Detection of atrial septal aneurysm on ECG based on Deep Learning algorithm (ANN)

Atrial Septal Aneurysm (ASA) is a real clinical challenge due to its possible association with other relevant conditions. The absence of specific symptoms or electrocardiogram (ECG) criteria explain why its diagnosis is very often qualified as incidental. The aim of this study is to assess ASA detection by Machine Learning (ML) on electrocardiogram (ECG) data. The study is a retrospective analysis of 233 individuals, including 123 with ASA confirmed by trans-thoracic Echocardiography (TTE) and 110 without ASA. Key ECG parameters were examined. An Artificial Neural Network (ANN) algorithm was trained on 80% of the dataset, with the remaining 20% for testing. Results demonstrated a model sensitivity of 73%, specificity of 84%, Positive Predictive Value (PPV) of 80%, Negative Predictive Value (NPV) of 73%, and an F-1 score of 0.79. The Receiver Operating Characteristic (ROC) curve exhibited an Area Under the Curve (AUC) of 0.8, indicative of excellent diagnostic test performance. This study shows that ASA detection by ECG using ML is possible, offering a potential opening for a broader clinical understanding and implications of this cardiac abnormality.