AI Breakthrough: Heart Conditions Accurately Diagnosed via Chest X-rays
Researchers at Osaka Metropolitan University have developed a groundbreaking AI application that accurately categorizes heart functions and identifies valvular heart disease using chest X-rays. Valvular heart disease, often leading to heart failure, is typically diagnosed with echocardiography, which requires specialized skills.
This AI model, trained on a multi-institutional dataset of 22,551 chest radiographs and echocardiograms, demonstrated impressive results. It successfully classified six types of valvular heart disease with an Area Under the Curve (AUC) ranging from 0.83 to 0.92. Notably, it achieved an AUC of 0.92 for detecting left ventricular ejection fraction, a crucial measure for monitoring cardiac function. This innovation promises to enhance diagnostic efficiency, particularly in areas lacking specialists and during nighttime emergencies.