HEAL: Assessing the Therapeutic Impact of Personalization, Closed-Loop Adaptation, and Generative AI in Music Therapy
编号:76
访问权限:仅限参会人
更新:2026-11-10 16:38:51 浏览:11次
Online
摘要
Over one billion people globally live with mental health conditions, yet 85% receive no formal treatment mostly due to high cost, low accessibility or social stigma. Music therapy delivered traditionally can be expensive and difficult to scale. Most related digital solutions rely on static playlists and self-reported emotions which limits their effectiveness. This paper presents HEAL, a randomized controlled study of an AI-enabled adaptive music and visual therapy platform, evaluating three system design variables- content personalization, closed-loop feedback using emotional and physiological signals and AI-generated versus human-composed content. Three independent between-subjects experiments (n= 54 per experiment, n=27 per group) measured outcomes via validated psychological instruments (GAD-7, PSS-10, VAS), objective physiological sensors (Arduino-based HR and GSR) and engagement metrics. Results demonstrated a significant decrease in reported anxiety and self-reported levels of stress in comparison to control conditions. Moderate to large effect size differences were noted across the experiments, with Cohen's d ranging from 0.63 to 1.28, several exceeding one standard deviation. Chi-squared analysis indicated statistically significant increases in participant re-engagement in personalized and adaptive conditions compared to non-personalized and static conditions, with values ranging from 6.48 to 7.81 and p values below 0.02. Additionally, results indicated that AI-generated content yielded similar if not greater reduction in symptoms compared to human-composed content, with a change in GAD-7 of -5.78, change in PSS-10 of -8.19, change in HR of -9.04, and p values below 0.001. Cross-experiment aggregation revealed a consistent pattern that personalized, adaptive and AI-driven interventions outperform controls across psychological, physiological and engagement domains. These findings establish AI-driven adaptive music therapy as a scalable, evidence-based alternative to traditional music therapy for global mental health access.
关键词
adaptive feedback,generative AI,affective computing,Personalization,AI,AI driven,ISO principle
稿件作者
Anusha Arora
Interlake High School USA
发表评论