A hospital capacity system powered by AI and developed by GE Healthcare is giving clinicians in Oregon near real-time information about ICU beds and other critical care resources in the state.

The system was forged during the Covid-19 pandemic.

“It gives us a bird’s-eye view of critical care,” said Helene Anderson, regional director of capacity and throughput at Providence Health & Services. “We know where we have capacity, and we can also coordinate with other services that wrap around the care.”

To show how it works, say there is an emergency in Oregon. A patient is fighting a life-threatening situation and the paramedics at the scene are worried. If the patient does not receive blood and intensive care soon, the outcome could be dire.

Fortunately, it’s not long before an air ambulance hovers overhead. A few minutes later, the patient is on board and arrowing towards the hospital, where an ICU and a team of medical personnel are ready. The air ambulance team has packed plentiful supplies of blood, allowing lifesaving treatment to begin in the cabin.

“I get the chills when I think about it,” said Anderson.

The story might have ended differently were it not for the artificial intelligence-enhanced software application developed by GE Healthcare that helps match critically ill Oregonians with lifesaving resources across their state.

The AI system, called the Oregon Capacity System (OCS), continually analyses near-real-time information about critical care capacity in the state. This helped the system identify the nearest ICU for the patient and alert the ambulance crew to pack several litres of the right blood type in the helicopter’s fridges.

The Oregon Health Science University (OHSU) in Portland led the development using a digital command centre, which GE Healthcare has implemented at hospitals globally to help optimise their operation.

Since March 2020, some 65 of the hospitals in Oregon – approximately 90% of the state’s beds – have used an early version of the app to increase the state’s major lifesaving resources, including ICUs, ventilators and negative pressure beds. That system corralled 4.2 million data points per day from the state’s hospital networks, updating once every five minutes with fresh intelligence about admissions, patient movements, nursing flows, ventilator usage and bed requests. When algorithms went to work on these data, Oregon’s clinicians were able to see key metrics and use them to make quicker, more informed operational decisions.

“It has performed brilliantly,” said Jeff Terry, who leads Clinical Command Centers, the GE Healthcare unit that developed the software. “It’s never gone down, the data have flowed, it’s been there when needed.”

It is possible the OCS may have helped boost Oregon’s capacity to deal with Covid-19. The Pacific Northwest state might have the lowest per capita supply of hospital beds of all US states, but as of October 2021 it also boasted some of the lowest case and death rates in the country.

But Oregon and GE aren’t finished. By the end of this year, through a partnership with Apprise Health Insights, the data and analytics subsidiary of the Oregon Association of Hospitals & Health Systems, an expanded version of the system will cover a wider range of information on emergency departments, ventilators and life support systems. The beefed-up OCS already has an updated classification system for the state’s ICUs, allowing hospitals to allocate critical care units by type and capability.

The statewide expansion, the first in the nation, is being funded in part by a federal grant provided by the Office of the Assistant Secretary for Preparedness & Response and the US Department of Health & Human Services to Oregon’s hospital association.

“Our system is designed to scale regionally, nationally and beyond if needed with the appropriate controls in place,” Terry said.

So far, the system has allowed Oregon’s hospitals to confine some painstaking analogue processes to the past. Before the pandemic, making decisions about critically ill Oregonians was a “telephone game,” said Terry. If a clinician wanted to know which of the state’s ICUs could support, for example, extracorporeal membrane oxygenation (ECMO) – a process where the blood of a patient whose lungs are not working is pumped through a machine that mimics the heart and the lungs – or find a free psychiatric bed within 150km, they rang round until they had an answer. Moreover, “the second you put that information into the system, it was outdated”, said Anderson.

Terry said the new automated system took just a couple of hours to install and did not collect any sensitive patient data.

“It is more reliable and secure than the analogue processes it replaces,” he said.

GE Healthcare engineers have spent the past 18 months fine-tuning the OCS with member organisations so it can match the right ICU bed to the right patient. Working alongside Oregon’s intensive care physicians, the engineers grouped Oregon’s ICUs into four buckets based on their capabilities.

“It’s not just about the number of ICU beds you have, because they’re not all the same,” said Terry. “It’s how many ICU beds you have that can take a vented patient, or a patient with a tracheostomy tube.”

Matching critically ill Oregonians with the right facility is crucial because of the state’s size, larger than the entire UK.

“There are vast expanses where it’s six hours by car to get from hospital A to hospital B,” said Terry.

The benefits of the OCS extend beyond Oregon. Providence Health has received patients from as far afield as Denver, Chicago and Calgary,

Anderson said: “We know, in the instance that we are asked, if we have capacity or not, and about our future capacity.”

Terry remembers how the app was used during the wildfires in the Pacific Northwest region last summer. As the fires came close, the OCS helped identify facilities with capacity to take ICU patients who were being evacuated from threatened hospitals.

Underpinning the statewide success is the collaboration and shared governance model run by Apprise Health Insights. Apprise has created an advisory group to help ensure the system is consistent across the state, gathers information from hospitals about ways to improve the system, and passes that wish list to developers.

“We have learned a lot from Covid-19 and the gaps in the healthcare system with inaccurate data,” said Andy Van Pelt, CEO of Apprise Health Insights. “We are proud to partner and bring this lifesaving asset to Oregon for an inside view of capacity across our healthcare continuum.”

The OCS could help lighten a huge administrative burden for Oregon. Already, clinicians don’t need to update spreadsheets two or three times per day, Terry said, saving the state’s facilities up to 43,000 hours of human work per year, which could translate into millions of dollars of savings annually. Those productivity savings will increase as OCS expands through 2022.

And the system could help improve hospital use rates where just a small increase can have a huge positive impact on the bottom line, said Terry.

“Moving that red line up from 90% utilisation to 95% takes a lot of cost out of the system,” he said.

Terry sees challenges ahead, not least a reluctance among major healthcare providers to reveal capacity information publicly. But he is optimistic that Oregon’s experience with sharing data, and a shared governance model such as the one Apprise operates will help guide the discussion on how such data sharing can be managed to accommodate all parties. In fact, Terry expects other US states could soon follow Oregon’s lead.