Researchers from Emory University’s School of Medicine and Georgia Institute of Technology are investigating the potential of artificial intelligence (AI), specifically natural language processing (NLP), to improve the efficiency of diagnoses and treatment in healthcare delivery. The COVID-19 pandemic led to a surge in telemedicine and electronic health record (EHR) messaging, allowing patients to report positive COVID-19 test results and receive treatment remotely. However, the lack of a digitized triage system caused a backlog of messages, delaying timely responses and access to treatment.
In a recent study published in JAMA Network Open, the research team developed an NLP model to classify patient-initiated EHR messages accurately. They evaluated the model’s performance at five Atlanta-area hospitals from March 30 to September 1, 2022. The study included 3,048 messages reporting positive COVID-19 test results. When a positive test was reported via EHR, the NLP model swiftly classified the patient message.
The findings revealed that the NLP model achieved a classification accuracy of 94% for patient messages. Moreover, faster responses to patient messages increased the likelihood of patients receiving antiviral medical prescriptions within a five-day treatment window.
Lead author Nell Mermin-Bunnell, a third-year student at Emory School of Medicine, expressed excitement about how NLP accurately and instantly triaged patient messages for positive COVID-19 tests, improving patient access to treatment. The authors also highlighted the potential to extend the model’s scope beyond COVID-19 diagnoses.
Co-author May Wang, Ph.D., a professor at Georgia Tech, emphasized the power of advanced NLP models in real-time identification of patients at risk of specific diseases. The study demonstrated that healthcare access speed can be significantly increased through such models.
The study was a collaborative effort among Emory University, Georgia Tech, and Switchboard, MD, a data science and AI company founded by physicians from Emory Healthcare. The NLP model used, called eCOV, was initially developed by Dr. Blake Anderson, CEO of Switchboard, MD, and an Emory primary care physician. The model aimed to organize incoming messages and alleviate the cognitive load on clinical staff, ultimately expediting patient care.
The researchers are now looking to assess the model’s impact on clinical outcomes. They believe that as AI becomes more integrated into healthcare delivery, it has the potential to reshape medicine. Dr. Anderson emphasized that NLP prioritizes human interactions rather than replacing them, addressing concerns about the use of AI in healthcare.
Source: Emory University