A Bibliometric Analysis of Music's Role in Promoting Well-Being in Health Science Research

2.1 Data collection

This study followed the PRISMA 2020 framework¹⁴ to ensure transparent data collection and screening processes ( Figure 1). We retrieved bibliographic records from Scopus, selected for its robust coverage of both music-related research and health sciences literature. On 21 February 2025, we developed a comprehensive search strategy targeting titles, abstracts, and keywords related to various well-being constructs. Our search methodology employed an iterative approach: first identifying root terms (e.g., “music,” “wellbeing”) and their frequently co-occurring keywords through preliminary analysis, then manually screening these terms to establish our final keyword set. The search was limited to English-language journal articles with no publication date restrictions and was last updated on 22 May 2025 (Appendix A). From an initial pool of 22,358 records, we included 16,585 publications for bibliometric analysis and selected 30 key articles for systematic review.

Figure 1. PRISMA flow diagram and methods (Adapted from Dzhunushalieva & Teuber, 2024 and Page et al., 2021).

^† Number of literature reviewed (n_review) was determined based on the balance of cosine similarities, piecewise regression by era, citation metrics, and thematic alignment of keywords.

2.2 Data cleaning

We implemented a structured data cleaning pipeline using Python 3.13.3 to standardize terminology and improve keyword consistency. Our processing workflow incorporated pandas¹⁵ for data manipulation and regular expressions for text normalization, supplemented by NLTK¹⁶ for lemmatization. To address terminological variations, we developed a custom synonym dictionary that consolidated equivalent terms (e.g., mapping “covid” and “coronavirus” to “covid-19”). The research team manually reviewed and corrected lemmatization artifacts (e.g., converting “social medium” to “social media”). Final keyword selection prioritized the most frequently occurring representative term from each conceptual group.

2.3 Analysis and virtualization

We conducted descriptive analyses using R version 4.4.3 with the bibliometrix package.¹⁷ The dataset was transformed into a bibliographic dataframe to examine key metrics including publication volume, document age, citation rates, and author productivity across tourism, marketing, and business management domains. Temporal publication trends were analyzed using Piecewise Linear Regression to identify inflection points in research output. For comparative visualization, we employed Python to generate figures illustrating both national-level productivity and international collaboration networks, enabling cross-field comparisons.

The clustering analysis employed a two-phase methodology. First, we constructed a keyword co-occurrence network using VOSviewer 1.6.20,¹⁸ with nodes representing keywords and edges reflecting co-occurrence strength. We implemented Louvain clustering^19–21 rather than the default method to better identify modular communities. The algorithm was executed 100 times with varying random seeds (1-100), with manual selection of the optimal solution based on alignment with expected thematic structures. In the second phase, we mapped articles to clusters using cosine similarity between article keyword vectors and cluster profiles. This approach, which considers both keyword composition and directional relationships,^11,22 provided more accurate thematic alignment than frequency-based methods by capturing deeper semantic connections while avoiding overrepresentation of generic terms.

The cluster interpretation process combined quantitative bibliometric mapping with qualitative analysis through a structured three-phase approach. Following the scanning, sensing, and substantiating framework,²³ we first identified preliminary themes by examining high-frequency keywords and article titles. Next, we analyzed representative abstracts and author keywords to discern conceptual relationships. Finally, we triangulated these findings with citation metrics, temporal trends, and disciplinary discourse to validate themes. Cluster labels emerged inductively through synthesis of dominant keywords, metadata patterns, and content analysis of representative publications.

To enhance rigor, we incorporated an iterative reflection process using ChatGPT-4, which improved thematic clarity and visual coherence while maintaining human oversight. This approach aligns with: (1) contemporary bibliometric practices emphasizing quantitative-qualitative integration,^22,24; (2) emerging LLM-assisted methodologies for complex data analysis¹⁹; and (3) transdisciplinary bibliometric principles that facilitate cross-domain interpretation.²⁵

3.1 Descriptive

Our temporal analysis using Piecewise Linear Regression identified four distinct research eras demarcated by inflection points at 1982.7, 2002.5, and 2017.5. The earliest period (Era1: 1898–1982) showed gradual growth with 0.31 publications/year, followed by accelerated output during Era2 (1983–2002; 8.03 publications/year). The field entered rapid expansion in Era3 (2003–2017; 50.46 publications/year), culminating in exponential growth in Era4 (2018–2024; 174.86 publications/year), as evidenced by the model’s exceptional fit (R² = .9962). Publication volume surged from 361 documents in Era1 to 11,681 in Era4 ( Table 1, Figure 3), paralleled by expansions in journal diversity, authorship networks, and keyword variety. While citation impact peaked in Era2 (37.93 citations/doc) and Era3 (37.18), Era4 showed reduced but still active engagement (9.92 citations/doc; 1.911 citations/year/doc). The field demonstrated progressive globalization through rising co-authorship (mean authors/doc: 1.2→4.6) and international collaboration (12%→43% of publications), reflecting its maturation into an interdisciplinary research domain.

Figure 2. Global distribution of publications by corresponding author and collaboration type.

Note. Bar chart and world map show the number of publications per country based on corresponding author counts.

Figure 3. Music and well-being research clusters (left) and annual trend analysis with piecewise regression (right).

Beyond overall publication growth, our country-level analysis reveals significant variations in research output and collaborative patterns. The United States dominates with 4,094 publications, substantially outpacing the United Kingdom (1,618) and China (1,380). Other key contributors include Australia (1,111), Canada (791), and Germany (701). Notably, while the United States and India maintain high publication volumes, they demonstrate relatively limited international collaboration (13.6% and 12.1% respectively). This contrasts sharply with nations like South Korea (29.8% internationally co-authored publications), Italy (28.0%), and the Netherlands (25.7%), which show stronger global research integration. These disparities highlight fundamental differences in both productivity and scholarly connectivity among leading nations, as visually summarized in Figure 2. The findings suggest that research volume alone does not necessarily correlate with international engagement, revealing distinct national approaches to music and wellbeing research.

3.2 Cluster-based bibliometric analysis

Our Louvain clustering analysis of author keyword co-occurrence networks identified five distinct thematic research clusters ( Figure 3, left). Using cosine similarity, we mapped these cluster-level keyword profiles to individual articles, enabling longitudinal tracking of thematic evolution. Two clusters showed particularly rapid growth during 2010–2024: Cluster 1 (Well-being, Motivation, and Spiritual Growth in Musical Experience) and Cluster 2 (Music Therapy for Anxiety, Pain, and Emotional Distress). Cluster 3 (Music Perception: Emotion, Cognition, and Identity) demonstrated steady expansion, while Clusters 4 (Music-based Rehabilitation and Wellness in Older Adults) and 5 (Hearing Health and Risk Awareness in Youth) maintained consistent annual growth. These trends collectively highlight the field’s growing emphasis on mental health applications, aging populations, and auditory well-being ( Figure 3, right), reflecting broader societal health priorities in music research.

Cluster 1: Well-being, Motivation, and Spiritual Growth in Musical Experience

This cluster demonstrates music’s capacity to facilitate personal transformation through emotional enrichment, existential exploration, and identity development in both individual and communal contexts. The research reveals two key dimensions of musical experience: (1) an innate psychobiological foundation, evidenced by early work showing humans’ natural ability to synchronize emotional states with musical expressions,²⁶ and (2) culturally-mediated transformative effects. Neurochemical studies elucidate music’s ability to stimulate motivational and social bonding systems, enhancing psychological resilience.²⁷ In collective settings like raves, music facilitates symbolic healing through altered states of consciousness and emotional catharsis.²⁸ Active musical engagement—including singing, rapping, and movement—promotes self-actualization by fostering creativity, self-esteem, and identity consolidation.⁶ Awe-inducing musical experiences further contribute to spiritual growth and meaning-making,²⁹ while intercultural musical practices strengthen community belonging and cross-cultural understanding.³⁰ Despite its transformative potential, this research stream remains under-cited relative to clinical applications, suggesting opportunities for greater scholarly recognition.

Cluster 2: Relieving Anxiety, Pain, and Emotional Distress through Music Therapy

This cluster examines the evidence-based application of music therapy for alleviating psychological and physical distress in clinical environments. While music’s calming effects have been historically acknowledged, structured clinical research only emerged in the 2000s through systematic reviews and controlled trials. A seminal review established that audio-delivered music interventions in hospital settings consistently reduce anxiety and improve mood, though physiological outcomes show more variability.³¹ The evidence base has since expanded significantly, with clinical guidelines now incorporating music therapy for breast cancer care—both for general stress reduction³² and specifically during chemotherapy to enhance emotional well-being and treatment tolerance.³³ Comparable therapeutic benefits have been documented in perinatal care, where music interventions decrease maternal anxiety and labor pain while improving birth experiences.³⁴ Comprehensive meta-analyses confirm these effects, demonstrating music therapy’s robust capacity to mitigate stress, regulate physiological arousal, and promote emotional balance across diverse clinical populations,³⁵ with additional benefits for fatigue management and sleep quality.³⁶ This progression from anecdotal use to evidence-based practice reflects the field’s maturation into a standardized therapeutic modality.

Cluster 3: Emotion, Cognition, and Individual Identity in Music Perception

This cluster elucidates the complex interplay between music perception, emotional processing, and cognitive functioning. Foundational research established that listeners demonstrate remarkable consistency in interpreting musical meaning through structured descriptors, indicating that musical expressiveness derives from shared symbolic and rhythmic conventions rather than subjective interpretation alone.³⁷ Neuroscientific evidence reveals music’s capacity to activate primal emotional networks, triggering memory recall, mood modulation, and physiological responses (e.g., chills)—findings that support evolutionary theories of music as a communication and social bonding mechanism.³⁸ Theoretical models have since delineated six distinct pathways for musical emotion induction (brainstem reflexes, conditioning, emotional contagion, visual imagery, episodic memory, and cognitive expectancy), emphasizing the need for mechanism-specific investigations.³⁹ Neurochemical research has further demonstrated dopamine-mediated effects on musical pleasure and motivation, providing direct evidence of music’s reward system engagement.⁴⁰ Contemporary work bifurcates into two productive directions: (1) applied research on music-based emotional regulation through personalized interventions,⁴¹ and (2) fundamental studies of music’s cognitive effects, including neural plasticity and therapeutic applications across brain functions.⁴² Together, these advances underscore music’s dual role as both an emotional modulator and cognitive stimulant.

Cluster 4: Rehabilitation and Wellness in Older Adults through Music

This cluster examines the evidence-based application of music therapy for geriatric rehabilitation, with particular focus on Parkinson’s disease and dementia populations. While initial research recognized music’s general calming effects, systematic therapeutic applications only gained empirical support in the 2000s. For Parkinson’s patients, active music therapy combining singing and rhythmic movement demonstrates significant improvements in motor symptoms (e.g., bradykinesia), emotional well-being, and functional capacity.⁴³ Complementary approaches using dance and rhythmic auditory stimulation enhance gait rehabilitation while simultaneously addressing cognitive-motor integration.⁴⁴ Emerging digital modalities show promise, as evidenced by a pilot study where virtual group music therapy reduced apathy and depression with strong participant adherence.⁴⁵ In dementia care, meta-analyses confirm music therapy’s efficacy in reducing behavioral symptoms (agitation, anxiety) and improving cognitive metrics and quality of life,^46,47 often through multimodal interventions incorporating movement. Recent technological advances have expanded therapeutic access through ICT-based programs that manage neuropsychiatric symptoms via interactive digital platforms.⁴⁸ These developments reflect music therapy’s evolution from adjunctive comfort measure to standardized rehabilitation component in geriatric care.

Cluster 5: Hearing Health and Risk Awareness in Young Music Listeners

This cluster investigates hearing health risks associated with recreational music exposure among youth populations. Pioneering experimental studies first established the physiological vulnerability to high-decibel stimuli and fundamental mechanisms of auditory conditioning.⁴⁹ Clinical research subsequently identified subclinical auditory pathologies—including hyperacusis and tinnitus—in young music listeners without measurable hearing loss.⁵⁰ Public health studies have since shifted focus to behavioral prevention, with systematic reviews identifying personal audio devices and live music venues as primary risk factors for noise-induced hearing loss (NIHL) and advocating for targeted education initiatives,^51,52 Emerging evidence reveals early-stage auditory damage in young musicians through advanced diagnostics, suggesting conventional audiometry may underestimate music-related hearing impairment.⁵³ While current rehabilitation options remain limited, novel interventions including gene therapy and precision medicine show promise for addressing irreversible auditory damage,⁵⁴ highlighting the need for both preventive strategies and advanced treatment development.

Following the scanning, sensing, and substantiating framework,²³ Table 2 presents seminal articles representative of each thematic cluster, identified through cosine similarity-based keyword alignment. The selected publications are organized chronologically by research era, showcasing influential works that exemplify each cluster’s conceptual focus. Citation metrics—including total citations (TC) and citations per year (CPY)—provide quantitative indicators of their sustained scholarly impact and temporal influence within the field.