Of utmost importance to managing a hospital system is a thorough understanding of key volumes. As the analytical culture develops at the Cleveland Clinic, there is a shift from reactive decision making to proactive decision making. This calls for a slew of statistical forecast models, inserted into the workflow and at the fingertips of leaders and key decision makers. However, just putting these models out there isn’t enough. There’s a need active end-user engagement and buy in, in order for forecasting methods to be accepted and used to make decisions. This requires consistent interaction between our data scientists and end users throughout the modeling and validation process.
This 20-minute presentation covers Cleveland Clinic’s complete journey from model inception through to end-user acceptance and use for weekly business management. You’ll hear about the forecast model build using SAS®, application of Monte Carlo simulations and programming, output visualization, and the struggle to achieve buy-in from end users.
Video highlights
0:32 – Driving analytic maturity
01:55 – The beginning
03:00 – Feature engineering and model building
08:13 – Are “complete” data really complete?
09:00 – Dealing with extensive data lag
10:37 – Are we ready to model?
12:20 – How can we leverage ‘schedule’ variables?
13:50 – Forecast Studio modeling
16:07 – Monte Carlo simulation
18:09 – Output and user engagement
Related Resources
Cleveland Clinic’s SASGF presentation (proceedings)
How Cleveland Clinic uses medical resource optimization to fight pandemics (on-demand webinar)
Forecasting models with the Time Series Forecasting system (product overview)
Looking inside SAS Forecast Studio (white paper)
