1. Introduction
Primary insomnia (PI) is a chronic sleep disorder characterized by difficulties in initiating or maintaining sleep, independent of comorbid medical or psychiatric conditions.[1] Epidemiological studies suggest a global prevalence of 10% to 15% in adults, with a higher incidence among females.[2] Cognitive behavioral therapy for insomnia [3,4] and novel pharmacological agents are recommended as first-line therapeutic options, but challenges such as limited long-term efficacy, drug dependency, and interindividual variability remain unresolved.[5–10] Current research emphasizes neural circuitry mechanisms, digital therapeutics, and biomarker discovery but lacks interdisciplinary synthesis, necessitating a systematic exploration to guide future investigations.
Bibliometrics, also known as literature metrics, is a quantitative method used to assess the impact of scholarly articles through citation and content analysis. It measures the influence of publications by counting citations and evaluating the reach and impact of research.[11–13] Current research status: recent advancements in bibliometrics have expanded beyond traditional citation counts to include article-level metrics and alternative metrics. Article-level metrics provide a more granular view of an article’s impact by tracking downloads, views, and social media mentions.
Alternative metrics, in particular, have gained popularity for their ability to measure the societal and digital influence of research. Furthermore, studies are increasingly focusing on developing standardized evaluation frameworks to better assess the quality and impact of scientific literature. Bibliometrics, a quantitative analysis of literature distribution, linkage, and evolution, provides critical insights into the structure of disciplinary knowledge and emerging trends.[14] Its application in medical research has grown significantly, particularly in predicting therapeutic trends and mapping collaborative networks.[13,15] However, bibliometric studies on PI remain scarce, with existing analyses often limited to single interventions (e.g., cognitive behavioral therapy [CBT] or pharmacotherapy) and lacking dynamic assessments of the overall knowledge framework.[16]
CiteSpace, a Java-based scientific mapping tool, excels at identifying research frontiers and paradigm shifts through co-citation analysis, keyword burst detection, and cluster visualization (Chen, 2016).[17] Compared to alternatives like VOSviewer (Visualizing Scientific Landscapes), its strength lies in dynamic network analysis, which enables the identification of pivotal studies (e.g., landmark studies) and interdisciplinary interfaces (e.g., neuroimmune crosstalk). This study utilizes CiteSpace 6.2.R4 to analyze 2826 articles from the Web of Science Core Collection (WOSCC, 2005–2025), aiming to address gaps in the systematic knowledge mapping of PI therapeutics and inform precision medicine strategies. As a specialized analytical platform built on Java architecture, CiteSpace enables sophisticated scientometric evaluation through dynamic knowledge mapping and temporal pattern identification within academic corpora, serving as an open-access solution for bibliometric exploration.
Created by Dr Chaomei Chen, it allows users to explore the development of knowledge domains through interactive visualizations. Key features include co-citation analysis, keyword clustering, burst detection, and network visualization, which help identify influential works, emerging trends, and intellectual turning points. CiteSpace supports data import from major databases such as Web of Science, Scopus, PubMed, and CNKI, making it versatile for bibliometric studies across disciplines.
It is widely used for literature reviews, mapping scientific fields, and understanding the evolution of research topics.
2. Materials and methods
2.1. Data extraction
Publications were retrieved from the WOSCC using a defined search strategy. The search covered publications from January 1, 2005, to March 7, 2025, the date of data retrieval. The search terms were set as TS=(“primary insomnia” OR “PI”) AND TS=(“therapeutic method” OR treatment OR therapy) AND language=(English).
The core term “primary insomnia” corresponds to the diagnostic category defined in the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition, Text Revision and the International Classification of Sleep Disorders, Second Edition. The Diagnostic and Statistical Manual of Mental Disorders, 5th edition (DSM-5, 2013) and the International Classification of Sleep Disorders, Third Edition (ICSD-3, 2014) later merged primary and comorbid insomnia into a single category, insomnia disorder.[8] A total of 2826 publications were retrieved and exported as text files, including “Full record with cited references.” The records were designated for export, with the comprehensive search process outlined in Figure 1.
2.2. Analysis method
We used CiteSpace v.6.2.R3 (64-bit) Beta Advanced for the bibliometric analysis. Initially, we performed heuristic and manual checks, which involved consolidating synonymous terms and merging countries (e.g., PEOPLES R CHINA → China; publications affiliated with Taiwan, China, were included in the China count). Institution names were kept as recorded in the Web of Science Organization-Enhanced field.
The data analysis revealed trends in annual publication counts, as well as identifying the most productive journals, authors, institutions, and countries. Co-occurrence analyses were also conducted for keywords, authors, references, and institutions. Furthermore, visualization analysis of trends in PI treatment was conducted.
In CiteSpace, the parameters were defined as follows: temporal slicing from 2005 to 2025; analytical parameters included temporal segmentation (annual intervals), comprehensive source term inclusion, sequential node selection protocol, and the pathfinder network pruning algorithm. The constructed network topology consisted of vertices representing distinct academic entities (author affiliations, institutional collaborations, international partnerships) interconnected through weighted inter-nodal connections. In the visualization, the period from 2005 to 2025 was represented by distinct colors differentiating the inner and outer sections.
Lines connecting nodes represented co-occurrence relationships, based on co-citation analysis. Purple circles represented centrality, with nodes exhibiting high centrality regarded as more significant.[18]
3. Results
3.1. Annual publication trends
A total of 2826 articles were retrieved from the WOSCC. The document types included in the analysis were articles, early access, proceedings papers, letters, review articles, meeting abstracts, and editorial materials. The number of articles published annually is shown in Figure 2.
The annual publication output on PI treatment from 2005 to 2024 exhibits a dynamic trajectory marked by nonlinear growth and critical inflection points. Initial exploration during the nascent phase (2005–2007) saw publications rise from 46 to 76 (65%), reflecting foundational research efforts. Subsequent fluctuations (2008–2016), including a minor decline in 2008 (76→75), potentially linked to global financial constraints,[19] transitioned into a consolidation period with a 6.7% compound annual growth rate, consistent with established technology adoption frameworks.[20] Accelerated expansion (2017–2022) culminated in a 55% surge (149 to 231 publications), driven by global investments in artificial intelligence/machine learning technologies such as the European Union’s Horizon 2020 and China’s Next-Generation AI Development Plan.[21]
3.2. Analysis of journals and cited journals
A total of 2826 publications on PI were identified across 774 academic journals. Table 1 lists the 10 most frequently co-cited journals. Journal quartiles were taken from the 2023 Journal Citation Reports, which rank journals by impact factor (IF) within each subject category and divide them into 4 equal groups: Q1 covers the top 25% of a category and Q2 the next 25%.[12] Nine of the 10 journals were ranked Q1, and the Journal of Sleep Research was ranked Q2.
Sleep (IF = 5.3) ranked first with 1801 co-citations, followed by Sleep Medicine (1487) and Sleep Medicine Reviews (1327). While The Lancet (IF = 98.4) ranks among the top 10 leading journals, its thematic relevance to sleep medicine requires critical evaluation. The Lancet and JAMA, together with The New England Journal of Medicine and The BMJ, are regarded as the 4 leading general medical journals.
The high co-citation counts of The Lancet (790) and JAMA (1021) underscore the centrality of clinical medicine journals in this field, consistent with recent meta-analyses on translational research trends.[10] In the co-cited journal network (Fig. 3), Sleep, Sleep Medicine, Sleep Medicine Reviews, Journal of Clinical Sleep Medicine, and Journal of Clinical Psychiatry form a densely connected core closely linked to JAMA and The American Journal of Psychiatry. Our analysis demonstrates frequent cross-citation among related journals, with medical research, clinical practice, and biological sciences forming a synergistic triad that systematically informs drug discovery and development.
3.3. Distribution of countries and institutions
A geographic distribution map of countries was generated using CiteSpace. The network consisted of 82 nodes and 125 links (Fig. 4).
The data show that 2826 publications originated from 594 countries or regions. The top 10 countries with the most publications are the United States, China, England, Canada, Australia, Germany, Japan, Sweden, France, and Italy (Table S1, Supplemental Digital Content 1). Regarding the total number of publications, the USA has published 1219 papers on PI research, ranking first in this domain, followed by the People’s Republic of China (China) with 491, England (286), Canada (221), Australia (203), Germany (192), Japan (121), Sweden (121), France (119), and Italy (118).
The United States demonstrates predominant scholarly productivity in PI research, evidenced by substantial publication volume and exceptional citation influence, underscoring its pivotal role in advancing disciplinary knowledge. While China ranks second in research output quantity, its current prominence within international collaboration networks remains comparatively limited, reflecting opportunities for enhanced academic integration. Figure S1, Supplemental Digital Content 2, shows that the listed countries are distributed across 4 major regions: North America (USA, Canada), Asia (China, Japan), Europe (England, Germany, Sweden, France, Italy), and Oceania (Australia).
This geographic distribution highlights their representation across diverse continents, reflecting a broad global scope. Figure 5 shows the distribution of research institutions, with 365 nodes and 632 links. The top 10 institutions in terms of the number of publications are the University of California System, Harvard University, Harvard University Medical Affiliates, the US Department of Veterans Affairs, Veterans Health Administration, Duke University, Harvard Medical School, University of London, Pennsylvania Commonwealth System of Higher Education, and the University of Pennsylvania (Table 2).
These names follow the Web of Science Organization-Enhanced field, which aggregates affiliated units at the system level. The University of California System combines 10 independent campuses, and several entries overlap as parent and subsidiary organizations (Harvard University, Harvard Medical School, and Harvard University Medical Affiliates; the US Department of Veterans Affairs and the Veterans Health Administration; Pennsylvania Commonwealth System of Higher Education and the University of Pennsylvania). A single article can be counted under more than one of these entries.
The first-place ranking of the University of California System reflects the combined output of its campuses and does not indicate that any single campus is the most productive institution. Harvard Medical School produced 88 scholarly works (7th in productivity) but achieved the highest centrality metric (0.29) in scientometric analysis, demonstrating its strategic influence despite moderate output volume. Network topology analysis revealed fragmented institutional collaboration, particularly among high-output research entities with significant citation impacts.
Given the expanding focus on emerging research domains, these findings underscore the necessity to strengthen multinational consortia and interdisciplinary partnerships to address current knowledge fragmentation. In the co-authorship network (Fig. 6), Thomas Roth had the largest node, indicating the highest publication output, followed by Andrew D.
Krystal, Colin A. Espie, Gerhard Andersson, Charles M. Morin, and Jack D.
Edinger. The co-cited author network contained 1041 nodes and 8578 links (Fig. S2, Supplemental Digital Content 3).
Charles M. Morin was the most frequently co-cited author (924 co-citations), followed by Daniel J. Buysse (776), Célyne H.
Bastien (561), Maurice M. Ohayon (466), and Dieter Riemann (442; Table 3).
3.4. Keyword and thematic evolution
Scholarly consensus suggests that emerging research frontiers can be identified through a systematic examination of keyword prevalence and network centrality metrics.[22] Using co-occurrence analysis techniques, researchers constructed a semantic network visualization consisting of 349 distinct nodes interconnected by 451 conceptual relationships. Figure 7 displays 10 distinct research themes, each differentiated by color coding and cluster numbering. Each cluster represents a research hotspot or trend within the field.
For instance, cluster #0 represents “sleep quality,” cluster #1 represents “major depressive disorder,” cluster #2 represents “obstructive sleep apnea,” cluster #3 represents “chronic pain,” cluster #4 represents “safety,” cluster #5 represents “older adults,” cluster #6 represents “mental health,” cluster #7 represents “orexin,” cluster #8 represents “systematic review,” and cluster #9 represents “sleep disorders.” These clusters not only reveal the diversity of research but also reflect the interconnections between different treatment methods and disease management strategies. The clustering showed a well-defined structure (modularity Q = 0.7989; weighted mean silhouette S = 0.9238). Based on frequency and centrality, the most popular keywords were randomized controlled trials, CBT, quality of life, primary insomnia, sleep disorders, primary care, systematic review, sleep quality, sleep disturbance, and chronic insomnia (Table S2, Supplemental Digital Content 4).
The frequency of co-occurrence reflects how often keywords appear together, while centrality measures the extent to which keywords attract attention. Nodes with a centrality value exceeding 0.1 are considered influential in the study. Among the keywords analyzed, “randomized controlled trial” exhibited the highest co-occurrence frequency (135), followed by “cognitive behavioral therapy” (108), and “quality of life” (87), which were the 3 keywords with the greatest co-occurrence.
The top 3 keywords in terms of centrality were primary insomnia (0.13), sleep disorders (0.12), and randomized controlled trials (0.11). As the gold standard methodology in clinical research, randomized controlled trials demonstrate unparalleled rigor in assessing therapeutic interventions for PI, systematically generating high-quality empirical evidence to validate treatment protocols.[23–25]
Figure 8 presents the top 25 keywords with citation bursts. Bursty keywords represent frequent temporal citations[26]; burst detection identifies the period in which the frequency of a keyword rises sharply.[27] The keyword “double-blind” had the most substantial citation bursts in 2014. The early phase (2005–2010) was characterized by pharmacological validation, as evidenced by high-strength terms such as “double-blind” (strength = 23) and “zolpidem” (strength = 17.02), reflecting a focus on empirical validation of therapeutic efficacy.
The intermediate phase (2011–2018) witnessed a paradigm shift toward mechanistic investigations and longitudinal interventions, exemplified by the emergence of “chronic primary insomnia” (strength = 9.31), indicating methodological diversification. In the recent phase (2019–2024), research priorities transitioned to evidence-based standardization, with terms like “clinical practice guideline” (strength = 9.32) highlighting the establishment of translational frameworks. Methodologically, peak burst strength (>15) was concentrated in early experimental terminology, while later phases exhibited stabilized intensity (mean = 7.2 ± 1.8), signaling disciplinary consolidation.
Interdisciplinary convergence was prominently demonstrated through the sustained co-occurrence of “psychiatric disorders” and “circadian rhythms” (2005–2013), underscoring the integration of neurobiological mechanisms into clinical applications. The field’s progression aligns with Price’s theory of scientific growth, transitioning from empirical validation to paradigm-driven standardization, demonstrating enhanced disciplinary maturity and cross-disciplinary synergy.
3.5. Analysis of cited references
Table 4 lists the top 10 publications based on co-citation frequency and centrality. By analyzing co-citation frequency, bibliometric analysis reveals that standardized diagnostic frameworks and evidence-based treatment guidelines dominate current research in sleep medicine and mental health. The DSM-5 (2022; co-citation = 138, centrality = 0.19), serves as a pivotal reference, with its high citation frequency underscoring the critical role of nosological standardization in interdisciplinary research.
Three major clinical guidelines[28–30] were collectively cited 306 times, reflecting the dual emphasis on pharmacological interventions (e.g., zolpidem) and non-pharmacological approaches like CBT. The growing prominence of technology-driven solutions is evidenced by van Straten’s meta-analysis on CBT efficacy (2021; citations = 67) and Espie’s randomized trial on digital therapeutics (2019; citations = 59). Notably, Carter’s critical discourse on DSM limitations (2014; centrality = 0.12) and Mittal’s exploration of psychomotor comorbidities (2011; citations = 72) highlight ongoing debates about diagnostic boundaries.
Future research should focus on 3 frontiers: integrating multimodal biomarkers (e.g., circadian gene expression) to refine insomnia subtyping; developing machine learning-powered personalized protocols to address the lack of long-term efficacy data (>5 years) in current guidelines; elucidating neuroimmune mechanisms linking antidepressants (e.g., tricyclics) to sleep architecture alterations, which is crucial for managing rising comorbidity rates. Emerging trends suggest a paradigm shift toward precision sleep medicine, necessitating large-scale longitudinal cohorts and open-access biomarker databases to bridge translational gaps.
Co-citation cluster analysis reveals a dynamic evolution of knowledge structures and thematic diversification in PI research (Table 5). The largest cluster (#0, size = 233, silhouette = 0.791), centered on “systematic reviews” and “network meta-analyses,” emphasizes the comparative efficacy of CBT for cancer survivors, reflecting the dominance of evidence-based paradigms.[31] Cluster #1 (size = 182, silhouette = 0.811) further validates the expansion of CBT into digital therapeutics,[32] highlighting technology-driven intervention innovations.[33] Early research, represented by cluster #4 (size = 104, silhouette = 0.936), focused on behavioral treatments and clinical practices for chronic insomnia, while cluster #3 (size = 110, silhouette = 0.912) marks the emergence of melatonergic drugs (e.g., prolonged-release melatonin) around 2005, signaling initial pharmacological explorations (Riemann et al., 2017). Notably, cluster #8 (size = 54, silhouette = 0.959) highlights breakthroughs in neuromodulatory therapies, particularly orexin receptor antagonists like suvorexant.[9] In addition, clusters #5 (size = 81, silhouette = 0.946) and #10 (size = 28, silhouette = 0.978) address comorbid mechanisms in cancer-related insomnia and obstructive sleep apnea, respectively, suggesting the potential for cross-disease research.
Collectively, the field is transitioning from single-behavioral interventions to multimodal integration (e.g., digital therapeutics, neuromodulation), but long-term efficacy data and biomarker validation remain critical challenges for future research. Co-citation burst analysis reveals temporal shifts in sleep medicine research priorities (Fig. 9).
Between 2006 and 2012, peak burst intensities focused on behavioral and pharmacological interventions for insomnia, exemplified by Morin CM’s seminal work (2006/2012; strength = 18.3–22.53) comparing CBT efficacy against benzodiazepines, alongside Roth T’s pharmacological validation studies (2006; strength = 14.72). These efforts established foundational evidence for dual therapeutic pathways. The 2014 to 2019 period witnessed paradigm standardization through clinical guidelines, with Qaseem A’s 2016 guideline synthesis (strength = 43.38) and Buysse DJ’s nosological refinements (2013; strength = 35.01) consolidating evidence-based frameworks.
Zachariae R (2016; strength = 20.51) explored psychophysiological insomnia subtypes, reflecting a shift toward mechanistic classification. After 2017, research shifted toward neurobiological mechanisms and digital innovation. Riemann D (2017; strength = 40.65) advanced circadian dysregulation models, while van Straten A (2018; strength = 19.21) conducted a meta-analysis on CBT’s long-term durability, bridging gaps between neurobiology and behavioral science.
Espie CA’s digital therapeutics trials (2019; strength = 17.9) and Yin X’s neuroimmune crosstalk investigations (2017; strength = 14.18) signaled emerging themes like “digital biomarkers” and “comorbidity-driven insomnia.” Post-2020, research frontiers emphasize precision medicine and multimodal integration. De Crescenzo F’s multi-omics profiling (2022; strength = 28.42) and Page MJ’s wearable-enabled dynamic phenotyping (2021; strength = 27.51) underscore efforts to decode insomnia heterogeneity. Edinger JD (2021; strength = 15.65) redefined obstructive sleep disorder entities, including chronic insomnia (Morin et al., 2006–2012) and obstructive sleep apnea management, while Trauer JM (2015; strength = 26.41) highlighted cancer-related insomnia as a critical comorbidity.
Breakthrough keywords such as “orexin antagonists” and “real-time actigraphy” signal a transition toward multimodal data integration and personalized intervention. This evolution underscores the field’s progression from symptom management to mechanism-driven precision frameworks, although challenges persist in operationalizing translational biomarkers across heterogeneous cohorts.
4. Discussion
This study analyzed 2826 articles on PI treatment published between January 1, 2005, and March 7, 2025. Since 2017, there has been a significant increase in the annual number of publications, surpassing 149 articles per year. Notably, in 2024, the annual publication count exceeded 285 articles.
This upward trend suggests that research on PI treatment is experiencing rapid growth and attracting increasing attention within the academic community. In the field of PI therapeutic research, the United States and China have emerged as dominant contributors, with the United States holding the leading position. Analysis of the top 10 research institutions revealed that 9 are based in the United States, with the remaining institution located in the United Kingdom.
Notably, the United States demonstrated predominant influence in international collaborations, as evidenced by its largest node size in the cooperation network. This metric substantiates its clear leadership in advancing global research on PI therapeutics.
Among co-cited journals, Sleep ranked first (1801 co-citations; IF = 5.3; Q1), followed by Sleep Medicine (1487) and Sleep Medicine Reviews (1327). The highest IFs among the 10 leading co-cited journals belonged to The Lancet (IF = 98.4) and JAMA (IF = 63.5). Charles M.
Morin was the most frequently co-cited author (924 co-citations), and Thomas Roth showed the highest publication output. Morin’s continued influence is reflected in 2 2023 publications, “Effect of Psychological and Medication Therapies for Insomnia on Daytime Functions: A Randomized Clinical Trial”[34] and “Insomnia Research: What Have We Learned and What Else Do We Need to Know?”[35] Further investigation is required to systematically characterize distinct insomnia phenotypes and empirically validate corresponding therapeutic interventions, advancing precision medicine approaches in sleep disorder management.
Co-citation clusters represent scholarly works jointly referenced across multiple publications, serving as conceptual cornerstones that delineate disciplinary boundaries in specialized research fields.[36,37] To construct a methodological framework for chronic insomnia intervention analysis, we curated the top 10 most frequently referenced publications based on citation density metrics. The seminal work by Morin CM and Buysse DJ (2005), published in Sleep, emerged as the predominant reference within this analytical cohort. Similarly, the validation study of the Insomnia Severity Index, published in Sleep Medicine, ranked among the most frequently co-cited references.[38] Citation bursts in referenced works signify cutting-edge research frontiers, reflecting acute scholarly attention within specialized domains.
Notably, the DSM-5, emerged as the most highly co-cited reference (n = 138) with a centrality score of 0.19, underscoring its foundational role in shaping modern insomnia research. ICSD-3, issued by the American Academy of Sleep Medicine, is the principal diagnostic manual in sleep research, and its criteria for chronic insomnia disorder closely parallel those of DSM-5.[39,40] The widespread adoption of these harmonized diagnostic criteria facilitated the standardization of insomnia definitions across studies, thereby enhancing cross-study comparability, improving patient stratification, and enabling more rigorous evaluation of treatment efficacy. This nosological consistency has served as a critical scaffold for downstream investigations into comorbid mechanisms, phenotype classification, and targeted therapeutic approaches – marking a significant medical-scientific contribution to the field.
Scientometric analysis of high-impact publications (Fig. 9) reveals behavioral interventions as the predominant insomnia therapeutic paradigm, with cognitive-behavioral protocols demonstrating particular efficacy in sleep architecture modulation. Melatonin drugs and orexin receptor antagonists underscore the prominence of neuromodulatory therapies as the mainstream approach in the current treatment of PI.[41–46] Notably, the persistent high frequency of keywords such as “cognitive behavioral therapy” (n = 108) and its prominence in co-citation clusters confirms that cognitive behavioral therapy for insomnia remains the cornerstone of evidence-based practice, while emerging technologies represent complementary or alternative approaches for specific patient populations.
Future priorities include elucidating melatonin receptor subtype specificity, optimizing behavioral-pharmacological synergies, and constructing dynamic comorbidity network models to advance precision sleep medicine paradigms. Emerging translational challenges demand large-scale biomarker validation and real-world adherence analytics to bridge efficacy-effectiveness gaps, establishing methodological guidelines for subsequent scholarly investigations and providing direction for future research in this field.
This study highlights 3 key methodological strengths in PI research analysis. First, to our knowledge, this is the first bibliometric analysis to map 2 decades of research on PI treatment, and the systematic application of bibliometric methods provides a comprehensive mapping of global research trends, offering scholars an evidence-based framework for understanding therapeutic developments. Second, exclusive reliance on CiteSpace for quantitative visualization enhances objectivity through standardized metric evaluation.
Third, bibliometric analysis goes beyond conventional narrative reviews by algorithmically identifying emerging research frontiers through co-occurrence network topology. However, several limitations must be acknowledged. First, the exclusive reliance on the WOSCC as the sole data source may introduce selection bias, potentially omitting important regional studies or clinical trials indexed in other major databases such as PubMed.
Second, the analysis was restricted to English-language publications, which may overlook valuable contributions from non-English-speaking regions. Third, although the search period was defined as 2005 to 2025, it should be noted that data for 2025 only cover studies published up to March 7th. As shown in Figure 2, the number of publications in 2025 (n = 41) is significantly lower than in 2024 (n = 285), and thus may misrepresent recent trends if interpreted as a full-year output.
Fourth, the search was built on the term “primary insomnia.” Studies published after the release of DSM-5 and ICSD-3 that describe their population only as “insomnia disorder” were not retrieved. Publication volume after 2013 is therefore likely underestimated, and recent trends mainly reflect studies that retained the PI construct. The burst of the keyword “insomnia disorder” in 2021 to 2024 (Fig.
8) shows that the dataset captures part of this terminological transition. Combining both terms in future bibliometric analyses would give a fuller picture of insomnia treatment research under current diagnostic classifications.[39,40] Fifth, institution names followed the system-level aggregation of the Web of Science, which affects institutional rankings as described in section 3.3. This analysis reflects academic publishing trends within WOSCC and should not be interpreted as a direct assessment of clinical efficacy or treatment priorities in real-world practice.
5. Conclusion
In summary, this study presents, for the first time, a comprehensive developmental trajectory and academic collaboration network in PI research through systematic bibliometric analysis. Using CiteSpace visualization techniques (version 6.2.R3) to track the dynamic evolution of research hotspots, the study revealed 3 key insights: international collaborations exhibited a “core-periphery” distribution pattern, with institutions such as Harvard University and the University of Pennsylvania serving as central hubs. Research themes transitioned through 3 distinct phases: “pathological mechanism exploration,” “pharmacotherapy optimization,” and “non-pharmacological intervention innovation.” Emerging clusters highlighted digital CBT and neuromodulation technologies as current research frontiers.
These results not only propose a structural framework for fostering interdisciplinary collaborations but also identify unresolved scientific challenges through knowledge mapping analysis, thereby guiding future theoretical advancements and methodological innovations in this field.
Acknowledgments
The authors would like to thank all participants in this study.
Author contributions
Conceptualization: Junchang Liu, Xiaojuan Li, Xingping Zhang.
Data curation: Xusheng Lu.
Formal analysis: Junchang Liu.
Investigation: Junchang Liu, Xiaojuan Li.
Methodology: Xingping Zhang.
Software: Junchang Liu, Xiaojuan Li, Gulaisaer Aikebaier, Xusheng Lu, Xinxin Gu.
Validation: Junchang Liu, Xiaojuan Li.
Visualization: Xingping Zhang.
Writing – original draft: Junchang Liu, Xiaojuan Li.
Writing – review & editing: Xingping Zhang.