AI May Predict Rheumatic Heart Disease as Accurately as a Cardiologist

A new study showed that artificial intelligence (AI) can potentially detect rheumatic heart disease (RHD) with the same level of accuracy as a cardiologist. The findings were published in the Journal of the American Heart Association.

Researchers from Children’s National Hospital analyzed 511 echocardiograms in children (229 normal, 282 had RHD), focusing on color Doppler images of the mitral valve. They then tested a variety of modalities in machine learning and trained a model to interpret ultrasound images of the heart to detect RHD.

The machine learning model identified the correct view with an average accuracy of 0.99, and it identified the correct systolic frame with an average accuracy of 0.94 (apical) and 0.93 (parasternal long axis). Moreover, it localized the left atrium with an average Dice coefficient of 0.88 (apical) and 0.90 (parasternal long axis). The researchers noted that the model showed an area under the receiver operating characteristic curve of 0.84, precision of 0.78, recall of 0.98, and F1 score of 0.87.

“This technology has the potential to extend the reach of a cardiologist to anywhere in the world,” said Kelsey Brown, MD, a cardiology fellow at Children’s National and co-lead author on the manuscript with Staff Scientist Pooneh Roshanitabrizi, PhD, in a press release. “In 1 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.”