How algorithms developed during COVID-19 are expanding the digital strategy at UMass Memorial

Dr. David McManus (left), chair of the Department of Medicine and Dr. Michael Gustafson, president, UMass Memorial Health.

Photo courtesy of UMass Memorial Health

UMass Memorial Health researchers and clinicians have expanded predictive analytics developed during the COVID-19 pandemic to clinical decision support for use in radiology, gastroenterology, at-home acute care and chronic disease management.

Using in-house proprietary developments, vendor support and other partnerships, UMass developed an algorithm that is accessible in the EHR. Data from wearable devices, images from CT scans, radiology and other sources are linked in the patient record. 

During the initial surge of COVID-19 in Massachusetts, physicians could look at the records of COVID-19 patients to check their respiratory status and access a prediction algorithm. They could determine which patients could be safely moved to a nearby field hospital in a downtown Worcester auditorium. This was needed due to overcrowded conditions at the hospital due to COVID-19.

The field hospital is now closed, though the number of COVID-19 cases is again on the rise due to the Delta variant. As of Friday, the hospital had 10 patients admitted for COVID-19, according to Dr. David McManus, director of the Atrial Fibrillation Treatment Program and chair of the Department of Medicine.

But innovation initiated during the pandemic is only growing.

The development of ambient clinical intelligence is able to learn and feed insights in real time, McManus said. This "background ambient" is like having someone looking out for you, he said. It's someone sitting inside the data source to see patterns that indicate a patient could be at risk for a fall, dementia or depression. It has proved useful in looking at patients in the hospital who are likely to go to the ICU.

In gastroenterology, AI is being used to predict what is likely to be cancer from images taken by endoscopy. Patterns in cardiology and radiology can also predict coronary disease. 

The Radiology Department is experimenting with artificial intelligence and machine learning that can assist in the reading and detection of small lumps. The technology is at the very nascent stage, but AI and predictive analytics are looking to expand rapidly, McManus said.

UMass Memorial is exploring strategic partnership to see if it makes sense to develop these technologies together.

"What my goal is, is to continue to develop algorithms that get smarter and smarter," McManus said.

Prior to COVID-19, these types of predictive analytics would have taken years to develop. 

"COVID forced us to collaborate more." McManus said. "It was proof of concept that we could do this for other health conditions outside of COVID." 

TELEHEALTH

The digital strategy at UMass Memorial, as at most hospitals, began with telehealth, driven by necessity and changes to reimbursement and regulatory policy.

Pre-pandemic, telehealth visits were at 2% of all visits. COVID-19 accelerated that to three out of four visits, according to Dr. Michael Gustafson, president of UMass Memorial Medical Center. 

"We quickly rolled out a telehealth platform that was integrated into the record," Gustafson said. "We got physicians trained."

Telehealth has stabilized to about 20% of visits done by phone or video, except in the area of behavioral health, which has been converted to 100% of patient visits. The no-show rate, compliance and continuity are so much better than relying on people to come in physically, according to Gustafson. 

Massachusetts has extended state payment parity for behavioral health permanently, he said.

HOSPITAL AT HOME

The latest innovation about to be implemented at UMass Memorial is the Hospital at Home program. 

Hospital at Home is almost entirely driven by remote acute care. Using the model applied to determine field hospital patients, physicians are able to identify those who can safely go home for care. 

"This is not Apple Watch use. This is a person who would otherwise be in a hospital," McManus said. "We're getting to that level of remote disease monitoring. It's going to expand rapidly to chronic disease management. It's a reason for optimism in medicine."

An estimated 30, 40 or 50 patients will get care at home in their beds, which results in better outcomes. It's also where they'd rather be, Gustafson said.

"That's another significant long-term impact to the pandemic, thinking about how we can move as much care out of the hospital," Gustafson said. "Moving inpatient surgery to outpatient for example. We're going all out in our Hospital at Home program, in part because we think it's so much better care for patients." 

HEALTH EQUITY

"The pandemic really was this incredible one-year experiment," Gustafson said. 

Health systems learned that the virus impacted populations very differently, with a much higher incidence in minority populations, he said. Once members of these populations had the disease, it was much more severe and produced higher mortality rates and longer hospitalizations. 

"When we got the treatment, there was a disparity in how people were able to access the vaccine," Gustafson said.

They learned that the online registration system didn't work equally well for all populations. So they went out and distributed the vaccine at local sites such as housing centers, community health centers, churches and places of work, to meet people where they were. 

RECOMMENDATIONS

Gustafson recommends that digital innovation become a part of the strategy of the health system or hospital at the highest level.  

"Digital health needs to be embedded in," he said. "It's not a vertical column by itself. It's an enabler of quality and safety, growth strategy, and population health."

Twitter: @SusanJMorse
Email the writer: susan.morse@himssmedia.com