Heart Rate Variability: A possible machine learning biomarker for mechanical circulatory device complications and heart recovery

Cardiovascular disease continues to be the number one cause of death in the United States, with heart failure patients expected to increase to >8 million by 2030. Mechanical circulatory support (MCS) devices are now better able to manage acute and chronic heart failure refractory to medical therapy, both as bridge to transplant or as bridge to destination. Despite significant advances in MCS device design and surgical implantation technique, it remains difficult to predict response to device therapy. Heart rate variability (HRV), measuring the variation in time interval between adjacent heartbeats, is an objective device diagnostic regularly recorded by various MCS devices that has been shown to have significant prognostic value for both sudden cardiac death as well as allcause mortality in congestive heart failure (CHF) patients. Limited studies have examined HRV indices as promising risk factors and predictors of complication and recovery from left ventricular assist device therapy in end-stage CHF patients. If paired with new advances in machine learning utilization in medicine, HRV represents a potential dynamic biomarker for monitoring and predicting patient status as more patients enter the mechanotrope era of MCS devices for destination therapy.