A machine learning approach for LQT1 vs LQT2 discrimination

Long QT syndrome (LQT) is a congenital disease caused by a mutation of genes that leads to a distortion and a prolongation of the T-wave on standard ECG. The present study proposes an algorithm to automatically discriminate between patients with type 1 or type 2 LQT syndrom. The core of the method is the modeling of the T-wave recomputed on its principal lead by a single parameterized function named Bi-Gaussian Function (BGF). From all the features computed from this model, a statistical analysis was performed to select only the most relevant ones for the discrimination. A classifier was then designed through a Linear Discriminant Analysis (LDA). A database composed of 410 LQTS patients whose genotype is known was used to train the classifier and evaluate its performances.