Predicting Surgery Duration with Neural Heteroscedastic Regression

Nathan Ng, Rodney A Gabriel, Julian McAuley, Charles Elkan, Zachary C Lipton

Feb 17, 2017 (modified: Mar 17, 2017) ICLR 2017 workshop submission readers: everyone
  • Abstract: Scheduling surgeries is a challenging task due to the fundamental uncertainty of the clinical environment, as well as the risks and costs associated with under- and over-booking. We investigate neural regression algorithms to estimate the parameters of surgery case durations, focusing on the issue of heteroscedasticity. We seek to simultaneously estimate the duration of each surgery, as well as a surgery-specific notion of our uncertainty about its duration. Estimating this uncertainty can lead to more nuanced and effective scheduling strategies, as we are able to schedule surgeries more efficiently while allowing an informed and case-specific margin of error. Using surgery records from the UC San Diego Health System, we demonstrate potential improvements on the order of 18% (in terms of minutes overbooked) compared to current scheduling techniques, as well as strong baselines that do not account for heteroscedasticity.
  • TL;DR: We accurately predict both the expectation and variability of surgery durations with heteroscedastic neural regression.
  • Conflicts:
  • Keywords: Deep learning, Supervised Learning