Using AI to assess syncope risk

One of the most common reasons for Emergency Department (ED) visits is syncope, a form of transient loss of consciousness. With numerous causes, some benign and some with serious, even deadly consequences, management is a major challenge, especially in the ED.

An interdisciplinary team of physicians from the Departments of Internal Medicine and Emergency Medicine is creating an effective point-of-care risk stratification tool that will help ED physicians and first-responders identify people presenting with syncope who are at highest risk of adverse outcomes, including sudden cardiac death and those who can benefit from hospitalization. This tool could have the potential to save lives, reduce the risk of syncope recurrence, prevent unnecessary hospitalizations for those at low risk, save money, and direct further testing for those in need.

Milena A. Gebska, MD, PhD, MME, clinical associate professor in Cardiovascular Medicine, Brian Olshansky, MD, Emeritus Professor in Cardiovascular Medicine, and their team received a 10-week, $15,000 pilot grant from the Iowa Initiative for Artificial Intelligence (IIAI). In collaboration with the IIAI, the Carver College of Medicine offered awards to six projects from the college that could benefit from Artificial Intelligence (AI) and Machine Learning (ML) expertise.

“I am truly honored and humbled by this award. I am incredibly excited to have the opportunity to work with such excellent multidisciplinary teams and mentor our trainees,” Gebska said. “I look forward to learning from Dr. Milan Sonka and the University of Iowa Initiative for Artificial Intelligence team.”

Their funded project, “Novel Machine Learning Algorithms to Risk Stratify Patients with Syncope Presenting to the Emergency Department,” aims to create a data-driven model with AI and ML that will categorize people with syncope into low-risk, intermediate-risk, and high-risk groups to better define which patients may be discharged from the ED and who can benefit from further testing.

While other institutions have attempted to create scoring models, these models have not improved outcomes of patients with syncope who are evaluated in the ED. The CCOM-IIAI research team suspects that unique aspects of AI/ML implementation will successfully improve syncope-related clinical decision-making.

This grant will serve as a pilot study for a larger randomized-controlled trial project investigating the effectiveness of AI/ML-based risk stratification tools.

“This pilot grant truly reflects strong collegiality across multiple disciplines and provides a powerful academic environment,” Olshansky said. “We stand together and hope to address critical gaps in managing patients with syncope.”

Gebska and Olshansky’s research team (in alphabetical order) includes: Antony Anandaraj, MD, cardiovascular fellow at Mercy One North Iowa Heart Center; Tyler Bullis, MD, second-year internal medicine resident; Aron Evans, MD, first-year internal medicine resident; Samuel Johnston, MD, clinical assistant professor in Cardiology and Electrophysiology; Sangil Lee, MD, MS, clinical associate professor of Emergency Medicine; Deepak Kumar Pasupula, MD, MPH, cardiovascular fellow at Mercy One North Iowa Heart Center; Giselle Statz, MD, newly graduated resident in internal medicine and rising cardiovascular fellow; and Jon N. Van Heukelom, MD, clinical professor of Emergency Medicine.

2 Responses

Leave a Reply