As the director of the Office of Preparedness and Response for OSF HealthCare, Troy Erbentraut, needs three things to ensure OSF HealthCare can adequately respond to a disaster: space, staff and stuff.
“I need to create plans that tell our facilities where patients can be placed, how many clinicians are needed to treat those patients and what resources – such as medical equipment – we require to best care for individuals,” said Erbentraut.
For a slow-developing public health event such as the novel coronavirus (COVID-19), Erbentraut says it’s beneficial to use predictive analytics models to understand how the health care system will be impacted. However, state and national models being used across the country did not do a good job of making projections for the individual areas OSF HealthCare serves.
As a result, Advanced Analytics developed the OSF Pandemic Impact Model to predict how many COVID-19 patients the health care system could see. This gives leaders and those in charge of disaster planning a better idea of what resources the organization will need to keep up with demand, and better serve patients accordingly.
OSF Pandemic Impact Models
The OSF Pandemic Impact Model is made up of different machine intelligence approaches that continually learn and update, leading to more accurate predictions.
“It uses daily data updates from the Illinois Department of Public Health and makes projections at the county level,” said Chris Franciskovich, director of Advanced Analytics. “It determines how many COVID-19 cases could be considered mild, severe or critical, and predicts how many established OSF HealthCare patients could get the virus.”
Advanced Analytics also built another version of the Pandemic Impact Model that produces all of the same outputs, but comes to those conclusions differently. It is built on observed growth rates inside of Illinois and each OSF HealthCare region.
“I send out an update that compares the projections we made with the actual number of COVID cases we saw on a given day based on the scenarios and different versions we are running,” said Franciskovich. “The observed values are typically within the range of our projections each day.”
Franciskovich says the model is constantly evolving as situations arise. For example, it had to be adjusted to account for increases in COVID-19 testing results. For Erbentraut, the data that comes out of the OSF Pandemic Impact model is invaluable.
“In the early weeks of the pandemic, we looked at maximizing all of our operating units based on state and national models for Illinois; we thought we were going to fill every bucket to capacity,” said Erbentraut. “But because of Advanced Analytics work, we’ve been able to take a breath and approach changes to our plans, policies and procedures more systematically and with a little more precision.”
Better planning leads to better care
With the OSF Pandemic Impact Model, disaster preparedness teams and leaders can not only get an idea of the virus’s impact across the organization, but they can also see how each region will be affected. From a staffing perspective, it means ensuring there are enough care teams available to care for each patient at each facility.
“We’ve demonstrated that with OSF HealthCare Little Company of Mary Medical Center, where they are experiencing many cases,” said Kim Blakey, director of Clinical Business Operations, Nursing Administration. “We were able to shift more than 100 nurses, in addition to other members of the care team, to the Evergreen Park hospital because we knew we wouldn’t need those resources to deal with high demand in other parts of the organization.”
Overall, the model has helped OSF HealthCare make better decisions across the health system. This means running normal operations for most facilities while planning to increase personnel, medical equipment and space for areas that need it most.