Artificial Intelligence Shows Promise, Comparable to Humans, in Early Detection of Rheumatic Heart Disease

News
Article

In a study involving over 500 children, artificial intelligence using echocardiography demonstrated accuracy comparable to expert cardiologists in detecting rheumatic heart disease, hinting at the prospect of a potentially transformative tool for early intervention and disease prevention.

Kelsey Brown, MD | Credit: LinkedIn

Kelsey Brown, MD
Credit: LinkedIn

Artificial intelligence could prove useful in the detection of rheumatic heart disease in children, according to the results of a new study.

An analysis comparing artificial intelligence using echocardiography against clinician detection among more than 500 children, results of the study suggest the artificial intelligence tool was as accurate as an expert cardiologist, with potential to improve with integration of additional data.1

“This technology has the potential to extend the reach of a cardiologist to anywhere in the world,” said co-lead investigator Kelsey Brown, MD, a cardiology fellow at Children’s National.2 “In one minute, anyone trained to use our system can screen a child to find out if their heart is demonstrating signs of RHD. This will lead them to more specialized care and a simple antibiotic to prevent this degenerative disease from critically damaging their hearts.”

Although rheumatic heart disease has a reduced prevalence in the US relative to other countries, early identification before onset of symptoms remains a crucial part of disease management by offering an opportunity for early intervention to prevent disease progression. As highlighted by investigators, use of artificial intelligence has shown promise in other arenas of cardiology, specifically calling attention to studies on identifying mitral regurgitation.3,4

With mitral regurgitation present in up to 90% of children with early rheumatic heart disease, investigators sought to develop and test an automated method for rheumatic heart disease detection. The automated method developed by investigators included harmonization of echocardiograms to left atrium during systole using convolutional neural networks and rheumatic heart disease detection using mitral regurgitation jet analysis and deep learning models with an attention mechanism.1

Investigators pointed out their deep learning algorithm incorporated 39 morphological features describing mitral regurgitation jet’s pattern, velocity, duration, and length, including 9 features specifically used for rheumatic heart disease prediction.1

Of note, all echocardiograms included in the study were obtained during 2020 as part of completion evaluations for the GOAL trial, which was a 2-year randomized controlled trial compared intramuscular penicillin G benzathine, as compared with no prophylaxis as a secondary prophylaxis in Ugandan children and adolescents with latent rheumatic heart disease.1

In total, 511 echocardiograms in children were used in the study. Among these, 229 were considered normal and 282 had rheumatic heart disease.

Upon analysis, investigators found the artificial intelligence tool identified the correct view with an average accuracy of 0.99 and the correct systolic frame with an average accuracy of 0.94 (apical) and 0.93 (parasternal long axis). Further analysis indicated the tool was able to localize the left atrium with a mean Dice coefficient of 0.88 (apical) and 0.9 (parasternal long axis).1

Compared to expert manual measurements, maximum mitral regurgitation measurements were similar. Investigators also called attention to a 9-feature mitral regurgitation analysis, which produced an area under the receiver operating characteristics curve of 0.93, precision of 0.83, recall of 0.92, and an F1 score of 0.87. Before concluding, investigators highlighted the deep learning model achieved an area under the receiver operating characteristics curve of 0.84, precision of 0.78, recall of 0.98, and F1 score of 0.87.1

“Our algorithm can see and make adjustments for the heart’s size as a continuously fluid variable,” added Pooneh Roshanitabrizi, PhD, co-lead author on the manuscript and staff scientist at Children’s National Hospital.2 “In the hands of healthcare workers, we expect the technology to amplify human capabilities to make calculations far more quickly and precisely than the human eye and brain, saving countless lives.”

A release from Children's National Hospital, stated members of the investigatory team intended to implement a pilot program incorporating artificial intellect into the echo screening process of children being check for rheumatic heart disease to further explore the utility of the technology.2

“Once this technology is built and distributed at a scale to address the need, we are optimistic that it holds great promise to bring highly accurate care to economically disadvantaged countries and help eradicate RHD around the world,” said Craig Sable, MD, interim division chief of Cardiology at Children’s National Hospital.2

References:

  1. Brown K, Roshanitabrizi P, Rwebembera J, et al. Using Artificial Intelligence for Rheumatic Heart Disease Detection by Echocardiography: Focus on Mitral Regurgitation. J Am Heart Assoc. 2024;13(2):e031257. doi:10.1161/JAHA.123.031257
  2. Children’s National Hospital. Novel AI platform detects rheumatic heart disease. Children’s National Hospital. January 16, 2024. Accessed January 26, 2024. https://innovationdistrict.childrensnational.org/novel-ai-platform-detects-rheumatic-heart-disease/.
  3. Tal R, Hamad Saied M, Zidani R, et al. Rheumatic fever in a developed country - is it still relevant? A retrospective, 25 years follow-up. Pediatr Rheumatol Online J. 2022;20(1):20. Published 2022 Mar 15. doi:10.1186/s12969-022-00678-7
  4. Edwards LA, Feng F, Iqbal M, et al. Machine Learning for Pediatric Echocardiographic Mitral Regurgitation Detection. J Am Soc Echocardiogr. 2023;36(1):96-104.e4. doi:10.1016/j.echo.2022.09.017

Recent Videos
Dilraj Grewal, MD: Development of MNV in Eyes with Geographic Atrophy in GATHER | Image Credit: Duke Eye Center
1 KOL is featured in this series.
1 KOL is featured in this series.
1 KOL is featured in this series.
Margaret Chang, MD: Two-Year Outcomes of the PDS for Diabetic Retinopathy | Image Credit: Retina Consultants Medical Group
Phase 2 Data Shows KP1077 Meaningfully Improves Idiopathic Hypersomnia Symptoms
Carl C. Awh, MD: | Image Credit:
Raj K. Maturi, MD: 4D-150 for nAMD in PRISM Population Extension Cohort | Image Credit: Retina Partners Midwest
1 KOL is featured in this Insights series.
Charles C. Wykoff, MD, PhD: Interim Analysis on Ixo-Vec Gene Therapy for nAMD | Image Credit: Retina Consultants of Texas
© 2024 MJH Life Sciences

All rights reserved.