Detecting and Reversing Myocardial Ischemia Using an Artificially Intelligent Bioelectronic Medicine

Myocardial ischemia is spontaneous, usually asymptomatic, and contributes to fatal cardiovascular consequences. Importantly, biological neural networks cannot reliably detect and correct myocardial ischemia on their own. Supplementing biological neural networks may enable reliable detection, and potentially even facilitate correction, of myocardial ischemia. In this study, we demonstrate an artificially intelligent and responsive bioelectronic medicine, where an artificial neural network (ANN) supplements biological neural networks enabling reliable detection and correction of myocardial ischemia (preclinical experiments in rats). This responsive bioelectronic medicine uses an ANN with a long short-term memory layer to decode spontaneous myocardial ischemia and autonomously trigger vagus nerve stimulation (VNS) for reducing chronotropy, afterload, and myocardial oxygen demand. We first used injections of cardiovascular stress inducing agents (dobutamine and norepinephrine) that produce a model of spontaneous myocardial ischemia. Next, ANNs were trained to decode spontaneous cardiovascular stress and myocardial ischemia, with an overall accuracy of ~92%. ANN-controlled VNS reversed the major biomarkers of myocardial ischemia with no side effects. In contrast, open-loop VNS or ANN controlled VNS following a caudal vagotomy essentially failed to reverse correlates of myocardial ischemia. Lastly, variants of ANNs were used to meet clinically relevant needs, including interpretable visualizations of stress pathophysiology and unsupervised detection of new emerging stress states. Together, this adaptive architecture provides clinically relevant insights as pathophysiology evolves. Overall, these results provide a first-time demonstration that ANNs can supplement deficient biological neural networks to dynamically detect and help bioelectronically reverse cardiovascular pathophysiology.