Professor Kim Kyunghoon's Team at Bundang Seoul National University Hospital
[Asia Economy Reporter Lee Gwan-ju] The pediatric team led by Professor Kim Kyung-hoon at Bundang Seoul National University Hospital announced on the 31st that they have developed an artificial intelligence (AI) model that detects abnormal breath sounds called 'wheezing' in children with respiratory diseases.
Wheezing refers to a respiratory sound characterized by a 'squeaking' noise from the chest each time the child breathes due to airway narrowing and pressure. In children with structurally narrow airways, wheezing often occurs due to asthma, bronchitis, and other conditions, making it one of the most important indicators for early diagnosis of pediatric respiratory diseases.
However, the current method for identifying wheezing remains the traditional 'auscultation' technique, where a stethoscope is placed on the chest to listen directly to breath sounds. Since this is not a test that produces objective numerical data, its accuracy can vary significantly depending on the physician’s experience and judgment, which is a limitation.
To address this issue, Professor Kim’s team embarked on developing an algorithm to distinguish wheezing sounds. The research team used breath sounds from 287 pediatric respiratory patients, cross-verified by pediatric respiratory specialists, for machine learning. Additionally, to maintain an appropriate level of AI learning capability, they applied a 34-layer ResNet artificial neural network technology.
As a result, the developed algorithm demonstrated a high accuracy of 91.2% and a precision (a measure of consistency under the same conditions) of 94.4%, showing sufficient accuracy and stability for clinical application. This analysis requires only a small amount of memory space, making it expected to be applicable to mobile devices in the future, allowing patient-specific condition monitoring without constraints of time and place.
Professor Kim explained, "Children are structurally prone to airway narrowing, making wheezing more likely to occur, and their alveolar surface area is smaller, so their ability to withstand respiratory diseases such as asthma is significantly lower than that of adults. This model will greatly help in early diagnosis of respiratory diseases like asthma to minimize sequelae and establish optimal treatment strategies tailored to individual conditions."
The research results were published in the latest issue of the online academic journal Scientific Reports by the Nature Publishing Group.
© The Asia Business Daily(www.asiae.co.kr). All rights reserved.
