Boosted Tree Ensembles for Predicting Postsurgical ICU Mortality

October 7, 2013


October 7, 2013


12:00 AM


Minneapolis, MN


INFORMS Annual Meeting

Real-time monitoring of patient conditions in the ICU environment is essential in supporting clinical decisions and ensuring optimal allocation of medical resources. This study focuses on accurately predicting in-hospital ICU patient mortality utilizing functional profile data. We demonstrate that a boosted trees ensemble model is well suited for the diverse data typologies present in ICU data and provides an interpretable and accurate model to aid clinical experts in critical ICU decisions.

Posted on:
October 7, 2013
1 minute read, 71 words
Conference Presentation
Research INFORMS
See Also:
The Passive Smart Home: Unobtrusive At-Home Behavior Monitoring
Unsupervised behavior change detection using passive sensor systems in the homes of older adults
Presenting Smart Home Activity Profiles at INFORMS 2017