Association of Music Therapy and Physiologic Indices in the Pediatric Cardiac Intensive Care Unit: Insights from Machine Learning and High-fidelity Physiologic Data

Music therapy is increasingly used in critically ill children, yet its real-time physiologic effects remain incompletely characterized. This study used high-fidelity physiology data to characterize the physiologic response to music therapy in pediatric cardiac intensive care unit patients. Physiologic data were extracted from Sickbay. Heart rate, mean arterial blood pressure, central venous pressure, arterial saturation, respiratory rate, end tidal carbon dioxide, and renal near infrared spectroscopy were analyzed from 1 h before through 4 h after the start of therapy. Percent change from baseline was modeled using multivariable regression adjusting for participation level, therapy technique, circulatory physiology, mechanical ventilation, mechanical circulatory support, sedation, and vasoactive inotrope score. Cluster analysis of session-level percent changes was conducted to identify physiologic response phenotypes. A total of 52 music therapy sessions across 33 unique patients were included in the final analyses. Music therapy was independent associated with changes in heart rate (+ 3%, p = 0.03), central venous pressure (+ 10%, p < 0.01), end tidal carbon dioxide (+ 11%, p < 0.01), and renal near infrared spectroscopy (+ 3%, p = 0.02). Cluster analysis revealed two physiologic response clusters: one demonstrating stable hemodynamics and improved adequacy of oxygen delivery and another demonstrating heightened respiratory activity but no improvement in the adequacy of oxygen delivery. Cluster 1, with improvement in the adequacy of oxygen delivery, was associated with active patient participation. Music therapy was independently associated with physiologic changes. In a majority of patients, the adequacy of oxygen delivery improved and was associated with active patient participation.