An AI-enabled ECG algorithm could speed identification and treatment of heart failure among individuals presenting with shortness of breath.
Results of a new study from the Mayo Clinic suggest an artificial intelligence-enabled electrocardiogram (ECG) could aid clinicians in emergency departments more accurately identify heart failure.
Findings from the study indicate the AI-enhanced ECG could improve identification of left ventricular systolic dysfunction in patients presenting the emergency departments with acute dyspnea.
"AI-enhanced ECGs are quicker and outperform current standard-of-care tests. Our results suggest that high-risk cardiac patients can be identified quicker in the emergency department and provides an opportunity to link them early to appropriate cardiovascular care," said lead investigator Demilade Adedinsewo, MD, MPH, chief fellow in the division of cardiovascular medicine at Mayo Clinic in Jacksonville, Florida, in a statement.
With the timely and accurate identification of systolic heart failure still posing a problem for emergency department clinicians, Adedinsewo and colleagues from various Mayo Clinic sites sought to determine whether incorporating an AI-ECG could improve identification of left ventricular systolic dysfunction. To do so, they designed a retrospective cohort study of all adult patients reporting dyspnea presenting to an emergency department at Mayo Clinic sites in Arizona, Florida, and Minnesota, as well as the Mayo Clinic Health Systems between May 2018-February 2019.
For inclusion in the analysis, patients needed to be at least 18 years of age and have at least 1 standard 10-second 12-lead ECG performed within 24 hours of the baseline visit. Patients were excluded if they had a known prior diagnosis of systolic, diastolic, or unspecified heart failure. Patients were also excluded if they did not undergo an echocardiogram within 30 days of the emergency department visit.
Initially, 21,309 patients were identified by investigators. Of these, 1606 were included in the current analysis. The median age of the population was 68 years, 47% were female, 91% were white.
The primary outcome of the study was the identification of new left ventricular systolic dysfunction, which was defined as a left ventricular ejection fraction of 35% or less, within 30 days of the initial visit through use of a deep learning network for ECGs. The secondary outcome of the study was identification of left ventricular ejection fraction less than 50% within 30 days using the same approach.
Investigators noted the algorithm used in the deep learning network employed a convolutional neural network (CNN) trained with Keras with a Tensorflow.
Upon analysis, the AI-ECG algorithm achieved an area under the receiver operating characteristic curve (AUC) of 0.89 (95% CI, 0.86-0.91) with an accuracy of 85.9% (95% CI, 84.1-87.6%) for detecting left ventricular ejection fraction of 35% or less. Results indicated the sensitivity, specificity, negative predictive value, and positive predictive value of the AI-ECG algorithm were 74%, 87%, 97%, and 40%, respectively.
For identifying left ventricular ejection fraction of less than 50%, the AI-ECG algorithm achieved an AUC of 0.85 (95% CI, 84.2-87.7%) with an accuracy of 86% (95% CI, 84.2-87.7%). Results indicated the sensitivity, specificity, negative predictive value, and positive predictive value of the AI-ECG algorithm were 63%, 91%, 92%, and 62%, respectively.
In a subgroup of patients who had NT-proBNP values, an NT-proBNP alone at a cutoff value of greater than 800 was able to identify new left ventricular systolic dysfunction with an AUC of 0.80 (95% CI, 0.76-0.84), which investigators suggest demonstrates the superior diagnostic value of the AI-ECG in this group of patients (P <.0001).
"Determining why someone has shortness of breath is challenging for emergency department physicians, and this AI-enabled ECG provides a rapid and effective method to screen these patients for left ventricular systolic dysfunction," added Adedinsewo, in the aforementioned statement.
This study, “An Artificial Intelligence-Enabled ECG Algorithm to Identify Patients with Left Ventricular Systolic Dysfunction Presenting to the Emergency Department with Dyspnea,” was published in Circulation: Arrhythmia and Electrophysiology.