You are here

Dynamic resource allocation and hospital Load balancing using optimization models and predictive analytics

COVID-19 Research Area(s): Mathematical Modelling and Operations

Imagine a patient presents to the ED of a hospital with COVID like symptoms. The current situation in N. America is such that in most hospitals one of the following actions (broadly) will be taken on this patient – a. asked to go home, take acetaminophen, drink lots of fluids and monitor for worsening symptoms, b.  admitted to a ward or c. admitted to an ICU.  Note that depending on the hospital, data suggests a significant number of patients in group (b) may need the resources in (c) at some later point. In this project, using presentation symptom data (along with CCD scores in an ED), demographic data of the patient and medical history, we dynamically (using updated information) predict the pathway of patients presenting and admitted to an ED. As a consequence, we also predict the Length of stay and resources consumed by a patient. The first and obvious use of this prediction is to align resources in advance in a hospital – it helps operational planning. It also is a key input to manpower modelling of critical care teams in a hospital. Next, Demand for COVID related hospitalization or testing services is highly heterogeneous (eg. high in zips w/ older populations, nursing facilities etc.). Under current patient hospital choice patterns we project this results in substantial overuse of smaller facilities that have limited flexibility and relative underuse of larger facilities with greater flexibility. Using the above ideas we optimize the load across hospitals by formulating a load balancing model. This last idea is currently used in MA, USA developed by faculty in Sloan. A crucial measure used in hospitals is patient to bed ratio. Currently, in overloaded systems such as the ones in Ontario and NYC, this number is much higher than 1. However, larger and more flexible hospitals can accommodate a higher ratio than a small rural hospital, thus the need for a load balancer.