Researchers from the Emory University School of Medicine and the Georgia Institute of Technology are investigating how the use of artificial intelligence (AI) could expand access and increase the efficiency of diagnoses and treatments for COVID-19.
The use of telemedicine and electronic health record (EHR) messaging increased significantly during the COVID-19 pandemic, and the widespread availability of at-home tests allowed patients to report a positive test and start treatment without visiting their physician’s office. Although this shift in health care delivery has many benefits, the researchers noted that the influx of messages without a digitized triage system can slow responses and delay access to timely treatments.
Now, the researchers say AI could help sift through these messages and streamline processes.
“We’re trying to take a mountain of incoming data and extract what’s most relevant for people who need to see it so patients can get care faster,” said senior author Blake Anderson, MD, CEO of Switchboard and an Emory primary care physician, in a press release.
The study, published in JAMA Open Network, examined how natural language processing (NLP) AI could speed up the time between a patient-initiated message, a physician response, and access to antiviral treatment for COVID-19. Building off of previously tested deep learning predictive models, the team developed a novel NLP model to classify patient-initiated EHR messages and evaluated its accuracy at 5 Atlanta-area hospitals between March 30 and September 1, 2022.
Over the course of the study, 3048 messages reported test results positive for COVID-19. The NLP model was initiated when a positive test was reported via EHR.
The study findings show that the NLP model classified patient messages with 94% accuracy. Additionally, when responses to patient messages occurred faster, patients were more likely to receive an antiviral prescription within a 5-day treatment window.
“We were excited to see how NLP accurately and instantaneously triaged patient messages reporting a positive COVID-19 test and helped improve patient access to treatment,” said study lead author Nell Mermin-Bunnell, a third-year student at the Emory School of Medicine, in the press release. “While this model proved effective for this specific application, there are opportunities to broaden the scope beyond COVID-19 diagnoses.”
The NLP model used in the study, called eCOV, was initially developed by Anderson. As more patients began using EHR to communicate with their care team, Anderson saw a need to better organize incoming messages to ease the load on clinical staff and alleviate burnout. Anderson and his colleagues conducted experiments to evaluate the model’s performance, and they identified an algorithm to account for the context of the message, not just keywords.
“The results illustrate the power of using advanced NLP models in accurately identifying patients at risk of a certain disease in real time,” said study co-author May Wang, PhD, professor, and Wallace H. Coulter Distinguished Faculty Fellow at Georgia Tech, in the press release. “It showed that the speed for patient access to health care can be significantly increased.”
Further analysis is needed to measure the impact of the model on clinical outcomes, according to the study authors. Even without these data, however, the researchers said it is becoming increasingly clear that AI has the potential to reshape how medicine is practiced as it is further integrated into mainstream health care delivery.
Emory, Georgia Tech use artificial intelligence to accelerate access to COVID-19 treatment. News release. Emory University. July 11, 2023. Accessed July 12, 2023. https://news.emory.edu/stories/2023/07/HS_JAMA_AI_EHR_study_11-07-2023/story.html