinfo/gov3009-l>.
Show replies by date