How AI could help doctors monitor children born with common congenital heart defect

by Yingshuang Gao

edited by Gaby Clark, reviewed by Robert Egan

Gaby Clark

Scientific Editor

Meet our editorial team
Behind our editorial process

Robert Egan

Senior Editor

Meet our editorial team
Behind our editorial process Editors' notes

This article has been reviewed according to Science X's editorial process and policies. Editors have highlighted the following attributes while ensuring the content's credibility:

fact-checked

peer-reviewed publication

trusted source

written by researcher(s)

proofread

The GIST Add as preferred source

Every echocardiogram is a moving story. For a baby born with a complex heart condition, the gray and black images on the ultrasound screen can influence some of the earliest and most important decisions a medical team makes: What exactly is wrong with the heart? How urgent is surgery? What should doctors watch for after repair?

In our recent work, we focused on tetralogy of Fallot, often shortened to TOF. It is one of the most common cyanotic congenital heart defects. The condition involves several structural abnormalities of the heart, and many children with TOF need careful evaluation, surgery and long-term follow-up. The research is published in the journal eBioMedicine.

Echocardiography is central to that process. It is widely used, noninvasive and rich in clinical information. But it is also demanding. Clinicians must identify the correct views, interpret moving images, measure small cardiac structures, and combine these pieces of information with the patient's clinical course. Even experienced clinicians can face heavy workloads, and interpretation can vary between observers.

That is where we thought artificial intelligence might be useful—not as a replacement for doctors, but as a tool to make parts of the workflow more consistent and informative.

Our team developed an AI-assisted framework called DynaTOF. I think of it as an attempt to connect two questions that are often handled separately. The first is: Can we help identify TOF from echocardiography more consistently? The second is: Can information available before surgery help us anticipate a child's postoperative recovery pattern and follow-up risk?

To answer these questions, we designed DynaTOF around several steps. First, the framework recognizes standard echocardiographic views, such as apical and parasternal views. This matters because AI cannot give meaningful output if it is looking at the wrong image view.

Second, the framework locates and measures key cardiac diameters from echocardiographic images. These measurements are familiar to clinicians, but automating parts of the process may help reduce repetitive manual work and improve consistency.

Third, DynaTOF combines two kinds of information: visual features from echocardiographic videos and quantitative information from cardiac measurements. In our study, this multimodal approach performed better than using either videos or measurements alone. That result was important to us because it reflects how clinicians think. They rarely rely on only one number or one frame. They integrate many clues.

The part of the work I find most personally meaningful is the recovery prediction. For many congenital heart diseases, surgery is not the end of care. Children need follow-up, and some may be at higher risk of complications or abnormal recovery patterns. We designed DynaTOF to use preoperative echocardiographic information, surgery type and follow-up timing to estimate possible postoperative recovery trajectories.

In practical terms, this means the framework tries to sketch a possible recovery pattern. It does not tell a family what will definitely happen. It does not replace clinical judgment. But it may help doctors identify which patients deserve closer attention and more careful monitoring.

This distinction is important. AI in medicine is sometimes presented as if it will make final decisions on its own. I do not see it that way. In pediatric cardiology, the most useful AI systems should support clinicians by organizing information, reducing avoidable variation and highlighting patterns that may otherwise be difficult to see.

Our study used data from multiple medical centers and included healthy controls, patients with conditions that can mimic TOF, and patients with confirmed TOF. We wanted the framework to face a more realistic diagnostic problem than simply separating healthy hearts from diseased hearts. In real clinical practice, the difficult cases are often the ones that look similar.

The results were encouraging. DynaTOF showed strong performance in supporting TOF assessment, predicting postoperative abnormal score patterns, and stratifying follow-up risk. But I also think it is important to be cautious. A model that performs well in a study still needs careful evaluation before it can be used broadly in clinical settings. Different hospitals, ultrasound machines, patient populations, and clinical workflows can all affect performance.

For me, the broader lesson is that medical AI should be built around the clinical pathway, not just around a single technical task. A child with TOF does not need only an image classification result. The child needs diagnosis, surgical planning, postoperative monitoring, and long-term care. If AI is to become genuinely useful, it should help connect these stages.

I hope our work contributes to that direction. The goal is not to make echocardiography less human. The goal is to give clinicians better tools so that more of their time and attention can go toward the children and families who need them.

This story is part of Science X Dialog, where researchers can report findings from their published research articles. Visit this page for information about Science X Dialog and how to participate.

Publication details

Qiang Gao et al, Echocardiography-based intelligent diagnosis and risk stratification management for tetralogy of Fallot, eBioMedicine (2026). DOI: 10.1016/j.ebiom.2026.106292

Journal information: EBioMedicine

Key medical concepts

Tetralogy of FallotEchocardiography

Clinical categories

CardiologyPediatricsChildren's health Who's behind this story?

Gaby Clark

MA in English, copy editor since 2021 with experience in higher education and health content. Dedicated to trustworthy science news. Full profile →

Robert Egan

Bachelor's in mathematical biology, Master's in creative writing. Well-traveled with unique perspectives on science and language. Full profile →

Yingshuang Gao is a PhD candidate in Statistics at Shanghai Jiao Tong University and a co-first author of the EBioMedicine study on DynaTOF. Gao’s research focuses on statistical modelling, medical artificial intelligence, echocardiographic data analysis and clinical decision-support methods for congenital heart disease.

Citation: How AI could help doctors monitor children born with common congenital heart defect (2026, June 24) retrieved 24 June 2026 from https://medicalxpress.com/news/2026-06-ai-doctors-children-born-common.html This document is subject to copyright. Apart from any fair dealing for the purpose of private study or research, no part may be reproduced without the written permission. The content is provided for information purposes only.