A Machine Learning approach for Atrial Fibrillation detection in telemonitored patients

PEDRO BARRERA; LORENZA VECINO SCHANDY; PAULA BONOMINI; CRISTIAN MATEOS; MATÍAS HIRSCH; LUCAS GRANA; SERGIO LIBERCZUK

Ciudad Autónoma de Buenos Aires

Congreso; XXIV Congreso de Bioingeniería - XIII Jornadas de Ingeniería Clínica; 2023

Universidad Arturo Jauretche - Sociedad Argentina de Bioingeniería

Atrial fibrillation (AF) is the most common type of cardiac arrhythmia [1]. As it is typically asymptomatic, it often goes undiagnosed until major complications arise, such as stroke [2]. Therefore, the development of rapid, economical, and widely accessible diagnostic tools for detecting AF at an early stage is crucial. Telemonitoring with machine learning-assisted devices shows promise in achieving this goal. This paper presents an algorithm that automatically detects AF in signals obtained by portable electrocardiographs connected to a telemonitoring platform via smartphones. The algorithm consists of two stages: a noise detection stage and an AF detection stage. The noise detection stage involves analyzing the ECG signals using 5-second windows with a 1-second shift. A support vector machine (SVM) classifier predicts the presence or absence of noise in each window, allowing for the detection of noisy and non-noisy segments of the signal. In the AF detection stage, the non-noisy segments are processed using a Pan-Tompkins algorithm to find the R peaks of the signal, and the corresponding RR interval series. This RR series is used to calculate five indices, which serve as input features of a logistic regression classifier that predicts the presence or absence of AF in the ECG signal. The algorithm was trained and tested using the Physionet Short Single-Lead AF Database (SSLAFDB) and achieved an accuracy of 87.9% and an F1-score of 87.7%. Further validation was performed by an external partner using two other databases, reporting an accuracy of 93.1% and 89.5% on two different databases.