In many low- and middle-income countries, children born with heart conditions often face an uphill battle. With up to 90% receiving less than optimal care, the reality is stark. Boston Children’s Hospital is using artificial intelligence (AI) to extend the reach of paediatric cardiology to places where advanced diagnostic tools are scarce.
Cardiology fellow Joshua Mayourian, MD, PhD, and cardiologist John Triedman, MD, started the Congenital Heart Artificial Intelligence (CHAI) Lab to bring together heart specialists, computer scientists, and data experts. Their goal? To create AI tools that help detect and manage paediatric heart conditions more efficiently. As Mayourian puts it, “We have plenty of resources here to provide excellent care, but in many regions, doctors simply don’t have the same capabilities. These AI tools can flag heart issues early, meaning children get the care they need sooner.”
A major area of focus in the CHAI Lab is the application of AI to electrocardiograms (ECGs). Although an ECG is a routine and affordable test, it holds a wealth of information about heart rhythms and structural conditions. By integrating AI, clinicians can interpret ECG data in ways that traditional methods might miss. Studies have shown that these AI-enhanced ECGs not only detect heart problems in children and adults with congenital heart disease but also often outperform standard commercial software when it comes to identifying complex conditions like Wolff-Parkinson-White syndrome and long QT syndrome.
The lab is continually refining its AI-ECG models to pick up on subtle signals that could indicate early signs of heart dysfunction—signals that might be overlooked by even experienced cardiologists. One such model can predict ventricular dysfunction, where the heart isn’t pumping as it should, using just the data from an ECG. “The AI is catching delicate variations in how the heart muscle activates, which might be linked to declining function,” explains Mayourian.
Another key advantage is the lab’s access to Boston Children’s vast, secure medical database, organised with the hospital’s unique Fyler code system. This treasure trove of decades worth of anonymised patient records and heart images is crucial for training AI tools to recognise patterns in congenital heart disease, informing both present and future care strategies.
Mayourian and Triedman see great promise in AI-ECGs, especially in areas where there’s a shortage of heart specialists. Since ECG machines are both cost-effective and widely available, these AI models can empower clinicians around the world to quickly identify which children need urgent attention, helping to make the best use of limited resources. By testing their models on diverse populations and gathering feedback from healthcare professionals and patients alike, the team is committed to ensuring their tools are both safe and reliable.
