How one health system harnessed AI, telehealth to combat coronavirus
When the United States reported its first COVID-19 case in January 2020, the Medical University of South Carolina activated its telehealth response – months before swathes of other health systems rapidly pivoted to virtual care.
By April, that response included virtual urgent care, remote patient-monitoring, continuous virtual monitoring for in-unit patients and a shift in ambulatory care services.
According to a new manuscript published in the Journal of the American Medical Informatics Association, 67,577 patients completed a virtual urgent care screening through the MUSC's COVID-retooled system between March 7 and April 22; 14,924 met the criteria for COVID-19 testing.
The VUC system had been in place since 2015, explained the team in the manuscript, with the goal of providing rapid access to care for low-risk clinical conditions.
"We redesigned our existing virtual urgent care with the overall goal of keeping patients with symptoms away from typical healthcare access points such as clinics, urgent care settings, and emergency departments in order to reduce risk," said the MUSC team.
The team noted that the asynchronous capability provided by the Zipnosis platform proved crucial "due to very high patient volumes during which it would have been impossible to accommodate through video or chat functions."
MUSC also repurposed research teams to support clinical care through the institution's Biomedical Informatics Center.
The BIMC team used a dedicated registry of patients with confirmed or potential COVID-19 diagnoses to support screening, testing, remote home-monitoring and convalescent phases. It also expanded the capabilities of Epic's MyChart system with Bluetooth pulse oximeters and digital thermometers available through a smartphone app.
"Finally, we applied artificial intelligence expertise to data coming from MUSC’s virtual urgent care platform (Zipnosis) which, while rich in clinical information, was locked in the 'free text' record," the MUSC team explained.
Researchers developed natural language processing tools and applied deep-learning neural networks to notes in order to predict COVID-19 diagnoses and prioritize testing schedules.
In order to preserve personal protective equipment, the team converted existing continuous virtual monitoring technology – previously used for patients at increased fall risk – to communicate with patients in units.
"This represents a significant reduction in healthcare worker exposure, time spent donning and doffing PPE, and conservation of a substantial amount of PPE assuming these workers would have otherwise gone into the patient’s room," wrote the researchers.
"The estimated cost savings for this program as of April 22, 2020 are $105,624, not including any potential savings from avoided COVID-19 cases among healthcare workers," they added.
Following changes at the federal level to facilitate access to telehealth – namely, relaxation in regulations from the Centers for Medicare and Medicaid Services and the Office of Civil Rights – MUSC also ramped up its outpatient virtual services.
Although the health system began by using the existing video system embedded in patient portals, it deployed a second video portal, Doxy.me, to assist with clinical technological needs.
"By the 30-day mark, clinic volumes for the enterprise were at 69% of the pre-pandemic levels, with 67% of the volume obtained through virtual care," researchers wrote. "Many departments had achieved 80% of pre-pandemic levels. Some departments such as psychiatry and primary care were exceeding their pre-pandemic clinical volumes."
Researchers predicted that the transition to video-based ambulatory care "will ultimately be the most enduring aspect of the telehealth implementations to date," though they noted barriers to adoption, such as some patients lacking access to the necessary technology.
MUSC argues that its existing infrastructure allowed it to scale up telehealth offerings within a relatively short period of time in response to the pandemic – a change that will undoubtedly prove indispensable as coronavirus case numbers in South Carolina continue to climb.
"This study describes the experiences and early results from one health system. It is possible the results are not generalizable to others," acknowledged the researchers.
"Yet, we suggest other health systems have the opportunity to learn from our experiences to inform their own work," they added.
Kat Jercich is senior editor of Healthcare IT News.
Healthcare IT News is a HIMSS Media publication.