Next Generation ECG: The Impact of Artificial Intelligence and Machine Learning

AbstractPurpose of ReviewDevelopment of artificial intelligence (AI) models, particularly the application of machine learning (ML) and deep learning (DL) architectures to enhance the current diagnostic armoury for the detection of cardiovascular diseases (CVD) has expanded exponentially over the past five years. We review the current AI landscape in CVD and the impact of applying AI models to the electrocardiogram (ECG).Recent FindingsResearchers continue to explore innovative methods utilising ML and DL to automate ECG diagnosis and gain insights into underlying cardiovascular physiology. The results from numerous studies including two randomised control trials (EAGLE and BEAGLE) demonstrates the wide-ranging potential for the application of AI models to both 12-lead and single lead ECG recordings. This allows a fixed timepoint recording to effectively be extrapolated into a continuous monitoring device to predict outcomes for CVD diseases such as left ventricular systolic dysfunction (LVSD), atrial fibrillation (AF), hypertrophic obstructive cardiomyopathy (HOCM), cardiac amyloidosis, valvular heart disease, pulmonary hypertension and channelopathies. A significant proportion of models developed utilise DL architectures such as convolutional neural networks (CNN). The area under the curve (AUC) for published studies range from 0.66 –0.99 with sensitivities and specificities between 26.29—95% and 79.5—96.6% respectively. These results suggest a possible role for AI-ECG...
Source: Current Cardiovascular Risk Reports - Category: Cardiology Source Type: research