Math Colloquium: Improving Healthcare with Queuing, Optimization, and Neural Networks

Wednesday, November 28, 2018

3:00 PM-4:00 PM

David Scheinker, Stanford University

Title: Improving healthcare with queuing, optimization, and neural networks (all the applied math I'd never heard of)

Abstract: Queuing theory, mathematical programming, machine learning, neural networks, and mechanism design are areas of active research and have a wide variety of practical applications. These tools are built on the foundation of mathematical analysis and probability theory, but are not often included in a standard mathematics curriculum. Despite their huge impact, it is easy to finished a PhD and a PostDoc in analysis knowing little about these methods (I did).

This talk gives a very brief mathematical overview of these methods; reviews some of the widely publicized recent successes of neural networks in healthcare; presents highlights from the speaker's work at a children's hospital; and states a few accessible open problems. The applications include machine learning to forecast surgical procedure duration; optimization to prevent delays in the operating room; deep neural networks to detect hypotension; and the design of 'fair' rules for allocating organs.

Contact Information

Georgi Medvedev
gsm29@drexel.edu