Most LVAD complications are not detected until late in the disease process when patients present with severe or life-threatening symptoms, resulting in the need to pursue high risk rescue strategies and prolonged hospital courses.

We will provide patients with digital stethoscopes and ask them to record and upload their cardiac tones to the HIPAA compliant Duke Box once per week.

We will relate the acoustic spectra of the recordings to clinical events and use this information to develop acoustic signatures, via statistical models, associated with LVAD complications. In future work, we will develop a mobile health platform for capturing patient-obtained LVAD sounds, applying our acoustic predictors, and notifying providers of impending LVAD complications based on model predictions.