Development and Evaluation of Individualized Music Therapy for Common Mental Disorders: Protocol for a Multistage Study

Abstract Background Pharmacotherapy for common mental disorders is frequently limited by adverse events and suboptimal adherence. While music therapy offers a promising nonpharmacological alternative, its clinical utility is currently constrained by limited accessibility, inconsistent efficacy, and a lack of mechanistic clarity. Objective This study aims to describe the development of individualized (receptive) music therapy (IMT), an artificial intelligence (AI)–enabled, neuroscience-guided intervention, and to evaluate its efficacy, safety, and underlying neurobiological mechanisms in adults with major depressive disorder, generalized anxiety disorder, and primary insomnia. Methods This multistage research program is being conducted at Peking Union Medical College Hospital and comprises four sequential studies: (1) a cross-sectional pilot study (n=20) to benchmark clinical and electroencephalography features, (2) a prospective cohort study (n=80) evaluating the efficacy and safety of nonindividualized receptive music therapy, (3) a pilot randomized clinical trial (n=300) comparing nonindividualized therapy with IMT over 8 weeks, and (4) a prospective validation study (n=60) of a treatment-response prediction model. Participants include adults aged 18 to 60 years with mild-to-moderate major depressive disorder, generalized anxiety disorder, or primary insomnia, along with healthy controls. In the nonindividualized arm, participants engage in daily 30-minute listening sessions using therapist-curated instrumental tracks designed to regulate mood. In the IMT arm, an AI generation pipeline creates bespoke instrumental tracks based on weekly participant preferences regarding tempo, instrumentation, and emotional valence. To equalize participant–researcher contact across arms and reduce attention and expectation bias, participants in the nonindividualized arm complete a weekly music-experience questionnaire matched in length and timing to the IMT arm’s weekly preference assessment, and treatment credibility and outcome expectancy are measured at baseline in both arms. The primary outcomes are the response rate at 8 weeks (defined as ≥50% reduction in Montgomery-Åsberg Depression Rating Scale [MADRS], Hamilton Anxiety Rating Scale [HAMA], or Pittsburgh Sleep Quality Index [PSQI] scores) and changes in quantitative electroencephalography characteristics. Secondary outcomes include changes in MADRS, HAMA, and PSQI scores from baseline to 8 weeks. Results Ethical approval was obtained from the Ethics Review Committee of Peking Union Medical College Hospital (I-24PJ0689). Written informed consent will be obtained from all participants. The recruitment started on May 24, 2024, and the study is expected to be completed by December 2027. Results will be disseminated via peer-reviewed publications, conference presentations, and stakeholder communications. Authorship will follow the International Committee of Medical Journal Editors (ICMJE) criteria. The participant-level dataset will be available upon reasonable request. Conclusions This protocol outlines a translational framework designed to address the “therapeutic ceiling” of traditional music therapy. By integrating generative AI with neurophysiological monitoring, this program aims to develop a scalable, precision-medicine approach to mental health care that is both clinically effective and biologically grounded.