Tracking the Evolution of Research Topics in Healthcare Informatics Research Using Keywords and MeSH Terms
Article information
Abstract
Objectives
This study analyzed publications in Healthcare Informatics Research (HIR) to identify trends and shifts in research focus within both the journal and the broader Korean medical informatics landscape. By examining keywords across these papers, the study aimed to elucidate evolving priorities and innovations in the field over time.
Methods
Data from 958 papers published between 1995 and 2024 were extracted from the HIR journal’s online archive. The analysis focused on English-language articles published since 2010 (n = 658) to examine publication trends using descriptive statistics. Keyword and Medical Subject Headings (MeSH) term analyses (term frequency-inverse document frequency, latent Dirichlet allocation, co-occurrence) were performed on a subset of articles with available abstracts (n = 632) to identify research themes and interrelationships. Inferential statistics, including chi-square and regression analysis, were applied to assess changes in research trends over time.
Results
Among 958 total papers identified (672 in English), analysis of 658 English articles published since 2010 revealed increasing publication trends, peaking between 2015 and 2018. Keyword and MeSH term analyses of 632 papers with abstracts highlighted persistent themes (e.g., health systems, electronic health records) alongside emerging topics (e.g., machine learning, telemedicine). Inferential analysis indicated no statistically significant changes in keyword distribution over time.
Conclusions
This study offers insights into the evolution of health informatics research in Korea, underscoring the role of HIR in documenting this progression. The findings reveal a balance between emerging technologies and foundational healthcare themes, demonstrating the field’s adaptability and sustained relevance. Future research should extend the analysis to other journals and further consider ethical implications and global developments.
I. Introduction
Medical informatics has evolved rapidly over recent decades, transforming from a niche discipline into a cornerstone of modern healthcare systems worldwide [1]. By integrating information technology with medical practice, this field has revolutionized patient care, healthcare operations, and data management [2]. The exponential growth of digital health technologies, artificial intelligence (AI) applications, and big data analytics in healthcare has spurred a surge in research output and an increasing number of specialized journals. According to a bibliometric analysis by DeShazo et al. [3], the number of medical informatics articles grew by an average of 12% annually between 1987 and 2006, with 4,655 unique journals publishing medical informatics content during that period. This growth reflects the expanding scope and importance of the field. More recent analysis by Liang et al. [4] corroborates this trend, reporting that annual global publications in medical informatics increased by 193.86%, from 1,987 in 2011 to 5,839 in 2020.
Within this context of rapid growth, the Healthcare Informatics Research (HIR) journal has played a pivotal role in South Korea, publishing seminal works that have shaped the national landscape of medical informatics for more than 15 years. As highlighted in the recent editorial “Milestones and Growth: The 30-Year Journey of Healthcare Informatics Research” by Chang [5], HIR evolved from its origins as the Journal of the Korean Society of Medical Informatics (JKOSMI) in 1995 into a leading international journal in the field. Both JKOSMI and HIR have consistently contributed to the dissemination of knowledge within the health informatics domain, and this trajectory closely mirrors the development of medical informatics in South Korea. It reflects the nation’s growing engagement with, and contributions to, this rapidly advancing discipline.
This study analyzed publications in HIR to identify trends and shifts in research focus within the journal and, by extension, the broader Korean medical informatics landscape. By examining the keywords used across these papers, we aimed to clarify the evolving priorities and innovations that have characterized the field over time in South Korea. This analysis will not only provide valuable insights into the development of medical informatics as a discipline in Korea but also inform future research directions and policy decisions.
II. Methods
1. Data Access
We accessed journal data through the HIR journal’s online archive, which provides open access to all published articles. Metadata and text content were systematically extracted for all 958 articles published between 1995 and 2024 using the Korea Citation Index database. Of these, 672 were identified as English-language publications. Because HIR has officially published exclusively in English since 2010, our primary analysis focused on the 658 English-language articles published between 2010 and 2024. The extracted dataset included article titles, author names, languages, abstracts, keywords, and publication dates.
2. Trends in Publication Frequency and Language Distribution
To examine publication trends, descriptive statistical analyses were conducted on the 658 English-language papers published between 2010 and 2024. We analyzed the annual distribution of publications, investigating changes in publication frequency and article types across this period.
3. Analysis of Keywords and MeSH Terms
Comprehensive keyword and Medical Subject Headings (MeSH) term analyses were conducted on a subset of 632 papers (from the 2010–2024 cohort) with available abstracts. For keyword analysis, both titles and abstracts were used. Initially, key terms were extracted using the term frequency-inverse document frequency (TF-IDF) method, which identifies words that are both frequent and distinctive within the corpus. We then generated word clouds to visually represent the most prominent keywords, offering an intuitive overview of dominant themes. Latent Dirichlet allocation (LDA) [6] was subsequently applied for topic modeling to identify major research themes across the papers. To assess temporal patterns in keyword usage, the top 25 keywords based on overall TF-IDF scores were selected. Their yearly TF-IDF scores were calculated, and K-means clustering was applied to group keywords exhibiting similar temporal patterns over the 2010–2024 period. The optimal number of clusters was determined using the elbow method.
For the MeSH term analysis, we examined author-submitted MeSH keywords from these 632 papers. Frequency analysis was conducted to identify the most frequently used MeSH terms in the field. A co-occurrence analysis was then performed to explore relationships among keywords, revealing patterns and associations within research topics. This approach enabled us to understand how authors categorized their studies using standardized medical vocabulary.
4. Inferential Statistical Analysis
To identify changes in research trends over time, we conducted two main inferential statistical analyses using Python version 3.11.3. Statistical computations were performed with the scipy.stats library (specifically chi2_contingency) and the statsmodels.api package for regression analyses. First, chisquare tests were used to assess the statistical significance of changes in keyword distributions across different time periods, determining whether observed variations were meaningful or occurred by chance. Second, linear regression analyses were conducted to examine temporal trends in major keywords and topics. This approach allowed us to quantify and visualize the evolution of research themes, identifying which topics were gaining or declining in prominence within the field.
5. Generative AI Use
We used GPT-4o to generate code for data analysis and Perplexity Pro to assist with English writing [7,8]. The use of generative AI did not influence the authors’ selection of research topics or methodological design. Instead, it improved efficiency in code generation and enhanced fluency in English writing.
III. Results
1. Descriptive Statistics
Our initial dataset comprised 958 research papers published between 1995 and 2024, including 672 written in English. Focusing on the period after the journal’s official transition to English in 2010, the descriptive analysis of publication trends (Figure 1) included 658 English-language papers published from 2010 to 2024. This period showed an overall upward trend beginning in 2010, peaking between 2015 and 2018, when 48 papers were published annually. After this peak, publication numbers stabilized, with 44 papers published per year in subsequent years. Subsequent keyword and MeSH term analyses were performed on the subset of 632 papers from this 2010–2024 cohort that included abstracts.
Distribution of research papers published in Healthcare Informatics Research from 2010 to 2024.
Annual distribution of research papers published in HIR from 2010 to 2024, categorized by article type. The stacked bars illustrate the number of publications for each type (editorial, review, original, case report, tutorial, communication, book review, application) per year.
2. Evolution of Research Themes Based on Keyword Analysis
The word cloud visualization in Figure 2 provides an intuitive representation of the most frequently occurring keywords in the research titles and abstracts of 632 papers published in HIR from 2010 to 2024. This visualization was generated using the top 100 keywords from the dataset, excluding common terms such as “results” and other general descriptors. Prominent words such as “health,” “data,” “system,” and “medical” dominate the word cloud, underscoring their centrality to the field’s focus on health systems, medical data, and information management. Keywords like “clinical,” “patients,” “study,” and “hospital” emphasize a strong interest in patient-centered care and hospital-based healthcare systems. In addition, terms such as “learning,” “technology,” “factors,” and “model” highlight growing attention to machine learning, technological innovation, and research methodology. Words including “method,” “analysis,” “nursing,” and “records” point to the field’s sustained commitment to methodological rigor and healthcare documentation. The inclusion of terms such as “telemedicine,” “safety,” “disease,” and “evaluation” demonstrates a continued interest in remote healthcare delivery, patient safety, disease management, and healthcare practice assessment.
Word cloud of the top 100 keywords from titles and abstracts.
Next, using the combined text from the titles and abstracts of these 632 papers, we extracted key terms through the TF-IDF method, which identifies words that are both frequent and distinctive within the corpus. To further explore keyword dynamics, the top 25 keywords were selected based on their overall TF-IDF scores. Using their annual TF-IDF trends, a clustering analysis was performed to group keywords exhibiting similar temporal patterns. The resulting four clusters provided a more nuanced understanding of the evolution of research themes.
Cluster 0 included terms such as “healthcare,” “learning,” “model,” and “research,” which showed gradual increases or stable prominence over time, reflecting their ongoing relevance in healthcare informatics. Cluster 1 encompassed terms like “data,” “health,” and “study,” which maintained consistently high TF-IDF scores throughout the study period, with marked peaks around 2014 and 2024, suggesting their persistent centrality in the field. Cluster 2 comprised keywords such as “application,” “care,” “hospital,” “nursing,” and “technology,” which exhibited fluctuating prominence, indicating shifts in research priorities across healthcare systems and technological domains. Cluster 3 contained terms including “clinical,” “information,” “medical,” and “system,” which showed a gradual decline in prominence after 2012, possibly reflecting a transition toward more specific or emerging research topics. This clustering analysis demonstrates the persistence of foundational concepts such as “data” and “health,” alongside the rise of newer themes like “technology” and “nursing.” It highlights the field’s dynamic progression, balancing enduring core topics with emerging areas of inquiry in health informatics (Figure 3).
Keywords trends over time for different clusters: (A) cluster 0, (B) cluster 1, (C) cluster 2, and (D) cluster 3.
3. Analysis of MeSH Term Trends and Relationships
Figure 4 illustrates the distribution of MeSH keywords across research papers from 2010 to 2024, capturing the evolving focus of health informatics research. The frequency of MeSH terms shows a noticeable increase beginning in 2010, peaking around 2017, followed by fluctuations that remain relatively high through 2024. This pattern reflects the growth and diversification of research activity in health informatics over time. Terms such as “telemedicine,” “medical informatics,” “machine learning,” and “electronic health records” appear consistently across multiple years, underscoring their enduring significance. “Telemedicine” exhibits a sharp rise around 2014, reflecting heightened interest in remote healthcare solutions. Similarly, “artificial intelligence” and “deep learning” gain prominence from 2016 onward, signaling the increasing integration of advanced computational approaches into medical research. The terms “smartphone” and “mobile applications” also rise in frequency during this period, indicating the incorporation of mobile technologies into healthcare delivery. In contrast, terms such as “patient safety” and “classification” show moderate fluctuations, while “electronic health records” and “health information systems” maintain steady relevance, underscoring their foundational roles in healthcare data management. Notably, keywords such as “data mining,” “personal health records,” and “information storage and retrieval” emerge intermittently, suggesting evolving research priorities and specialized topics. Overall, the diversity and changing frequency of MeSH terms reflect an expanding scope of health informatics research— encompassing data analytics, machine learning, mobile health, and telemedicine—which aligns with emerging challenges and priorities in the field.
Temporal distribution of Medical Subject Headings (MeSH) terms in research papers from 2010 to 2024.
In our inferential statistical analysis, the chi-square test revealed no significant changes in keyword distribution over time (χ2 = 49.31, p = 0.656), suggesting that the observed variations may be random. Additionally, a regression analysis of the MeSH term “health personnel” indicated a potential positive trend over time, with an R2 value of 0.245; however, the trend did not reach statistical significance (p = 0.061). These results suggest that there is no strong evidence of systematic or directional change in research trends across the analyzed period.
Figure 5 presents a topic modeling visualization generated using LDA, which identified five principal research themes based on MeSH keywords across the analyzed papers. The first topic, “Health/Medical Informatics Systems in Korea,” represents the foundational concepts of the field. Keywords such as “Health,” “Medical,” “Information,” “Electronic,” “Korea,” and “Systems” highlight the focus on electronic health records and informatics systems within the Korean context. The second topic, “Machine Learning in Medical Data Analysis,” centers on advanced computational methods. Representative terms such as “Learning,” “Machine,” “Deep,” “Data,” “Prediction,” and “Classification” underscore the growing importance of computational methods for extracting insights from health data and improving system performance. The third topic, “Clinical Data Modeling and Systems,” focuses on the application of data analysis in clinical settings. Keywords such as “Data,” “Clinical,” “Development,” “Model,” and “Prediction” indicate a strong and persistent interest in developing predictive models and systems based on clinical data. The fourth topic, “Healthcare Information Technology and Management,” emphasizes the practical application and management of hospital information systems. Terms including “Application,” “Information,” “System,” “Records,” “Hospital,” and “Management” reflect the growing adoption of mobile technologies to support patient engagement, education, and health system efficiency. The fifth topic, “Systematic Reviews of Health Technology in Korea,” focuses on the evaluation and overview of healthcare technologies. Keywords such as “Health,” “Review,” “Systematic,” “Korea,” and “Quality” capture ongoing research efforts to enhance information infrastructure and data-driven management across healthcare settings. Collectively, these topics reveal a strong emphasis on leveraging digital technologies—particularly machine learning and electronic health record systems—to address key healthcare challenges. The recurrent prominence of keywords related to data analysis, systems, and patient data further underscores a sustained shift toward digitization, data-centric systems, and patient-centered care in health informatics research.
Visualization of latent Dirichlet allocation topic modeling using MeSH terms.
Figure 6 presents the filtered MeSH keyword co-occurrence network, which includes only those keywords appearing at least ten times. The node sizes are proportional to keyword frequency, while edge widths represent the frequency of co-occurrence between keywords. Prominent terms such as “telemedicine,” “machine learning,” and “electronic health records” dominate the network, reflecting their central role in health informatics research. Strong linkages are evident between “machine learning” and “artificial Intelligence,” as well as between “telemedicine” and “data mining,” suggesting frequent co-occurrence and shared research foci. Additional significant keywords, including “patient safety,” “mobile applications,” “deep learning,” and “data mining,” emphasize the growing focus on technological innovation and data-driven methodologies in healthcare. The network also reveals emerging topics such as “COVID-19” and “remote consultation,” highlighting their recent and increasing importance in the field.
Filtered keyword co-occurrence network.
Figure 7 displays a heatmap of co-occurrence frequencies among the top 20 MeSH terms, providing a visual overview of their interconnections within the research domain. The color intensity corresponds to the strength of co-occurrence, with darker shades indicating more frequent associations. Notable relationships include “electronic health records” co-occurring significantly with both “medical informatics” and “patient safety,” underscoring their central position within healthcare informatics research. Similarly, “machine learning” and “artificial intelligence” exhibit strong co-occurrence, reflecting their pivotal roles in advancing data-driven healthcare solutions. Another key association is observed between “mobile applications” and “telemedicine,” highlighting the ongoing emphasis on mobile technologies and their integration into remote healthcare delivery. Collectively, these patterns reveal the major thematic clusters and interdisciplinary linkages shaping contemporary health informatics research.
Co-occurrence heatmap of Medical Subject Headings (MeSH) terms (top 20 terms).
IV. Discussion
The bibliometric analysis of HIR journal publications from 2010 to 2024 provides valuable insights into the evolution of health informatics research, offering a distinctive perspective on the development of this field in Korea and its global contributions. This comprehensive examination of HIR’s publications not only synthesizes key findings and their implications for the discipline but also highlights the journal’s pivotal role in documenting and shaping the trajectory of health informatics in Korea over the past three decades. Since its inception, HIR has closely mirrored the growth and maturation of medical informatics in South Korea, serving as a barometer for the country’s expanding engagement with and contributions to this rapidly evolving domain. The keyword trends observed in this analysis vividly illustrate the changing landscape of health informatics research in Korea, showcasing the nation’s transition from early adoption to innovation in areas such as telemedicine, artificial intelligence, and mobile health applications. This evolution aligns closely with Korea’s broader national initiatives in digital health and smart healthcare, emphasizing the synergy between academic research and policy priorities [9].
This study complements the recent editorial by Chang [5], which offered a valuable historical overview and reflection on the milestones of HIR throughout its 30-year journey. While the editorial provided a qualitative perspective on the journal’s growth and impact, our study contributes a distinct quantitative analysis focusing on the evolution of research topics since 2010—the period marking HIR’s transition to exclusive English-language publication. By employing bibliometric methods such as TF-IDF keyword analysis, LDA topic modeling, and MeSH term co-occurrence analysis on a defined corpus of recent articles, this research delivers data-driven insights into thematic shifts and emerging trends represented in HIR. Consequently, our work builds on the historical foundation provided by Chang [5] by offering a detailed, methodologically rigorous examination of the contemporary research landscape documented in the journal.
The analysis revealed a steady increase in publications beginning in 2010, peaking at 48 papers annually between 2015 and 2018. This growth suggests a period of rapid expansion in the field, likely driven by increased funding opportunities, technological progress, and supportive policy initiatives. The analysis revealed a steady increase in publications beginning in 2010, peaking at 48 papers annually between 2015 and 2018. A modest decline in publication numbers was observed in subsequent years, stabilizing at 44 papers published annually. Importantly, the period from 2020 to 2022 coincided with the global COVID-19 pandemic and widespread lockdowns. Despite these challenges, publication rates remained highly stable, with 44 papers published consistently in 2020, 2021, and 2022 —demonstrating the field’s resilience and adaptability [10].
Our analysis of keywords and research trends underscores the dynamic and evolving nature of health informatics. Clustering analysis revealed that foundational topics such as “data” and “health” remained central to the field, while emerging areas such as “technology” and “nursing” gained increasing prominence. Terms like “learning,” “model,” and “research” appeared consistently over time, whereas keywords such as “clinical” and “system” declined in prominence after 2012, suggesting a shift toward more specialized and technologically advanced areas of inquiry. The word cloud visualization further highlighted the most frequent keywords—”health,” “data,” “system,” and “medical”—underscoring the discipline’s emphasis on health systems, data management, and information technology. Additionally, terms such as “telemedicine,” “safety,” and “evaluation” pointed to sustained interest in remote healthcare, patient safety, and the assessment of healthcare practices. The appearance of terms like “learning” and “technology” reinforced the growing integration of artificial intelligence and machine learning into healthcare research. Moreover, the recent emergence of pandemic-related keywords such as “COVID” reflected the field’s responsiveness to global health crises, illustrating its capacity to adapt to urgent societal needs. Overall, research activity between 2010 and 2024 steadily increased, affirming the growing recognition of health informatics as a critical and interdisciplinary component of modern healthcare systems.
The MeSH term analysis provided further insights into the evolution of health informatics research from 2010 to 2024. Core themes such as “telemedicine,” “medical informatics,” “machine learning,” and “electronic health records” maintained consistent prominence, demonstrating their foundational significance within the discipline. Topic modeling identified five major research themes—medical informatics and artificial intelligence, machine learning and natural language processing, telemedicine and remote health solutions, health information systems and obesity research, and healthcare systems and information management—reflecting the field’s emphasis on integrating advanced technologies with healthcare delivery systems. The co-occurrence analysis of MeSH terms revealed strong interconnections between technological and healthcare domains, particularly among “telemedicine,” “machine learning,” “artificial intelligence,” and “electronic health records.” Although chi-square analysis found no statistically significant systematic change in keyword distribution over time (p = 0.656), regression analysis indicated emerging trends in specific areas, such as an increasing focus on health personnel. Collectively, these findings highlight the evolution of health informatics toward the integration of innovative technologies while maintaining a focus on core healthcare systems and patient-centered care— positioning the field to address the complex challenges of contemporary healthcare effectively.
Analyzing both author keywords from titles and abstracts alongside assigned MeSH terms provided complementary perspectives on the research landscape. Keywords extracted directly from text capture the specific language and emerging concepts used by authors, offering a direct reflection of terminology trends. In contrast, MeSH terms represent a standardized, controlled vocabulary that enables analysis based on established medical concepts and their hierarchical relationships. Because MeSH encompasses both broader categories (e.g., “artificial intelligence”) and narrower, more specific terms (e.g., “machine learning”), its use requires careful interpretation. Nevertheless, incorporating MeSH proved highly valuable: it enabled structured analyses such as topic modeling (Figure 5), which revealed distinct themes grounded in standardized definitions, and co-occurrence analyses (Figures 6 and 7), which visualized the strength of relationships among key medical informatics concepts. Together, these approaches—integrating both free-text keywords and structured MeSH terms—provided a more comprehensive and nuanced understanding of the evolution of research trends than either method could have achieved alone.
This study has several limitations. First, the analysis is limited to publications from a single journal, HIR, which may not fully represent the broader health informatics field. Second, the reliance on MeSH terms for classification may overlook subtle distinctions and emerging topics not yet incorporated into the controlled vocabulary. Furthermore, assessing the long-term impacts of global events such as the COVID-19 pandemic will require continued observation and future data collection. Future research should expand the scope to include publications from multiple journals to obtain a more holistic view of global research trends. In addition, further exploration of ethical considerations and user acceptance surrounding emerging technologies, as well as longitudinal evaluation of the pandemic’s influence on research priorities, would yield deeper insights into the adaptability and responsiveness of the health informatics field.
This bibliometric analysis provides a comprehensive overview of research trends within HIR, emphasizing the dynamic yet stable nature of the health informatics research ecosystem. The findings highlight the field’s capacity to balance rapid technological innovation—such as telemedicine and artificial intelligence—with enduring core themes that remain central to healthcare research. Since its transition to an English-only format in 2010, HIR has played a critical role in disseminating medical informatics knowledge, mirroring South Korea’s growing engagement and leadership in this domain. By contextualizing our findings within HIR’s 30-year trajectory, this study not only complements prior historical analyses but also contributes to the broader discourse on the role of specialized journals in advancing medical informatics. Identifying emerging areas of focus and aligning research priorities with international developments will be essential for guiding future investigations and informing evidence-based policy decisions.
Notes
Conflict of Interest
Kye Hwa Lee and Hyejung Chang are editorial members of Healthcare Informatics Research; however, they did not involve in the peer reviewer selection, evaluation, and decision process of this article. Otherwise, no potential conflict of interest relevant to this article was reported.
Acknowledgments
We would like to express our gratitude to Soyeong Park for her valuable contributions to this research, including data collection and consulting. Her expertise and efforts greatly enhanced the quality of this study.
This research was supported by a grant of the Korea Health Technology R&D Project through the Korea Health Industry Development Institute (KHIDI), funded by the Ministry of Health & Welfare, Republic of Korea (grant number: HR21C0198) and 2024IP0015-1 from the Asan Institute for Life Sciences, Asan Medical Center, Seoul, Korea