Researchers in Pittsburgh found that an AI tool outperformed the three most common practices for analyzing ECGs of patients being treated for chest pain, reclassifying one of every three patients.
An AI tool used in three Pittsburgh hospitals was able to diagnose and reclassify 33% of patients being treated for chest pain, improving on the standard practice for identifying heart attacks and potentially saving lives.
The technology, developed by researchers in Toronto, analyzes ECG readings for subtle clues that are often overlooked, leading to delays in detection and treatment. Researchers from the University of Pittsburgh compared the model against the three gold standards for assessing cardiac events and found that the AI tool performed better than all three.
[See also: Cedars-Sinai Team Develops AI Tool to Predict Heart Attacks.]
“When a patient comes into the hospital with chest pain, the first question we ask is whether the patient is having a heart attack or not," Salah Al-Zati, PhD, RN, an associate professor in the Pitt School of Nursing and of emergency medical and cardiology in the School of Medicine, said in a press release issued by UPMC. "It seems like that should be straightforward, but when it’s not clear from the ECG, it can take up to 24 hours to complete additional tests. Our model helps address this major challenge by improving risk assessment so that patients can get appropriate care without delay.”
Al-Zaiti was part of the team that tested the technology on 4,026 patients treated for chest pain at the Pittsburgh hospitals and co-authored the results of the study, which was recently published in Nature Medicine. Those results were independently validated with 3,287 patients from a different health system.
The study compared the technology against experienced clinician interpretations of an ECG, commercial ECG algorithms, and the HEART score, which factors in age, risk factors, and other considerations prior to diagnosis. The model outperformed all three standards, reclassifying one of every three patients into low, intermediate, or high risk.
The study has implications not only for ED treatment, but for those who are first on the scene to treat patients with chest pain.
“This information can help guide EMS medical decisions such as initiating certain treatments in the field or alerting hospitals that a high-risk patient is incoming,” Christian Martin-Gill, MD, MPH, chief of the Emergency Medical Services division at UPMC and co-author of the study, said in the press release. “On the flip side, it’s also exciting that it can help identify low-risk patients who don’t need to go to a hospital with a specialized cardiac facility, which could improve prehospital triage.”
Martin-Gill and his team are testing that concept in the next phase of their research. They're working with the City of Pittsburgh Bureau of Emergency Services to deploy the model through the cloud to hospital command centers, which can direct risk assessments back to EMS teams in the field for more timely diagnosis and treatment.
Eric Wicklund is the associate content manager and senior editor for Innovation at HealthLeaders.
KEY TAKEAWAYS
The three gold standards for assessing cardiac events are experienced clinician interpretation of an ECG, commercial ECG algorithms, and the HEART score, which takes into account other factors including patient age and risk factors.
An AI tool developed in Toronto assesses ECG readings in more detail, looking for subtle changes that are often difficult to spot.
Researchers used this tool on patients in 3 Pittsburgh hospitals and found that it outperformed all three gold standards, reclassifying 33% of the patients into different risk categories.