How does research output and impact in medical informatics vary among EU member states? – A bibliometric analysis

Introduction

Medical informatics is an interdisciplinary field that plays an important role in modern healthcare systems by integrating data science, technology, and clinical practices to improve health outcomes. It includes a wide range of research areas, such as health information systems, telemedicine, artificial intelligence, applications in diagnostics, digital health solutions, and continuity of care [1], 2].

This field’s growing significance was especially evident during the COVID-19 pandemic when digital health technologies were heavily used to ensure continuity of care through telehealth and remote patient monitoring [3], [4], [5], [6]. In addition, during the pandemic, researchers started exploring real-life use cases of artificial intelligence (AI) for triaging patients, diagnosing, and guiding treatment. Mann et al. (2022) investigated the deployment of 66 AI applications during the COVID-19 outbreak but found that limited independent evaluations of their effectiveness were available. In addition, they noted that many of these applications were used mostly in the United States, China, and other high-income countries [7].

In the European Union, medical informatics research has emerged as a domain of focus. Various studies approach the Union’s actual or potential role in this space. Aarestrup et al. (2020) discuss the EU’s initiative to create a digital health infrastructure that would aid in data sharing and analysis across member states [8]. Similarly, Kondylakis et al. (2023) examine multiple EU-funded projects aimed at developing big data infrastructures to support AI in medical imaging [9]. Previous research has also been published discussing the challenges and opportunities of big data in healthcare, even offering recommendations for an EU action plan [10]. Additionally, the Union has performed its own studies on how to use AI safely for medical applications [11], culminating with the adoption of the regulation 2024/1689 of the European Parliament and of the Council of 13 June 2024, colloquially known as the EU AI Act, which lays down rules on artificial intelligence use cases, including medical ones [12]. These efforts reflect significant focus and investments in this area of research.

The research output in medical informatics allows us to also evaluate if the publications in this field follow the open-access publishing trend, and in doing so facilitate knowledge dissemination and equitable access to scientific findings. The EU’s support for open-access publishing, as advocated by initiatives like PlanS and cOAlition S, has further catalyzed the global shift toward transparency and inclusiveness in scientific research. This trend aligns with previous observations of increased open-access publishing [13], [14], [15]. Therefore, a metric we consider in the present paper is whether the research output in medical informatics follows the OA momentum seen in other areas of research [16].

Another important aspect analyzed in this study is whether the research output disparities between EU member states highlighted in previous studies still hold today. Peterson (2021) has highlighted the uneven distribution of research and development funding across geographic regions [17]. Western and Eastern European countries frequently secure uneven funding, resulting in evident differences in research volume and impact. Gallo et al. (2021) analyzed the distribution of biomedical and health research funding in the EU’s Horizon 2020 framework and found significant inequalities, with EU-15 countries securing 87 % of the funding while EU-13 countries received only 3 %, reflecting persistent gaps in research infrastructure and capacity​ [18]. Similarly, Polašek and Kern (2012) observed substantial disparities in medical informatics output across 33 European countries, linking higher research productivity to GDP levels and highlighting Switzerland, the Netherlands, and the Scandinavian countries as leaders in the field [19]​. Unim et al. (2022) further explored the role of research networks in enhancing health data collection and sharing, identifying 57 research networks across Europe. While most networks contributed to improved data quality and accessibility, significant gaps in data-sharing practices and adherence to metadata standards were noted, posing challenges to health information integration and policy planning [20]​.

The bibliometric analysis presented in our study provides an overall evaluation of the medical informatics research published in 47 journals identified in the Web of Science, having EU-affiliated authors between 2018 and 2022. These 47 journals represent all identified journals that publish research in English indexed by WoS under the medical informatics standardized category, covering all relevant research for this field. The 2018–2022 timeframe was selected because we wanted to capture trends in this field in recent years, but also cover the time frame that was marked by the COVID-19 pandemic, which influenced research output and innovation in digital health technologies, telemedicine, and AI applications in healthcare. The focus on metrics such as cumulative impact factors, collaboration types, and publishing models offers insights into this evolving research field, but more importantly, it allows us to create a general idea of how each EU country ranks in terms of research productivity and impact in this niche. While our study aimed to elucidate the volume and impact of publications, we also analyzed the collaborative networks and thematic trends that define this field in the European context.

Our research contributes to a better understanding of the development of medical informatics in the EU. It offers insights for researchers, policymakers, and funding bodies, highlighting the need to address regional disparities, invest in collaborations, and promote easy access to research outputs. These findings are meant to provide a starting point for future investigations into the factors that influence the quality and impact of medical informatics research within the European Union, especially considering the constantly changing healthcare challenges and technological advancements.

Methods

Data collection

We utilized Clarivate’s InCites to identify journals classified under the Medical Informatics research area according to the Web of Science schema. Subsequently, we applied specific filters as outlined in Figure 1 (time span, research area, location, document type, language). The workflow depicted in Figure 1 ensures we are working with a consistent dataset and avoiding inconsistencies that can result from the InCites data being updated separately from Web of Science.

Figure 1:

Workflow used to identify medical informatics-focused journals that published papers in English, indexed by WoS under the “medical informatics” category between 2018 and 2022 published by EU-affiliated authors, using InCites.

Journal selection criteria: we have selected all journals that are indexed by Web of Science, that have published research articles in English within 2018–2022, and that are categorized by Clarivate within the medical informatics category, having published at least one paper authored by researchers affiliated with institutions in European Union member states; schema: Web of Science; source type: journals. A total of 47 journals met the selection criteria. Data was retrieved in May 2024. We have compiled the list of journals, their abbreviation, and impact factors throughout 2018–2022 using Clarivate’s Journal Citation Reports in Supplementary Table S1.

The 2018–2022 timeframe was selected to ensure consistency and comparability in impact metrics. At the time of data retrieval, many journals had not yet received their 2023 Journal Impact Factors (JIF) in Clarivate’s Journal Citation Reports. Additionally, Clarivate revised its JIF calculation methodology in 2023, expanding the range of indexed journals [21]. Including 2023 data would have introduced inconsistencies in cumulative metrics and affected longitudinal comparability.

Results

Research output overview

The dataset includes 6,620 papers published between 2018 and 2022 in 47 journals categorized under “medical informatics” in Clarivate’s Web of Science. These papers are in English and have at least one author affiliated with an institution in the European Union.

Figure 2 presents a correlation between the number of papers published per EU country and the cumulative yearly impact factors (CIF) between 2018 and 2022. Each article is counted once for every country with at least one affiliated author. If an article has co-authors from multiple countries, it is counted once for each country represented. The cumulative impact factor is particularly suitable for analyzing research output at a country level because it reflects both the quantity and the quality of publications by integrating the impact factor of journals in which the articles were published. Unlike citation counts, which can be skewed by a few highly cited articles, CIF provides a broader overview of the collective influence of a country’s research output. Similarly, altmetrics focus on social media attention and non-traditional sources, which do not represent the academic quality or relevance of publications in this field.

Figure 2:

Academic output and impact across EU member states (2018–2022). (A) Number of articles published by each EU member state from 2018 to 2022. Darker blue shades represent higher publication volumes. (B) Cumulative impact factors of articles published by each EU member state from 2018 to 2022. Darker green shades correspond to higher cumulative impact factors.

We note that from 2018 to 2022, the number of academic articles and their cumulative impact factors across the 27 EU member states experienced significant growth. The total number of yearly published articles rose steadily from 1,170 in 2018 to a peak of 1,931 in 2021, before a slight decline to 1,781 in 2022. Similarly, yearly cumulative impact factors increased sharply by 267 % from 2018 to 2021, reaching 11,787, followed by a modest decrease of 13.2 % in 2022 (Supplementary Table S3). A strong positive correlation (r=0.93) between article numbers and cumulative impact factors was to be expected, and confirms the link between research volume and overall impact. However, the faster growth in impact factors compared to articles (53 vs. 12.3 % annually on average) highlights a strategic shift towards quality (Supplementary Table S4). Germany, France, Italy, Spain, and the Netherlands emerged as leading contributors in both output and impact. Other nations like Ireland and Finland showed remarkable relative growth in impact factors, indicating a focus on research quality. In contrast, countries such as Bulgaria and Croatia maintained consistently low output and impact, pointing toward potential disparities in academic infrastructure and/or interest in medical informatics.

Regarding the preferred publishing model, over 75 % (n=5,000) of the 6,620 papers were published under an Open Access (OA) model. A clear trend emerges when examining publishing patterns between 2018 and 2022: the number of OA papers steadily increased each year, while the number of traditionally published papers declined (Figure 3). This trend aligns with the global shift toward Open Access (OA) publishing. Studies have shown a significant increase in OA publications over recent years. For instance, Demeter et al. (2021) conducted a comparative empirical analysis across seven world regions and nine academic disciplines, revealing a substantial rise in OA publishing [24]. Similarly, Schiltz (2018) discussed the objectives of cOAlition S in promoting full and immediate OA, highlighting the movement’s momentum [25]. Koutras (2015) also examined the role of OA in the context of globalization, underscoring its growing importance in disseminating research [26].

Figure 3:

Comparison of open access (OA) vs. traditionally published papers through 2018–2022, from the dataset; horizontal axis: time, vertical axis: number of articles.

The most productive authors by the number of papers co-authored from the dataset (data from InCited as of 2 June 2024; person ID type group: WoS Researcher Profile) are visualized in Supplementary Figure S1 alongside the cumulative citations their articles received. Cumulative citations are determined by InCites and account from the time of publication up to Apr 30, 2024. We used InCites’ WoS Researcher Profile because this function groups authors under unified profiles and minimizes cases where the same author appears with different name variations.

In terms of the most productive affiliations publishing research on medical informatics between 2018 and 2022 (Supplementary Figure S2), Netherlands-based and France-based institutions are leading the way, occupying six of the top 10 positions (three each). The most productive affiliations is INSERM (France) with 319 papers authored by researchers affiliated there, followed by Karolinska Institutet (Sweden) with 233 papers. In terms of the most cited, the University of London (UK) ranks the highest with 4,634 cumulative citations for its 229 articles.

Collaboration metrics

Most EU countries have a high percentage of international collaborations, averaging 66.01 %. This highlights strong cross-border research cooperation within the EU and beyond. Domestic collaborations average 18.68 %, indicating a good level of national cooperation between different institutions from the same country. Internal collaborations are slightly lower at an average of 12.71 %, suggesting that within-institution collaboration is less common compared to cross-institution collaboration. Single-author papers are relatively rare, averaging only 2.59 %, which implies that most research is collaborative. In addition, industry collaboration (as defined in chapter 2.2.3 Collaboration Metrics) is present but not predominant, averaging 5.53 %. This indicates that while there is some engagement with the private sector, it is not the main focus of most research efforts. Luxembourg has by far the highest rate of industry collaborations, with 20.59 % (which only amounts to 7 out of 34 papers), followed by Belgium with 13.75 % (59 of 429 articles). However, in terms of the number of documents that have industry collaborations, Germany leads with 144 out of the 1403 total (Supplementary Table S5).

In Figure 4 we illustrate the distribution of collaboration types across EU member states in the medical informatics field from 2018 to 2022. The stacked bar chart highlights significant variations in collaboration preferences among countries. International collaborations dominate the research output for most nations, particularly in Belgium, Austria, and Cyprus, where they represent the largest proportion of publications. Domestic collaborations are highlighted in the dataset in countries like Belgium and Germany, which could either point toward robust intra-national research networks or maybe even national funding policies and research programs encouraging such partnerships. Internal collaborations and single-author papers are comparatively rare, with internal collaborations more prominent in smaller countries such as Croatia and Bulgaria. These aspects show the reliance of most EU countries on international partnerships.

Figure 4:

Collaboration types by country, ordered by total articles. Horizontal axis: countries. Vertical axis: number of articles based on the type of collaboration. The sum of the collaborative articles equals the total number of papers published within 2018–2022 by each country.

The collaborative co-authorship network can be visualized in Supplementary Figure S3. There are 17 distinct clusters determined by VOSviewer. Interestingly, four of the top ten authors from Supplementary Figure S2 are in close proximity in the network, having established strong collaborations: Andersson G., Riper H., Botella C., and Carlbring P. The rest of the top ten authors are not in close proximity to each other or any other from the top 10 list, suggesting they have their own distinct collaborative groups.

Figure 5 illustrates the collaboration networks between countries based on the papers in the dataset. Prominent countries in the network include Spain, France, and Germany, which exhibit extensive connections with many other countries. While this paper focuses on the EU countries, we note that the United States is also featured prominently in this network, indicating substantial collaborative efforts with the EU. Various clusters of countries can be observed, suggesting regional collaboration trends. For instance, the green cluster indicates strong collaboration among Northern and Western European countries such as Sweden, Norway, Finland, and Belgium, but also non-EU members such as Iceland and the UK. Another cluster, highlighted in red, includes extensive collaborations between France, Canada, and various Asian countries like China, Thailand, Malaysia, and India.

Figure 5:

Co-authorship network map based on the geographic locations of authors’ affiliations, visualized with VOSviewer (v 1.6.20). Each node represents a country, with the size of the node indicating the volume of research output. The lines between nodes represent co-authorship links, where thicker lines denote stronger collaborative relationships.

Keyword mapping

Figure 6 illustrates several interconnected clusters, each representing distinct thematic areas within the field, based on the author keywords of the 6,620 papers in the dataset. The first prominent cluster (red) includes terms related to “machine learning” and “deep learning,” signifying a strong interest in related research indexed under the medical informatics WoS category. The second significant cluster (purple) focuses on “ehealth” and “mhealth,” highlighting the increasing importance of mobile health technologies and digital health interventions. The “covid-19” cluster reflects the extensive research response to the pandemic, integrating aspects of epidemiology, public health, and contact tracing. Additionally, keywords like “natural language processing” and “survival analysis” indicate the utilization of advanced computational techniques in health data analysis. The concept map in Figure 6 further confirms the interdisciplinary nature of medical informatics research in general and in the EU.

Figure 6:

Keyword map using VOSviewer (v 1.6.20) based on the author keywords from the dataset.

Publishing venues

Journals diversity and distribution

We determined the correlation coefficient (Supplementary Table S6) between the number of articles published by each country and the number of distinct journals (in InCites) in which said papers were published by each country: r=0.756 (p<0.001). This indicates a positive relationship between the two variables: as the number of articles increases, the number of distinct journals in which they are published also tends to increase.

In Figure 7 we highlight the number of total articles published by each country between 2018 and 2022, and the number of total distinct journals in which each country has published. The papers published by researchers affiliated in Germany (n=1,403), The Netherlands (n=1,154), and Spain (n=1,098) exhibit the highest journal diversity (Germany n=39 distinct journals, The Netherlands n=35; Spain n=32). However, interestingly, even though there is a positive correlation between the volume of articles and the diversity of journals, we observe that there are countries that publish a much lower number of papers, yet maintain almost the same diversity of journals, as it is the case with Denmark (n=32), Finland (n=32), Portugal (n=32), Austria (n=31), and Ireland (n=29). It is possible that the researchers from these countries have a strategic approach to maximizing the impact and outreach of their articles, publishing in a diverse range of journals, in an attempt to reach wider audiences.

Figure 7:

The number of articles per EU country (black bard) vs. the number of distinct journals in which they are published (blue line); ordered by article number. Horizontal axis: EU countries. Left-vertical axis: number of articles. Right-vertical axis: number of distinct journals.

The diversity of journals in which authors publish can fluctuate over time based on many factors – from publishing trends to funding restrictions. However, it is noteworthy to see how much fluctuations we observe over time for each of the 27 EU countries. In Supplementary Figure S4 we visualize the aforementioned aspects (data in Supplementary Table S7). We note that some countries have wider fluctuations than others. Austria and Belgium show consistent publication numbers, indicating stable engagement in a similar range of journals each year. Austria’s journal diversity remains around 21–22 journals annually, while Belgium fluctuates slightly between 19 and 24 journals. Similarly, Germany and France also exhibit stability, with Germany maintaining around 35–36 journals per year and France ranging from 26 to 31 journals. Italy and Spain show slight fluctuations but maintain a relatively high number of journals, with Italy ranging from 20 to 24 and Spain from 29 to 32. On the other hand, we observe wider fluctuations in instances such as for Bulgaria and Croatia-based authors. Bulgaria published in only one journal in 2018 (10 papers), namely the Journal of Evaluation in Clinical Practice, but increased to five journals by 2022 (6 papers) – meaning it published less than in 2018 but in a much more diverse portfolio of journals. Croatia’s numbers range from 3 to 6 journals, indicating modest variability. Cyprus shows consistent numbers but with slight year-to-year changes, maintaining between 4 and 6 journals.

Top journals

While there are 47 distinct journals in our dataset, authors published most of their research in a select few. The top 15 journals by the number of papers published from the total 6,620 in the dataset are showcased in Figure 8. The articles within these 15 journals (5,205 papers) amount to 78.6 % of the total volume published in the medical informatics WoS category from EU authors between 2018 and 2022.

Figure 8:

Top 15 journals by the number and percentage of medical informatics publications from EU-affiliated authors (2018–2022). Journal names are abbreviated according to Clarivate’s abbreviations.

Out of these top 15 journals, about half have relatively high impact factors (Supplementary Table S8) compared to the 3.81 average journal impact factor (JIF) of the 47 total journals.

Discussions

Our study provides a bibliometric evaluation of medical informatics publications among European Union countries between 2018 and 2022, highlighting several trends and challenges in the field. However, there are important limitations that must be acknowledged, which also point to potential for future research.

The analysis relied exclusively on the Web of Science and InCites databases, which, while comprehensive, do not include all academic literature on medical informatics. It is possible other databases, such as Scopus, PubMed, orOpenAlex, could offer complementary perspectives or broader coverage. We focused on English-language publications because medical informatics is an interdisciplinary field that often involves collaboration among researchers from different countries, making English the likely preferred language for scholarly communication. This assumption is supported by the data shown in Figure 1: applying the English language filter reduced the number of articles only slightly, from 6,626 to 6,620, and the number of journals from 49 to 47. These minimal exclusions suggest that only a very small fraction of the relevant literature in the Medical Informatics category was published in other languages.

Our study centers on EU member states, which limits the generalizability of its findings to other regions. While this approach allows for targeted analysis, future research could benefit from comparative studies that include non-EU countries to provide a broader perspective on global trends in medical informatics research. As we note in Chapter 3.3 Collaboration Metrics, non-EU countries, such as the USA, are also highly involved in medical informatics research.

Another limitation of this study is the reliance on metrics such as impact factors, cumulative impact factors, and the number of articles as indicators of research productivity and quality. While these metrics provide valuable insights into the volume and visibility of a country’s output, they do not reflect the qualitative nature of the research produced, which could be analyzed through approaches such as peer-review assessments, expert panels, or real-world implementation reports, to evaluate the actual impact of medical informatics innovations on healthcare outcomes. High publication counts or impact factors may indicate productivity and recognition within academic circles but fail to capture the real-world implications of the research, such as its contribution to improving healthcare systems, patient outcomes, or public health policies. Similarly, countries with lower publication counts may still produce research of significant societal or clinical value that is not adequately reflected in these metrics.

Investigating the role of funding in medical informatics research could be an interesting future research direction for similar articles as this one. However, we note that there are immediate challenges that need to be overcome, especially in data availability and processing due to the limitations of the WoS database in aggregating funding information. While funding sources are often listed, inconsistencies in how they are reported pose significant hurdles for data analysis. For instance, funding entities linked to the EU are not consistently categorized as a single entity, or aggregated under an “EU umbrella” in any way. Each EU-funded project is listed as a distinct funding entry, making it difficult to systematically track the EU’s contributions to this field. Additionally, errors such as typographical mistakes (e.g., “Euorpean Union” instead of “European Union”, which is something we have observed in the funding information for some articles in the dataset), and variations in reporting styles further complicate the extraction of meaningful insights. Some authors include funding acknowledgments within the acknowledgments section instead of the designated funding section, adding to the inconsistency. These issues limit the ability to analyze the impact of specific funding bodies on research outputs. Furthermore, while our study focused on bibliometric indicators, we recognize the value of incorporating complementary metrics in future research. In addition to the aforementioned funding metrics, other indicators such as the number of medical informatics EU-funded projects, national and EU-level public policies supporting digital health innovation, pilot programs implemented across member states, or patent activity could provide a more holistic understanding of how research output translates into practical applications. Future work integrating these data sources could better capture the relationship between academic production, policy adoption, and real-world innovation in healthcare technologies across the EU.

Our findings show the existence of research output and impact disparities among EU member states in the field of medical informatics based on publication data between 2018 and 2022. Western European nations, particularly Germany, the Netherlands, and Spain dominate in terms of publication volume and cumulative impact factors. Iammarino et al. (2018) emphasize that high-income regions benefit from agglomeration economies, which encourages innovation and reinforces their research dominance, while low-income regions face systemic barriers such as skill shortages and outmigration [28]. This gap highlights persistent regional inequities in research infrastructure, funding availability, and capacity for international collaboration, as is further echoed in studies analyzing the distribution of Horizon 2020 funding, where EU-15 countries received 87 % of the funding compared to only 3 % for EU-13 countries [18]. Additionally, Polašek and Kern (2012) have previously linked research productivity to GDP levels, pointing out that wealthier nations inherently have a greater capacity to support academic pursuits, which, according to our research, seems to be the case with the medical informatics research niche as well [19].

In light of these findings, we suggest that policymakers take a more proactive role in reducing regional disparities by implementing targeted funding programs for EU-13 countries. These programs could include capacity-building grants, infrastructure investments, and dedicated funding calls for interdisciplinary health tech projects led by institutions in underrepresented regions. Additionally, cross-border collaboration could be encouraged through EU-level initiatives that incentivize partnerships between research-intensive institutions in EU-15 countries and emerging centers in EU-13 countries. Policymakers could also consider expanding support for mobility programs, joint PhD supervision, and collaborative data-sharing platforms to facilitate sustained engagement and long-term collaboration beyond individual projects.

Conclusions

Our study aimed to provide an overview of medical informatics research output within the European Union, from 2018 to 2022. We achieved the main goal by conducting a bibliometric analysis of publications categorized by Clarivate under WoS′ medical informatics category, across 47 journals that publish research in English, utilizing tools like InCites and VOSviewer to evaluate research output, collaboration patterns, and impact metrics among EU member states.

One of the central findings of our study is the persistent disparity in research output and impact among EU countries. Nations like Germany, France, and the Netherlands consistently demonstrate high productivity and cumulative impact, while others such as Bulgaria and Croatia contribute fewer publications and show lower engagement. These differences likely reflect underlying variations in research infrastructure, access to funding, and national investment in digital health technologies. Such disparities warrant attention from policymakers and funding agencies aiming to foster a more equitable research landscape within the EU.

Additionally, our study highlights the critical role of international collaboration, with over 66 % of publications resulting from cross-border partnerships. This finding underlines the interconnected nature of medical informatics and the necessity for shared expertise, especially in addressing complex healthcare challenges. However, industry collaborations remain limited (5.53 % of publications), suggesting a potential disconnect between academic research and private-sector application in this field. The observed shift toward open-access publishing also reflects broader trends in scientific communication, supporting transparency and wider dissemination of medical knowledge.

It is important to note that while our findings reveal disparities and patterns in medical informatics research output, they do not establish a direct causal link between publication trends and national commitments to healthcare technology advancement. Future research could explore this relationship more directly by integrating bibliometric data with policy analysis, investment levels, or implementation indicators such as national pilot programs and adoption rates of digital health solutions.

In summary, this study contributes to advancing knowledge in medical informatics by providing a data-driven mapping of the field’s evolution within the EU before and during the COVID-19 pandemic. It sheds light on imbalances in research output, shows the importance of international cooperation, and identifies shortcomings in public-private collaboration patterns. Our findings offer valuable information for researchers, funders, and policymakers seeking to support more inclusive, collaborative, and impactful medical informatics research across Europe.

Supplementary: Figure S1: Top 10 authors by the number of WoS-indexed medical informatics papers through 2018–2022; Figure S2: Top 10 affiliations from the dataset in terms of research output; Figure S3: Co-authorship visualization map; Figure S4: Consistency of journal diversity visualized; Table S1: List of Journals, Abbreviations and Their Impact Factors; Table S2: Articles and Cumulative Impact Factors Per Journal by Country; Table S3: Total Cumulative Impact Factor per Country; Table S4: Annual Growth % and Impact per Country; Table S5: Industry Collaboration Metrics Across EU Countries; Table S6: Correlation Between Article Volume and Journal Diversity by Country; Table S7: Annual Journal Diversity by Country; Table S8: The top 15 journals ranked by the total volume of articles published.