Background Cardiovascular diseases, particularly coronary heart disease (CAD), are the leading causes of death in industrialized nations. The 'gold standard' for diagnosing ischemic heart disease is invasive coronary angiography, despite its associated risks, costs, and extensive effort. An emerging solution is the utilization of artificial intelligence (AI) in 3D vectorcardiography for non-invasive detection of obstructive coronary artery disease, incorporating deep neural networks within a supervised learning model.

Objective This study aims to assess the sensitivity and specificity of this AI-enhanced 3D vectorcardiography for the non-invasive identification of obstructive coronary artery disease, in comparison to invasive coronary angiography.

Methods A prospective blinded study enrolled 183 patients suspected of cardiac ischemia and CAD, all diagnosed via invasive coronary angiography. Prior to coronary angiography, each participant underwent an AI-driven 3D vectorcardiography. A deep neural network AI algorithm calculated various parameters, including a perfusion index, from this data, and subsequently computed a probability of obstructive CAD. The separate investigators evaluating the AI-driven 3D vectorcardiography and the coronary angiographies remained unaware of each other's findings.

Results The AI-enhanced 3D vectorcardiography prediction of obstructive CAD yielded an area under the curve (AUC) of the receiver operating characteristic (ROC) of 0.621, comparable to the standard risk prediction model based on age, gender, and standard risk factors (P=NS). However, incorporating these parameters with standard risk factors significantly improved the prediction power with an AUC of 0.719.

Conclusion AI-driven 3D vectorcardiography represents a straightforward, non-invasive diagnostic tool for detection of obstructive coronary artery disease. Further studies are required to assess this technique's applicability to other cardiovascular diseases.