Detect heart defects in babies and children

A team from the Department of Engineering has won first prize in the annual PhysioNet Computing in Cardiology Challenge. Participants were asked to design and implement open source algorithms to identify heart murmurs from echo sound recordings.

Noninvasive heart assessment can identify congenital and acquired heart defects in children. Lack of diagnosis, especially in poorer countries with high birth rates, is a major health problem. The data underlying the challenge was collected in two mass screening programs for patients under the age of 21 in northeastern Brazil.

PhysioNet was established in 1999 under the auspices of the National Institutes of Health (NIH) in the United States. It conducts and promotes biomedical research and education, in part by providing free access to vast collections of physiological and clinical data and associated open source software. It has organized annual biomedical challenges since 1999, focusing research on unsolved problems in clinical and basic sciences.

The king’s team consisted of two doctoral students, Yujia Xu and Xinqi Bao, and two academics from the engineering department, Dr. Ernest Kamavuako, senior lecturer and Dr. Hak-Keung (HK) Lam, reader.

Commenting on the challenge, Ernest said:

“I am very proud that our two PhD students won first place in such a competitive challenge with a new signal processing approach.”

Hak-Keung added, “We are delighted that our HearTech+ team was ranked 1st in the 2022 PhysioNet Computing in Cardiology Challenge. Ernest and I are very proud that our PhD students, Yujia Yu and Xinqi Bao, demonstrated excellent ability research to develop state-of-the-art machine learning algorithms to detect heart murmur from phonocardiogram recordings to diagnose congenital heart disease which affects approximately 1% of newborns.

This was an opportunity to analyze lots of patient data and develop an algorithm that recognizes patterns. Such an algorithm paves the way for innovative and life-saving miniaturized devices for detecting and monitoring heart disease.