AI-driven music intervention based on five-tone theory for anxiety: a preliminary pre-post feasibility study

Music therapy, as a pivotal non-pharmacological intervention, faces a critical challenge in designing personalized treatment protocols. Current clinical applications of the traditional Chinese five-tone theory to explore the relationship between music and emotion lack empirical validation and scalable implementation. This study integrates the five-tone theory with artificial intelligence (AI) to develop an AI-driven music modulation system based on real-time physiological feedback, aiming to examine its efficacy in alleviating anxiety symptoms and modulating the autonomic nervous system. A single-group pretest-posttest design was employed, with 40 university students exhibiting moderate anxiety (M_age = 21.5) enrolled as participants. Each underwent a 20-min intervention session. The system employed the Jue tone as the foundational pitch, with real-time dynamic modulation guided by heart rate variability (HRV) and electrodermal activity (EDA) data. The Hamilton Anxiety Rating Scale (HAMA), high-frequency HRV (HF-HRV), and skin conductance level (SCL) were employed in pre-post measurements. Post-intervention statistical analyses revealed a statistically significant reduction in HAMA scores (M_pre = 18.2 vs. M_post = 11.5, p < 0.001), accompanied by significant increases in HF-HRV and significant decreases in SCL (both p < 0.001). The preliminary findings suggest that the AI-powered music intervention integrating the five-tone theory with biofeedback mechanisms may have significant anxiolytic effects. This highlights the potential for incorporating traditional cultural wisdom into modern digital therapeutics and warrants further investigation through controlled trials.