by Yale School of Medicine

Baseline AI-ECG screening for LVSD and all-cause mortality risk. Adjusted cumulative hazard curves for all-cause mortality stratified by the AI-ECG LVSD probability. Credit: Circulation: Cardiovascular Quality and Outcomes (2024). DOI: 10.1161/CIRCOUTCOMES.124.011504

In a new study published in Circulation: Cardiovascular Quality and Outcomes, researchers at Yale School of Medicine say an artificial intelligence (AI) tool can use electrocardiographic (ECG) images to define the risk of cardiac dysfunction for patients undergoing cancer treatments.

"There are certain treatments for breast cancer and lymphomas, such as anthracyclines and trastuzumab, that have cardiotoxic effects," said Rohan Khera, MD, Assistant Professor of Medicine, Director of the Cardiovascular Data Science (CarDS) Lab, and senior author of the study. "1 in 10 individuals receiving these drugs individually, and nearly 1 in 3 using the drugs in combination, have adverse cardiac reactions."

After identifying 1,550 patients who had received anthracyclines or trastuzumab to treat breast cancer or non-Hodgkin lymphoma, researchers applied an AI model of left ventricular systolic dysfunction (LVSD) to ECG images of patients to sort them into low-, intermediate-, and high-risk groups.

The team says compared with low-risk patients, patients in the high-risk group had 3.4 times the risk of cardiac dysfunction after cancer treatment, and 13.5 times the risk of a left ventricular ejection fraction (LVEF) below 40%.

The new AI-driven method could help quickly make safe clinical decisions for patients with breast cancer or non-Hodgkin lymphoma at risk of heart dysfunction, especially for providers in under-resourced communities. "Identifying those at heightened risk on a simple tool like an ECG can allow such risk assessment to be extended across low- and high-resource settings and substantially reduce the burden for risk assessment," Khera says.

Researchers say AI models have the potential to improve diagnostics and care across the industry. "We were able to find a way to provide information that typically required an advanced test by using AI applied to a simple and scalable diagnostic," Khera said.

"These signatures are not discernable to even experts, highlighting the role of AI in augmenting human capacity," added Evangelos K. Oikonomou, fellow in cardiovascular medicine and the first author of the study.

More information: Evangelos K. Oikonomou et al, Artificial Intelligence-Enhanced Risk Stratification of Cancer Therapeutics-Related Cardiac Dysfunction Using Electrocardiographic Images, Circulation: Cardiovascular Quality and Outcomes (2024). DOI: 10.1161/CIRCOUTCOMES.124.011504

Journal information: Circulation: Cardiovascular Quality and Outcomes 

Provided by Yale School of Medicine