The encyclopedia · R&D & Science · Operational decision · 2011–2014
Grady remodelled its emergency department flow instead of building more beds
Georgia Tech and Grady coupled machine learning, simulation and optimization to cut ED length of stay about 33%.
Grady Health System (Grady Memorial Hospital)
the move
Emergency departments clog because patients arrive unpredictably and some stay longer than necessary, waiting on decisions. Adding beds is expensive and often doesn't relieve the congestion that's really about flow.
Georgia Tech's Eva Lee and the Grady team built an ED decision support system coupling machine learning, simulation and optimization. It predicts which patients will need admission and optimizes how they flow through the department, routing appropriate short-stay cases to a clinical decision unit.
The system lets administrators globally re-optimize workflow rather than react to the latest surge, turning the existing space and staff into a more efficient whole.
why it works
- It optimized the whole department's workflow, not a single queue, so the pieces moved together.
- It repurposed existing resources into a clinical decision unit rather than building new capacity.
- Machine learning predicted demand, and the model turned that into routing and staffing choices.
- It cut length of stay, which decongests the ED without more beds or staff.
what transfers
Before spending on capacity, model the flow and reallocate what you have — often the bottleneck is where work is routed, not how many beds or staff exist.
what came after
The system helped reduce length of stay at Grady by roughly 33%, and the repurposed clinical decision unit cut ED readmissions by about 28%. The work was a finalist for the 2014 INFORMS Franz Edelman Award and also earned a special second-place recognition at the 2013 Daniel H. Wagner Prize, with the paper published in Interfaces.
references
- Lee's Project Named One of Six Finalists for the INFORMS Franz Edelman Prize
- Transforming Hospital Emergency Department Workflow and Patient Care
spotted an error? The archive wants to know.