Exploring the Potential of ML in Congenital Heart Defect Diagnosis: A Scoping Review

ABSTRACT

ML is a method of data analysis and prediction that relies on previous information. It has been widely used in healthcare to enhance diagnostic categorization by the uncovering of hidden patterns in data that may be invisible to the naked eye. One of the most exciting areas of clinical application for ML is in the field of congenital heart defect (CHD) diagnosis, where rapid and precise analysis is of the utmost importance. The purpose of this scoping review is to provide a concise summary of the clinical relevance and use of ML techniques in paediatric cardiology research, with a special emphasis on methods that aim to improve CHD diagnosis and evaluation. Forty percent of the found full-text papers between 2015 and 2021 focused on improving CHD diagnosis and evaluation. With an overall diagnosis accuracy of > 0.80, the most popular algorithms utilised were deep learning and support vector machine. The main clinical applications were classification of auscultatory heart sounds, transthoracic echocardiograms, and cardiac magnetic resonance imaging. This review of the literature gives a thorough overview of numerous applications and suggests areas for further research.

FULL TEXT