Cardiac Response to Live Music Performance: Computing Techniques for Feature Extraction and Analysis

Overview & Synopsis

Strong emotions and mental stress have been linked to potentially deadly arrhythmias. Music evokes strong emotion through the regulation of tension and release and the modulation of changes and transitions. We exploit this in a novel study involving patients with implanted cardiac defibrillators to study the impact of live music performance on cardiac electrophysiology. The patients' heart rates are artificially fixed with pacing at the higher of 80 beats per minute or 10 above the heart's intrinsic rate. We make continuous recordings directly from the heart muscle whilst the patients are listening to a short classical music concert, which is concurrently recorded in a separate stream. The participants provide annotations of perceived boundaries/transitions and felt tension. The recorded cardiac and music information is further processed to extract relevant features. Here, we describe the experiment design, and the mathematical and computing techniques used to represent and abstract the features from the recorded data. Cardiac reaction is measured by the action potential duration (APD), approximated using the action recovery interval (ARI). The expressive parameters extracted from the music include the time varying loudness, tempo, and harmonic tension. The synchronized information layers allow for detailed analysis of immediate cardiac response to dynamically varying expressive nuances in performed music.