Seminer: “Physiology-Based Anticipative ICU Management,” Mehmet Yasin Ulukuş (University of Pittsburg), EA-409, 13:40 11 Ocak (EN)

Title: Physiology-Based Anticipative ICU Management by Mehmet Yasin Ulukuş , University of Pittsburg

January 11, 13:40


Intensive care units (ICUs) are specialized hospital departments that provide temporary support to critically ill patients. ICU costs have been increasing significantly, and represent around 20% of all hospital operating costs. ICUs are tightly connected to other hospital units such as operating rooms and emergency departments. Improved ICU workflow management improves patient flow in the entire hospital, reduce mortalities and readmissions, and help reduce healthcare expenses. Better predictions of performance measures are essential for optimal management of units. In that regard, we construct a new Transfer Readiness Score to estimate readmission and death probabilities upon transfer to a lower level of care unit. On our dataset, our score provides better estimations than all other models in the literature.
We further consider the transfer operations of patients from an ICU to a downstream unit. We investigate anticipative bed requests that can be made before a patient is ready for transfer. Patient health is described via proposed transfer readiness score that we incorporate into an infinite horizon Markov decision process model, which we solve through an approximation algorithm. We establish the existence of a threshold-type request policy for a single patient version of the problem. Our numerical results indicate that an anticipative transfer request policy can significantly improve the system performance, e.g., a 10-bed unit under our proposed policy performs as well as a 12-bed unit under current practice. We further investigate the sensitivity of policy change upon cost parameter estimation errors by using robust models, and demonstrate that proactive strategies are more beneficial than reactive current policy in most of the scenarios.

Mehmet Yasin Ulukus is a PhD candidate in the Department of Industrial Engineering at the University of Pittsburgh. He earned his MS and BS in Industrial Engineering from Bogazici University. His research interests span decision-making problems under uncertainty with focus on predictive health and healthcare operations. His methodological interests include stochastic processes, Markov decision processes (MDPs), and queueing theory. His current research focuses on data driven management of intensive care units.