Achieving 91% forecasting accuracy

Published by Hiro Kawashima on December 13, 2018

You've prepared yourself for a brisk but sunny winter day only to be blindsided by twelve inches of snow. Or you brought your swimwear to the beach expecting 80 degree weather only to encounter a sudden drop in temperatures. How accurate are weather forecasts? A typical 10-day weather forecast is only accurate about 50% of the time.

The forecasts you rely on to run your hospital or department can't be as unreliable as weather forecasts. The decisions you make effect patient outcomes, staff experience, and hospital operations. We developed the Agilecast forecasting engine to ensure that you have the most accurate and reliable patient volume and staffing need forecasts. 

We recently reviewed 18 months of patient volume and staffing data from 60 units at a large hospital system on the East Coast. The data set we examined including a variety of departments including emergency departments (general and pediatrics), intensive care units, and med surge units. In addition, the data set also included all clinical and administrative position types. We leveraged our Agilecast forecasting engine on the data set to forecast a theoretical month and compared our results with actual patient volumes and staffing demand. 

Here are some highlights from the data analysis:

  • Amongst all departments and position types, we achieved a 96% forecasting accuracy for the next shift and a 91% forecasting accuracy 30-days out. 
  • For most positions, we achieved a 100% forecasting accuracy both for the next shift and 30-days outs.
Next post: Improve unit coverage and decrease costs. Sounds too good to be true?

Take the guesswork out of staffing

Schedule a demo and learn more about how predictive analytics can help you staff better