Aug. 01, 2025
Mayo Clinic's artificial intelligence (AI) cardiology team is researching and applying new approaches to risk prediction and diagnosis for adults and children with serious or complex heart problems. Using AI algorithms to analyze electrocardiogram (ECG) data, for example, can potentially enhance the detection, diagnosis and management of various cardiac conditions.
Using the basic building blocks of AI systems, known as neural networks, Mayo Clinic researchers train computer systems by introducing hundreds of thousands of sets of similar readings for AI to analyze. AI becomes experienced in looking at a specific problem, resulting in the system reading test results, detecting conditions and then predicting potential future conditions.
Hundreds of AI projects at Mayo Clinic are currently in various stages of development, including collaborative research for advancing the use of AI-enabled ECGs and analysis to improve diagnostic accuracy and patient outcomes.
Using AI-enabled ECGs to detect right and left ventricular dysfunction in pediatric patients
The pathogenesis of ventricular dysfunction can vary, and the symptoms are often nonspecific, which can lead to delayed diagnosis and underdetection. In a study published in the May 2019 issue of The Journal of Pediatrics, half of the pediatric patients with new‐onset heart failure were missed at first presentation and underwent significant nonrelevant treatment and testing.
Currently, there are existing AI algorithms that can identify left ventricular systolic dysfunction (LVSD) and right ventricular systolic dysfunction (RVSD) in adults. Because the efficacy of the use of these algorithms in pediatric patients is uncertain, Mayo Clinic Children's researchers have developed novel AI-enabled ECG algorithms for LVSD and RVSD detection in children. The results were published in the November 2024 issue of the Journal of the American Heart Association.
"Early detection of LVSD and RVSD in children can lead to initiation of medical therapy that improves heart failure symptoms and reduces mortality rate," says Talha Niaz, M.B.B.S., a cardiovascular geneticist and pediatric cardiologist at Mayo Clinic Children's in Rochester, Minnesota. "The artificial intelligence-enabled ECG algorithms we developed demonstrate accurate detection for both LVSD and RVSD in pediatric patients."
From a cohort of 10,142 unique pediatric patients, Mayo Clinic researchers trained novel AI models to accurately detect LVSD and RVSD from ECGs in the pediatric population. All models demonstrated excellent area under the curve (AUC) for detection of abnormalities: severe LVSD (LVEF ≤ 35%) achieved an AUC of 0.93; moderate LVSD (LVEF < 50%) reached an AUC of 0.88; and the model for detecting any degree of RVSD achieved an AUC of 0.90. The AI-enabled ECG model identified some patients as having heart dysfunction despite normal function at the time, and they were later found to be over three times more likely to develop dysfunction in the future.
AI-enabled ECG screening for LVSD and RVSD can provide a low‐cost, easily accessible option for evaluating pediatric patients in various clinical settings, to help identify individuals in need of ECG for definitive diagnosis.
Using AI-enabled ECG analysis for pediatric sex estimation and the influence of pubertal development
In a study published in the July 2024 issue of Nature, Mayo Clinic researchers tested the ability of AI-enabled ECGs to predict sex in the pediatric population and study the influence of pubertal development.
"In this study, we saw that AI can pick up on the effects of the different hormones and their effects on the human cardiovascular system," says Dr. Niaz. "It was also able to determine the pivotal time frame where you can actually see the true effects of the hormones on the cardiovascular system as it matures."
There are certain cardiovascular conditions that have different risk profiles between males and females after puberty, including long QT syndrome, aortic aneurysms and sudden cardiac death. "The results of this study highlight the potential to understand when hormones can truly affect the cardiovascular system and the opportunity for future applications with other conditions that are affected by puberty," says Dr. Niaz.
The study concluded that AI-enabled interpretation of the 12-lead ECG can accurately estimate sex in peripubertal and postpubertal children, but not in prepubertal children. "We showed the robustness of AI-ECG in sex discrimination in the postpubertal group, having trained and validated on a heterogeneous pediatric population," says Dr. Niaz. "Further research is necessary to evaluate the utility of AI-predicted ECG sex to more accurately assess pubertal status in patients for sex-risk stratification in certain pediatric cardiovascular diseases. Additionally, investigation is needed into how AI-ECG can be utilized among patients with hormonal or pubertal disorders, where traditional markers may be less reliable."
Through continued research, refining AI algorithms and addressing challenges collaboratively, AI-enabled ECG has the potential to revolutionize cardiovascular diagnostics and management, leading to earlier and more-accurate diagnoses, individualized treatments and improved patient outcomes.
Additional research
Research in progress
- Artificial intelligence-enabled electrocardiogram for age estimation in pediatric and congenital heart disease patients.
- AI-ECG and detection of congenital heart defects in children.
For more information
Puri K, et al. Missed diagnosis of new-onset systolic heart failure at first presentation in children with no known heart disease. The Journal of Pediatrics. 2019;208:258.
Anjewierden S, et al. Detection of right and left ventricular dysfunction in pediatric patients using artificial intelligence-enabled ECGs. Journal of the American Heart Association. 2024;13:e035201.
O'Sullivan D, et al. Pediatric sex estimation using AI-enabled ECG analysis: influence of pubertal development. Nature 2024;7:176.
Refer a patient to Mayo Clinic.