Sage Journals: Discover world-class research

Abstract

Background

Stroke is one of the leading causes of mortality and long-term disability worldwide. In recent years, integrated rehabilitation models that combine virtual reality (VR) technology with standardized exercise therapy have emerged, demonstrating promising potential in improving recovery outcomes.

Objective

This bibliometric review systematically analyzes global literature on virtual reality combined with exercise therapy for stroke rehabilitation to map the knowledge landscape, identify research hotspots and evolutionary trends, and inform future research, clinical practice, and policy.

Methods

Relevant studies on VR combined with exercise therapy for stroke rehabilitation were retrieved from the Web of Science database, covering the period from database inception to 2025. Bibliometric and visualization analyses were conducted using CiteSpace and VOSviewer to assess publication trends, country, institutional contributions, authors and co-cited authors networks, highly cited references, core journals, and the evolution of research hotspots.

Results

A total of 1,687 articles were identified, showing a steady upward publication trend. China ranked first in publication volume, while the United States had the highest total citation count. Researchers such as Calabrò, De Luca, and Naro from the IRCCS Centro Neurolesi in Messina, Italy, made notable contributions, particularly in VR-robotics combined rehabilitation. The Journal of NeuroEngineering and Rehabilitation published the largest number of articles in this field. Keyword burst analysis indicated two distinct phases: before 2021, research primarily focused on conventional rehabilitation methods and clinical trials; after 2021, attention shifted towards the integration of emerging technologies in stroke rehabilitation, including machine learning and immersive VR, reflecting growing scholarly interest in novel rehabilitation strategies.

Conclusions

This study provides a comprehensive bibliometric analysis of VR combined with exercise therapy in stroke rehabilitation, identifying key research hotspots, emerging trends, and existing limitations. The findings could offer theoretical insights and data-driven evidence to support future research and clinical applications in this field.

Keywords

stroke rehabilitation virtual reality exercise therapy bibliometric analysis

1. Introduction

Stroke, as an acute cerebrovascular disease, has become one of the leading causes of death and long-term disability worldwide. According to the latest data from the World Stroke Organization, stroke ranks as the second leading cause of death and the third leading cause of disability globally.¹ The substantial health and socioeconomic burden it imposes exerts a sustained and escalating pressure on healthcare systems worldwide. In 2019, there were over 12.2 million new cases of stroke globally, with more than 100 million survivors. In terms of mortality, approximately 6.5 million people die from stroke each year, accounting for 11.6% of all deaths in 2019.² More alarmingly, the impact of stroke extends far beyond mortality, resulting in the loss of over 143 million disability-adjusted life years (DALYs) annually. This metric quantifies the years of healthy life lost due to premature death and disability, underscoring the profound and lasting disruption stroke inflicts on human well-being.² Economically, the global cost of stroke exceeds USD 721 billion, equivalent to approximately 0.66% of the world Gross Domestic Product, placing a significant financial strain on societies.¹

Stroke survivors frequently experience severe sequelae, particularly motor impairments. It is estimated that up to 80% of patients exhibit varying degrees of limb motor dysfunction,³ manifesting as muscle weakness and impaired coordination. These physical impairments, affecting the trunk as well as the upper and lower limbs, severely compromise capacity of patients for daily activities and self-care, thereby substantially diminishing their quality of life.⁴

Conventional rehabilitation approaches, such as physical therapy and occupational therapy, remain the cornerstone of functional recovery, yet their effectiveness is often hindered by intrinsic limitations. Neuroscientific evidence indicates that neural functional reorganization depends on high-intensity, high-repetition, task-oriented training.⁵ Although traditional rehabilitation methods can enhance muscle strength and mobility, they are characterized by significant constraints: they are resource-intensive (requiring specialized equipment and therapists), time-consuming,^6,7 and monotonous, which frequently leads to reduced patient motivation and adherence, ultimately limiting long-term outcomes.⁸ These limitations often result in a plateau in functional improvement after a certain rehabilitation period. Therefore, there is an urgent need for novel rehabilitation strategies capable of overcoming the constraints of traditional approaches.

Against this backdrop, virtual reality (VR) technology has been introduced into the field of stroke rehabilitation as an emerging rehabilitation method, aiming to promote the reorganization of neural motor pathways and alleviate motor dysfunction.⁹ VR is typically defined as an interactive computer simulation system that can track the movements of users in real time and create an immersive experience through multimodal feedback such as visual and auditory cues.³ This characteristic not only significantly enhances intrinsic motivation and engagement, enabling patients to unconsciously complete training volumes far exceeding those of traditional therapies, but also provides immediate, quantifiable motor feedback, and dynamically adjusts task difficulty based on patient ability, thereby optimizing motor learning efficiency.¹⁰ More importantly, VR is not only a tool to enhance training motivation but also an effective “driver” for stimulating neural plasticity. Research indicates that when patients engage in high-intensity, repetitive training in a virtual environment and receive rich multisensory feedback, the neural networks responsible for motor planning and execution in the brain are continuously and efficiently activated, thereby promoting the reorganization and functional compensation of damaged neural pathways.^11,12 However, the standalone application of VR still has limitations, such as the lack of individualized exercise prescriptions and insufficient physiological feedback. In recent years, a new rehabilitation model combining VR technology with standardized exercise therapy has emerged and shown potential for improving rehabilitation outcomes.

Clinical research evidence indicates that compared to conventional rehabilitation alone, combining VR training can significantly enhance therapeutic efficacy. For example, a systematic review found that adding VR training to conventional occupational therapy significantly improved upper limb motor function, functional independence, and activities of daily living in stroke patients.¹³ Another study demonstrated that VR also has positive effects on the recovery of upper and lower limb function, gait, and balance, with more significant efficacy when combined with traditional therapy. Additionally, limited evidence suggests that VR may also have positive effects on patients with cognitive impairments.¹⁴ However, the rapid development in this field is accompanied by high heterogeneity among studies: significant differences exist in VR system types, intervention protocols, and participant characteristics across studies, leading to inconsistent results and posing challenges for clinical application and the determination of future research directions.

Bibliometrics provides an objective and quantitative analytical tool for this purpose. By systematically analyzing author, institutional, keyword, and citation information from a large-scale academic literature corpus, knowledge maps can be constructed to identify research hotspots and emerging trends.¹⁵ Based on this, the study aims to apply bibliometric methods to systematically review the global literature on the application of VR combined with exercise therapy in stroke rehabilitation, with the goal of mapping the knowledge landscape in this field, identifying research hotspots and evolutionary trends, and providing theoretical reference for future scientific research, clinical practice, and policy-making.

2. Materials and methods

2.1. Search strategy

The data for this study were sourced from the Web of Science database. As an internationally recognized authoritative academic database, Web of Science has become one of the most favored data sources in the field of bibliometric research due to the high quality and broad coverage of its included literature.^16–18 Additionally, articles related to the study topic were retrieved from the Science Citation Index Expanded (SCIE) and Social Sciences Citation Index (SSCI) databases within the Web of Science Core Collection (WOSCC) from its inception to June 15, 2025. SCIE and SSCI are sub-databases of WOSCC, comprising journals in the field of basic scientific research globally, and cover neuroscience and medical research relevant to the study topic. To obtain literature explicitly using relevant terminology, the search query was set as follows: “(((((((((((((((((TS=(Strokes)) OR TS=(Cerebrovascular Accident)) OR TS=(Cerebrovascular Accidents)) OR TS=(Cerebral Stroke)) OR TS=(Cerebral Strokes)) OR TS=(Cerebrovascular Apoplexy)) OR TS=(Brain Vascular Accident)) OR TS=(Brain Vascular Accidents)) OR TS=(Cerebrovascular Stroke)) OR TS=(Cerebrovascular Strokes)) OR TS=(Apoplexy)) OR TS=(CVA (Cerebrovascular Accident))) OR TS=(CVAs (Cerebrovascular Accident))) OR TS=(Acute Stroke)) OR TS=(Acute Strokes)) OR TS=(Acute Cerebrovascular Accident)) OR TS=(Acute Cerebrovascular Accidents)) AND ((((((((((((((TS=(Exercises)) OR TS=(Exercise)) OR TS=(Physical Exercise)) OR TS=(Physical Exercises)) OR TS=(Aerobic Exercise)) OR TS=(Aerobic Exercises)) OR TS=(Isometric Exercises)) OR TS=(Isometric Exercise)) OR TS=(Acute Exercise)) OR TS=(Acute Exercises)) OR TS=(Training)) OR TS=(Trainings)) OR TS=(Physical Activity)) OR TS=(Physical Activities)) AND (((((((((((TS=(Virtual Reality)) OR TS=(Educational Virtual Realities)) OR TS=(Educational Virtual Reality)) OR TS=(Instructional Virtual Realities)) OR TS=(Instructional Virtual Reality)) OR TS=(VR)) OR TS=(Virtual Environment)) OR TS=(Immersive Virtual Reality)) OR TS=(Non-Immersive Virtual Reality)) OR TS=(Simulation)) OR TS=(Serious Games))”. Additionally, in this study, “exercise therapy” refers to exercise-based stroke rehabilitation interventions, including structured exercise, physical training, and task-oriented forms of physical activity. Similarly, “virtual reality” denotes computer-generated, interactive rehabilitation environments that simulate or augment real-world tasks. In the bibliometric analysis, this category encompasses immersive and non-immersive virtual reality, virtual environments, simulation-based systems, and serious games.

2.2. Inclusion and exclusion criteria

All included studies were peer-reviewed publications. An initial screening was conducted by reviewing the titles and abstracts, followed by a full-text assessment to determine eligibility based on the inclusion criteria. The exclusion criteria were as follows: (1) studies that did not involve VR combined with exercise therapy; (2) studies focusing on diseases unrelated to stroke; (3) publications in the form of letters, commentaries, conference abstracts, or retracted articles; (4) articles not published in English.

2.3. Bibliometric and visual analysis

Articles retrieved from the Web of Science database, including full records and cited references, were exported in plain text format with the filename “download_X.txt.” After data deduplication and merging, the plain text files were imported into VOSviewer (version 1.6.18) and CiteSpace (version 6.3.R1) for subsequent analysis and visualization.

2.4. Statistical analysis

In this study, statistical indicators include publication year distribution, country and institutional contributions, author and co-cited authors relationships, journal analysis, reference analysis, and keyword analysis. For each category of indicator, first conducted frequency-based analysis, then utilized visual maps to illustrate the strength of relationships and developmental trends among different nodes. Interrelationships between indicators were primarily reflected through node size and line thickness: node size represented the frequency of occurrence or citation frequency of that indicator, while line thickness indicated the strength of co-occurrence or co-citation relationships.

2.5. Ethical considerations

This study falls within the scope of bibliometric research. As it is classified as a bibliometric analysis, formal ethical approval was not required. Bibliometric studies involve quantitative analyses of publicly accessible scholarly outputs, such as journal articles, patents, and conference proceedings, and do not entail: (1) direct interaction with human participants; (2) collection of personal, sensitive, or identifiable information; or (3) intervention in human behavior or health. According to international guidelines (e.g., the Declaration of Helsinki) and institutional policies, studies of this nature are generally exempt from ethical review, as they pose no risk to privacy, autonomy, or well-being.

3. Result

3.1. Trends in the number of publications

A total of 1,934 records were retrieved from the WOSCC database. After excluding non-academic publications, non-English literature, and studies outside the scope of the research objectives, 1,687 articles were finally included (Figure 1 presents the literature search and screening flowchart). Among these, 1,406 were original research articles and 281 were review papers.

Figure 1.

The literature search and screening flowchart.

Although the search period covered publications from the database’s inception to June 15, 2025, Figure 2 shows that this topic first appeared in 2011. Overall, research on VR combined with exercise therapy in stroke rehabilitation has demonstrated a steadily increasing publication trend, with a particularly rapid growth from 2019 to 2022. The number of publications peaked in 2022, reaching 200 articles. However, the annual output declined from 2023 to 2025, especially in 2025, likely because the year was incomplete at the time of the search, resulting in some articles not yet being published or indexed.

Figure 2.

Number of articles published.

3.2. Distribution of countries

Statistical analysis revealed that research on VR combined with exercise therapy in stroke rehabilitation has been published across 85 countries. For clarity of presentation, the top 34 countries were visualized (Figure 3). The top 10 countries in terms of publication are listed in Table 1. The results show that China leads with 354 publications, followed by the United States with 330. The number of publications from other countries drops significantly thereafter, with Italy at 153 and South Korea at 136. In terms of total citations, the United States leads with 10,999 citations, followed by China (5,331 citations) and Italy (4,666 citations). Notably, in terms of citations per paper, Australia leads with 65.62 citations, followed by the England (49.29 citations) and the United States (33.33 citations).

Figure 3.

Country distribution of VR combined exercise therapy in stroke rehabilitation.

Table 1.

Top 10 countries/regions contributing to publications on VR and exercise in stroke.

Rank	Country/region	Counts	Citations	Citations per paper
1	China	354	5331	15.06
2	USA	330	10999	33.33
3	Italy	153	4666	30.50
4	South korea	136	3630	26.69
5	England	106	5225	49.29
6	Canada	95	2786	29.33
7	Germany	88	2063	23.44
8	Spain	88	2790	31.70
9	Switzerland	79	2253	28.52
10	Australia	74	4856	65.62

In the visualized network, different colors represent different clusters, and the lines indicate cooperative relationships between countries, with thicker lines indicating closer cooperative relationships. The results show that the United States and China have the closest cooperative relationship, forming an important hub in the academic cooperation network in this field.

3.3. Analysis of institutions

This dataset involves 2,710 institutions. Table 2 lists the top 10 institutions with the most significant contributions. Among them, IRCCS Centro Neurolesi Bonino Pulejo in Italy published the most papers, with 27, followed by McGill University in Canada and Sahmyook University in South Korea, both with 26 papers. It is worth noting that McGill University also had the most citations with 1,364, averaging 52.46 citations per article. In addition, among the top 10 institutions, three institutions in Switzerland were included, including 21 articles from the Swiss Federal Institute of Technology, 18 articles from the University of Zurich, and 15 articles from the University of Bern. Two institutions in China were also included, with 16 articles from Hong Kong Polytechnic University and 15 articles from the Chinese Academy of Sciences. Figure 4 illustrates the inter-institutional collaboration network for VR-combined exercise therapy in stroke treatment. Among these institutions, four institutions-IRCCS Centro Neurolesi Bonino Pulejo, McGill University, Sahmyook University, and the Swiss Federal Institute of Technology-form four representative networks. The connections between these networks highlight a highly cohesive and dynamic collaborative framework. These institutions collectively form a closely collaborative network to further advance the development of this research field.

Table 2.

Top 10 institutions contributing to publications on VR and exercise in stroke.

Rank	Institution	Country	Count	Citations
1	IRCCS Centro Neurolesi Bonino Pulejo	Italy	27	1039
2	McGill University	Canada	26	1364
3	Sahmyook University	South Korea	26	927
4	Swiss Federal Institute of Technology	Switzerland	21	763
5	Tel Aviv University	Israel	20	811
6	University of Zurich	Switzerland	18	730
7	Northeastern University	USA	18	476
8	Hong Kong Polytechnic University	China	16	361
9	University of Bern	Switzerland	15	504
10	Chinese Academy of Sciences	China	15	261

Figure 4.

Collaborative networks between institutions.

3.4. Analysis of authors and co-cited authors

Based on the author collaboration network (Figure 5(a)) and author co-cited network (Figure 5(b)), the global research landscape of VR combined with exercise therapy for stroke rehabilitation was analyzed. Table 3 presents the top 10 authors ranked by both publication output and co-cited counts. The three most prolific authors were Calabro, Rocco Salvatore (35 publications), De Luca, Rosaria (24 publications), and Naro, Antonino (18 publications). In terms of co-cited, Laver, K.E. ranked first with 365 citations, followed by Saposnik, G. (339 citations) and Kwakkel, G. (317 citations).

Figure 5.

(a) Active authors. (b) Active co-cited authors.

Table 3.

The top 10 authors and co-cited authors.

Rank	Author	Counts	Co-cited author	Counts
1	Calabro, Rocco Salvatore	35	Laver, Ke	365
2	De Luca, Rosaria	24	Saposnik, G	339
3	Naro, Antonino	18	Kwakkel, G	317
4	Manuli, Alfredo	14	Langhorne, P	203
5	Maggio, Maria Grazia	14	Levin, Mf	191
6	Qiu, Qinyin	12	Holden, Mk	168
7	Alcaniz, Mariano	11	Piron, I	153
8	Llorens, Roberto	11	Wolf, Sl	151
9	Riener, Robert	11	Feigin, Vl	147
10	Bramanti, Alessia	10	Duncan, Pw	142

3.5. Analysis of journals

In this study, 1,687 articles were published in 553 academic journals. Table 4 shows the top 10 journals, which accounted for 21.28% of the publications and contributed 359 articles. Among them, the Journal of Neuroengineering and Rehabilitation published 79 articles, which were cited 5,115 times; Frontiers in Neurology published 42 related articles, which were cited 587 times; IEEE Transactions on Neural Systems and Rehabilitation Engineering published 32 articles, which were cited a total of 813 times. Additionally, among the top 10 journals, 5 belong to Q1. The highest IF over the past 5 years was 6, primarily related to neuroengineering and rehabilitation. The influence of academic publications in a research field is quantified by the number of co-citations. Co-citation analysis was conducted using VOSviewer. As shown in Figure 6, the co-citation network consists of 8 clusters. The size of the nodes represents the number of co-citations, and the lines between nodes indicate co-citation relationships. Among these, the red clusters primarily involve articles related to rehabilitation engineering, while the blue clusters are mostly associated with clinical rehabilitation.

Table 4.

The top 10 journals that published articles.

Rank	Journal title	Country	Counts	Citations	JCR	IF_5
1	Journal of Neuroengineering and Rehabilitation	England	93	5115	Q1	6
2	Frontiers in Neurology	Switzerland	42	587	Q1	3.3
3	IEEE Transactions On Neural Systems and Rehabilitation Engineering	United States	32	813	Q1	5.7
4	Neurorehabilitation	Ireland	29	779	Q2	2.3
5	Jmir Serious Games	Canada	28	484	Q1	4.8
6	Frontiers in Human Neuroscience	Switzerland	28	1486	Q2	3.3
7	Sensors	Switzerland	28	386	Q2	3.7
8	Archives of Physical Medicine and Rehabilitation	United States	26	1483	Q1	4.1
9	Disability and Rehabilitation	England	24	776	Q2	2.8
10	Healthcare	Switzerland	24	270	Q2	2.8

Note. JCR: Journal Citation Reports; IF: impact factor; Q: Quartile; IEEE: Institute of Electrical and Electronics Engineers.

Figure 6.

Co-citation analysis clustering map of journal.

3.6. Cited reference analysis

The number of citations for an article reflects the level of interest researchers have in the topic. The more citations an article receives, the more important it is considered to be. Figure 7 shows a visual analysis of the references. Nodes represent the authors and years of the articles, with node size indicating citation frequency. Lines between nodes indicate that two articles were cited together, with line thickness indicating the strength of the relationship. Table 5 presents the top 10 most cited articles. Among them, Laver K (2011) has the highest number of citations, with 191 citations; Saposnik G (2011) has 125 citations, involving original research and meta-analyses on VR in stroke rehabilitation.

Figure 7.

Co-citation analysis of cited references.

Table 5.

The top 10 cited references.

Rank	First author	Title	Year	Citations	Journal
1	Laver Ke	Virtual reality for stroke rehabilitation	2011	191	cochrane db syst rev
2	Saposnik G	Virtual Reality in Stroke Rehabilitation: A Meta-Analysis and Implications for Clinicians	2011	125	stroke
3	Fuglmeyer Ar	The post-stroke hemiplegic patient. 1. a method for evaluation of physical performance	1975	122	scand j rehabil med
4	Saposnik G	Effectiveness of virtual reality using Wii gaming technology in stroke rehabilitation: a pilot randomized clinical trial and proof of principle	2010	114	stroke
5	Langhorne P	Motor recovery after stroke: a systematic review	2009	95	lancet neurol
6	Henderson A	Virtual reality in stroke rehabilitation: a systematic review of its effectiveness for upper limb motor recovery	2007	92	top stroke rehabil
7	Holden Mk	Virtual environments for motor rehabilitation: review	2005	88	cyberpsychol behav
8	Lohse Kr	Virtual Reality Therapy for Adults Post-Stroke: A Systematic Review and Meta-Analysis Exploring Virtual Environments and Commercial Games in Therapy	2014	86	plos one
9	Kleim Ja	Principles of experience-dependent neural plasticity: implications for rehabilitation after brain damage	2008	83	j speech lang hear r
10	Langhorne P	Stroke rehabilitation	2011	77	lancet

3.7. Analysis of keywords

As integral article labels, keywords not only enable readers to rapidly grasp the main theme and core content of a study, but also serve as a crucial basis for identifying research hotspots and emerging trends within a field.¹⁹ Analyzing keywords thus facilitates a more comprehensive understanding of the developmental trajectory of the relevant discipline. In this study, VOSviewer was employed to conduct co-occurrence and clustering analyses of the keywords of article (Figure 8). The results revealed that the keywords were categorized into four clusters, among which the red, blue, and green clusters were the most prominent. Specifically, the red cluster primarily pertains to the anatomical regions targeted by the application of VR in stroke rehabilitation; the green cluster relates to the characteristics of the patient population and the evaluation indicators used in stroke rehabilitation; and the blue cluster focuses on the methodologies employed in stroke rehabilitation. Table 6 lists the top 10 important keywords. Stroke was the most frequently used term, appearing 897 times, followed by virtual reality (828 times) and rehabilitation (688 times).

Figure 8.

Keywords co-word network and clustering.

Table 6.

The top 10 keywords.

Rank	Keyword	Counts	Rank	Keyword	Counts
1	stroke	897	6	balance	214
2	virtual reality	828	7	exercise	174
3	rehabilitation	688	8	gait	167
4	recovery	258	9	reliability	162
5	therapy	217	10	performance	152

With the continuous development of this research field, its hotspots have evolved over time. Solely relying on keyword clustering analysis presents certain limitations, as it cannot reveal temporal variations in keyword usage. Keyword bursts refer to a marked increase in the frequency of a given keyword within a specific time period. Analyzing keyword bursts provides valuable insights into the developmental trajectory of VR combined with exercise therapy in stroke rehabilitation, as well as the identification of stage-specific research focuses. In the present study, CiteSpace was employed to perform a keyword burst analysis, and the top 20 burst keywords are presented in Figure 9. Among them, “ambulation” and “postural balance” exhibited the longest burst durations, lasting nearly six years. Furthermore, results indicated that before 2021, burst keywords were predominantly associated with conventional rehabilitation methods and clinical trials. After 2021, however, research attention shifted toward the application of emerging technologies in stroke rehabilitation, including machine learning, immersive VR, and game, suggesting a growing scholarly emphasis on the exploration and adoption of novel rehabilitation techniques.

Figure 9.

Keywords with bursts.

4. Discussion

CiteSpace and VOSviewer software were utilized in this study to conduct a systematic analysis of the current state of research on VR combined with exercise therapy in stroke rehabilitation. From the establishment of the database to 2025, a total of 1,687 relevant literature were retrieved, including 1,406 original studies (83.34%) and 281 review articles. Contemporary research predominantly focuses on clinical trials to validate the therapeutic efficacy of this intervention in stroke rehabilitation. Over the past 15 years, the number of publications in this field has shown an overall upward trend. The earliest application of VR in stroke rehabilitation dates back to 2001, when D. Jack et al. introduced it into hand function rehabilitation training for stroke patients. The results showed that most hand parameters improved significantly during training, and patients reported positive subjective evaluations,²⁰ marking the beginning of VR research in stroke rehabilitation. Notably, research on VR combined with exercise therapy was not formally published until 2011. According to the results in Figure 2, the development of this topic was relatively slow from 2011 to 2018, but entered a rapid growth phase starting in 2019. This may be closely related to the advent of the AI era, the significant reduction in VR hardware and software development costs, and the improved standardization of exercise therapy combination models. By 2022, the number of related publications reached 200; although there was a slight decline in 2023 and 2024, the numbers remained higher than in previous years, indicating that this field will maintain a high level of interest in the foreseeable future.

Regarding geographical distribution, although related research spans 85 countries worldwide, underscoring the widespread global interest in this interdisciplinary field, there remains a significant disparity in research productivity and academic impact. This pattern not only reflects differences in research and development investment levels among countries but also reveals the divergent paths and priorities adopted by different nations in addressing public health challenges, formulating science and technology development strategies, and building innovation systems. The findings of this study reveal that the field presents a multipolar landscape spearheaded by China and the United States, with active participation from multiple countries. As shown in Table 1, China ranks first with 354 papers, followed closely by the United States with 330 papers, with the two countries constitute the primary echelon of research output in this field. The concentration of research outcomes within a select few countries suggests that the field has relatively high technical and resource barriers to entry. Research combining VR with exercise therapy not only requires advanced VR technology and equipment but also the deep integration of multidisciplinary fields such as neuroscience, rehabilitation medicine, biomechanics, and computer science, as well as sufficient research funding and clinical trial resources.

China leads the world in the number of publications in this field, a phenomenon driven by profound social and policy factors. Its substantial research output can be understood as a response to a severe public health crisis, strongly supported by national strategic initiatives. Specifically, the significant clinical demand serves as the fundamental driving force behind research development. Currently, China bears the highest global burden of stroke.²¹ Stroke has become the leading cause of premature death and DALYs loss among the Chinese population.^22,23 As the rapid acceleration of population aging, this burden is expected to continue to worsen. Additionally, the national strategic planning provides strong “top-down” policy support and resource allocation for this demand. The Chinese governments “Healthy China 2030” planning outline elevates national health to a national strategic level, proposing to strengthen health science and technology innovation to drive improvements in healthcare quality and equity through technological progress. Under this macro policy framework, the prevention, treatment, and rehabilitation of major chronic diseases such as stroke have become key areas of development. In contrast, the United States leverages its mature neuro-rehabilitation research system, long-term stable biomedical funding mechanisms, interdisciplinary collaboration platforms, and robust methodological capabilities to gain significant advantages in research design, mechanism exploration, hardware equipment, standard development, and high-impact output. Consequently, China excels in “clinical demand-driven approaches” and “large-scale application validation,” while the United States holds a central position in “theoretical innovation, technological integration, and high-quality evidence production.” This structural complementarity enables both nations to more readily establish hub status within international collaborative networks.

From the perspective of academic influence (measured by citation data), the data shows that scholarly impact is not strictly linearly correlated with the quantity of research output but reflects the unique research ecosystems and strategic priorities of different countries. The United States leads significantly in total citation frequency with 10,999, far exceeding the 5,331 of China, reflecting its strong influence. The United States leadership is rooted in its long-term, large-scale, and sustained investment in biomedical research. As the largest public biomedical research funding agency globally, the National Institutes of Health allocates tens of billions of dollars annually, laying a solid foundation for scientific breakthroughs and innovative therapies.²⁴ Meanwhile, Australia and England demonstrate a model of high-impact efficiency. Australia leads the world with an average citation frequency of 65.62 per paper, followed closely by the England at 49.29, both significantly higher than the United States (33.33) and China (15.06). This outstanding performance stems from the meticulous design of national research strategies. For example, the Australian Medical Research and Innovation Strategy, guided by the Medical Research Future Fund (MRFF), aims to promote research translation and impact, as well as health and economic benefits through innovative research. This “impact-oriented” funding approach tends to support research projects with significant transformative potential, fostering a culture of “quality over quantity” within the research community.

In the field of stroke rehabilitation, research on VR combined with exercise therapy has been primarily driven by a small number of highly productive authors and their teams. Bibliometric analysis reveals that Calabrò, De Luca, Naro, and others at the Messina IRCCS Brain Injury Center in Italy exhibit high publication activity, suggesting a degree of team concentration in knowledge production within this domain. Further analysis of institutional distribution reveals that Italy research teams focus more on VR combined with robotic training and neurorehabilitation technologies,²⁵ while teams from Canada and Australia concentrate more on randomized controlled trial designs, rehabilitation assessment, and evidence-based reviews.^26–28 These findings indicate that distinct research specializations have emerged among different countries and institutions in this field, with these variations collectively shaping the current structure and developmental trajectory of research topics.

A bibliometric analysis shows that research on combined VR and exercise therapy in stroke rehabilitation is concentrated in a small number of institutions and author cohorts. Statistically, Calabrò (35 publications) is the most prolific author, with other high-output authors including De Luca and Naro, who are affiliated with Italian institutions. Co-author analysis reveals that top authors form several closely collaborating clusters, characterized by frequent intra-institutional and domestic partnerships. For example, members of the Messina IRCCS team (e.g., Calabrò, De Luca, Naro) frequently collaborate as co-authors; while the McGill and Toronto teams (e.g., Levin, Saposnik) focus on Canadian projects. Based on the above, the collaborative networks formed by authors and their institutions significantly shape both research themes and analytical depth: Italian teams primarily focus on VR-assisted robotic training and cognitive rehabilitation, while Canadian teams tend toward large-sample RCTs and systematic reviews. Collaboration networks are highly concentrated internally but relatively limited internationally. Therefore, future research should encourage broader collaboration among teams from different countries and disciplinary backgrounds to enhance the external validity of evidence and reduce research bias arising from single-team dominance.

Co-occurrence analysis of keywords shows that terms such as “stroke,” “virtual reality,” “rehabilitation,” “recovery,” and “therapy” appear most frequently, reflecting the core research themes of the field. The results of keyword clustering divide the keywords into several thematic clusters, with the cluster related to rehabilitation methods (blue) being the most prominent. CiteSpace keyword burst analysis indicates that from 2011 to 2021, research hotspots were concentrated on traditional rehabilitation methods and related clinical assessments; however, after 2021, emerging technologies such as “immersive virtual reality,” “machine learning,” and “game” began to appear frequently. This trend indicates that research focus in this field is gradually shifting from traditional training models toward immersive interaction, gamified interventions, and data-driven intelligent rehabilitation.^29–34 This evolutionary pathway demonstrates that current research no longer limits itself to validating single intervention formats, but instead places greater emphasis on digital technology integration, personalized rehabilitation, and multimodal intervention strategies.

Specifically, the emerging technology clusters that emerged after 2021 essentially represent a response to the limitations of traditional VR applications. While early VR rehabilitation research demonstrated certain therapeutic effects, its clinical adoption has long been constrained by several issues. First, traditional VR systems primarily feature pre-set scenes and fixed tasks, resulting in limited motion engagement, reduced realism in task scenarios, and inadequate adaptability of training content. This leads to insufficient patient immersion and diminished motivation for training.³⁵ Second, early VR placed greater emphasis on visual feedback and task completion. While this enhanced training engagement, it provided insufficient support for upper limb fine motor skills, postural control, movement quality monitoring, and functional transfer.³⁶ Third, many traditional VR interventions rely on specialized venues and high costs, and lack sufficient capabilities for continuous training and remote monitoring, thus limiting their value in long-term rehabilitation management.³⁶ In addition, traditional systems are mostly based on preset uniform programs, which cannot be dynamically adjusted according to the patient functional level and rehabilitation progress, making it difficult to formulate precise individualized plans.

Against this background, the emergence of emerging directions such as immersive VR, robotics, gamified interventions, and machine learning actually corresponds to clinical gaps at different levels. Immersive VR enhances presence and environmental interaction through head-mounted displays and multimodal feedback, helping to improve patient attention engagement, motor participation, and task scenario realism. This addresses the shortcomings of traditional screen-based VR in terms of insufficient immersion and training motivation decay.³⁷ The integration of robotics with VR not only delivers more stable, repeatable high-intensity training but also improves dose control and movement quality management through force feedback, path constraint, and motion assistance. These features compensate for the limitations of single VR in physical guidance and fine kinematic adjustments.³⁸ The value of gamified intervention lies not only in increasing enjoyment, but also in its convenience, reward mechanisms, instant feedback, task upgrades, and challenge structures, which enable long-term rehabilitation management and enhance patient motivation for sustained training.^39,40 A systematic review of 169 studies conducted between 2019 and 2023 found that various gamified rehabilitation devices, not only promote recovery of motor and cognitive functions but also significantly improve patient mood, social engagement, and satisfaction.⁴¹

The machine learning and artificial intelligence enables rehabilitation systems to transition from uniform program-driven approaches to data-driven individualized interventions. For instance, by analyzing motion data, sensor signals, and training performance, these systems can identify patient capability thresholds, dynamically adjust task difficulty, predict recovery trajectories, and assist in developing more precise and personalized rehabilitation plans.^42,43 A review of 704 publications revealed that artificial intelligence applications have evolved beyond proof-of-concept stages to encompass diverse technologies, including supervised learning, neural networks, and natural language processing. In upper limb functional assessment, sensors and deep learning models enable detailed analysis of patient movements, thereby supporting the development of personalized rehabilitation plans.⁴⁴

The development of emerging technologies is not merely a simple overlay on traditional VR, but rather an evolution toward greater personalization, intelligence, compliance, and real-world functional recovery. From a clinical translation perspective, this technological evolution holds significant practical implications. A core challenge in stroke rehabilitation lies in the high heterogeneity of patient recovery processes. Factors such as lesion location, disease stage, cognitive status, and family support levels significantly influence training responses, making standardized rehabilitation protocols often inadequate for individual needs. The emergence of emerging technology clusters indicates researchers are attempting to establish more adaptive rehabilitation systems. Their goal extends beyond improving short-term scale scores to enhancing patient movement execution, participation capacity, and long-term self-management abilities in real-life environments. Particularly in settings with uneven rehabilitation resource distribution, shortages of specialized therapists, and challenges in long-term follow-up, digital rehabilitation technologies, with their potential for remote monitoring, data feedback, and personalized adjustments, may offer new implementation pathways extending stroke rehabilitation from hospitals to communities and homes. Therefore, the value of these emerging research directions lies not only in enhanced therapeutic efficacy but also in their potential to reshape the organizational structure and delivery models of stroke rehabilitation services.

It is worth further emphasizing that the clinical translation of machine learning and artificial intelligence in stroke rehabilitation faces a critical bottleneck: insufficient explainability. When artificial intelligence models are applied to assess movement quality, adaptively adjust task difficulty, and predict treatment efficacy during VR training, clinicians require not only “what conclusions the model provides” but also an understanding of “why the model reached those conclusions.” Otherwise, black-box outputs struggle to establish clinical trust and limit their application in real-world scenarios. Consequently, recent research has shifted from solely pursuing predictive performance toward incorporating explainable AI (XAI) frameworks. This enables model outputs to provide transparent, auditable evidence chains aligned with clinical logic. For instance, studies have combined ensemble learning with feature optimization and statistical significance analysis, further integrating interpretative methods like SHAP and LIME to identify key features and enhance decision transparency, achieving explainable high-performance prediction.⁴⁵ Similarly, a combined strategy of “hybrid modeling and multiple interpretability tools (e.g., SHAP, LIME, or ICE)” has also been employed to enhance the credibility and comprehensibility of health-related predictive tasks.^46–48

Meanwhile, explainability approaches are expanding from traditional data models to deep learning. Visualization techniques like Grad-CAM or Grad-CAM++ can highlight key regions or patterns relied upon by deep model decisions, providing visual evidence for complex models.⁴⁹ This trend holds significant implications for VR stroke rehabilitation: As visual tracking, pose estimation, and high-dimensional kinematic data increasingly apply to movement therapy assessment and dosage regulation, understanding which action segments or movement features models focus on will become crucial for therapists to validate, correct, and adjust intervention plans. In the future, artificial intelligence-based VR movement therapy hybrid models are likely to extend further into community and home settings. Training data will exhibit characteristics such as continuous collection, remote transmission, and cross-scenario usage, thereby highlighting privacy protection requirements. Research integrating privacy protection mechanisms with explainable prediction frameworks offers a viable pathway to balance privacy and transparency in health monitoring and risk assessment.⁵⁰ Overall, incorporating XAI holds promise for advancing this field from an accurate yet opaque algorithmic paradigm toward intelligent rehabilitation systems with enhanced clinical credibility, auditability, and deployability.

In summary, the current literature emphasizes the research trend of using technological innovation to enhance the effectiveness of exercise therapy, offering novel paradigms and translational potential for clinical practice in stroke rehabilitation. Despite employing a relatively comprehensive methodology, this study has certain limitations. First, constrained by the strict requirements of bibliometric visualization tools regarding data formats and citation links, we conducted statistical analysis only on literature from the WOSCC database. While WOSCC holds high authority in bibliometrics, reliance on a single database may overlook clinical trial literature published exclusively in specialized rehabilitation or clinical journals indexed solely in PubMed or Scopus. Consequently, this limitation may introduce bias into “research hotspots” and “keyword emergence” analyses, potentially skewing existing trends toward publications related to engineering techniques and methodology. This could, to some extent, underestimate research dynamics concerning clinical efficacy. Future studies may consider supplementary analyses in specialized clinical databases to cross-validate these trends. Second, the inclusion criteria were limited to English-language literature, which may have resulted in some degree of publication bias. Third, our search strategy only covered peer-reviewed articles and reviews, excluding academic monographs, conference proceedings, and gray literature from the analysis, which may have omitted some valuable information. Future research should focus more on literature from multiple databases and languages to provide a more comprehensive and accurate representation of the current state of research in this field.

5. Conclusion

This study utilized CiteSpace and VOSviewer software to conduct a comprehensive analysis and visualization of the application of VR-combined exercise therapy in the field of stroke rehabilitation from 2011 to 2025. A comprehensive summary was conducted by focusing on aspects such as the countries of publication, institutions, number of papers, authors and co-cited authors, journals, keywords, and high citation rates. The data showed that the number of papers has been increasing year by year, with China and the United States leading in terms of the number of papers published. It is worth noting that most of the included studies were published in journals with relatively low IF, indicating that future research needs to further improve innovation and comprehensive standardized scientific experiments. In addition, through the analysis of literature on VR combined exercise therapy in the field of stroke rehabilitation, this study highlights the latest research trends and future research directions. These insights will help researchers gain a deeper understanding of the development trends and hot spots in this field, as well as potential future research directions.

Footnotes

ORCID iD

Yuzhuo Dan

Consent to participate

This article does not contain any studies with human or animal participants.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by The Special Project of Science and Technology Research of Sichuan Provincial Administration of Traditional Chinese Medicine (No. 25MSZX269); Xinglin Scholar Research Premotion Project of Chengdu University of TCM (No. CCCX2024008).

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Appendix

References

Feigin

Brainin

Norrving

, et al. World Stroke Organization (WSO): Global Stroke Fact Sheet 2022. International journal of stroke : official journal of the International Stroke Society 2022; 17: 18–29. 2022/01/07. https://doi.org/10.1177/17474930211065917

Collaborators

. Global, regional, and national burden of stroke and its risk factors, 1990-2019: a systematic analysis for the Global Burden of Disease Study 2019. The Lancet Neurology 2021; 20: 795–820. 2021/09/07. https://doi.org/10.1016/s1474-4422(21)00252-0

Demeco

Zola

Frizziero

, et al. Immersive Virtual Reality in Post-Stroke Rehabilitation: A Systematic Review. Sensors (Basel, Switzerland) 2023; 23: 1–19. 2023/02/12. https://doi.org/10.3390/s23031712

Teasell

Rice

Richardson

, et al. The next revolution in stroke care. Expert review of neurotherapeutics 2014; 14: 1307–1314. 2014/10/22. https://doi.org/10.1586/14737175.2014.968130

Guo

Q-F

, et al. Virtual reality for neurorehabilitation: A bibliometric analysis of knowledge structure and theme trends. Front Public Health 2022; 10: 1–13. 2022/11/10. https://doi.org/10.3389/fpubh.2022.1042618

Norouzi-Gheidari

Hernandez

Archambault

, et al. Feasibility, Safety and Efficacy of a Virtual Reality Exergame System to Supplement Upper Extremity Rehabilitation Post-Stroke: A Pilot Rand omized Clinical Trial and Proof of Principle. IJERPH 2019; 17: 1–11. 2019/12/23. https://doi.org/10.3390/ijerph17010113

Patel

Fluet

Qiu

, et al. Intensive virtual reality and robotic based upper limb training compar ed to usual care, and associated cortical reorganization, in the acute and early sub-acute periods post-stroke: a feasibility study. J NeuroEngineering Rehabil 2019; 16: 1–12. 2019/07/17. https://doi.org/10.1186/s12984-019-0563-3

Huang

Chen

, et al. Evaluating the effect and mechanism of upper limb motor function recov ery induced by immersive virtual-reality-based rehabilitation for suba cute stroke subjects: study protocol for a randomized controlled trial. Trials 2019; 20: 1–9. 2019/02/06. https://doi.org/10.1186/s13063-019-3177-y

You

Jang

Kim

Y-H

, et al. Virtual Reality–Induced Cortical Reorganization and Associated Locomot or Recovery in Chronic Stroke. Stroke 2005; 36: 1166–1171. 2025/05/12. https://doi.org/10.1161/01.str.0000162715.43417.91

10.

Zhang

. The Technology to Enhance Patient Motivation in Virtual Reality Rehabilitation: A Review. Games for health journal 2024; 13: 215–233. 2024/08/19. https://doi.org/10.1089/g4h.2023.0069

11.

Nath

Singh

Saini

, et al. Clinical potential and neuroplastic effect of targeted virtual reality based intervention for distal upper limb in post-stroke rehabilitation: a pilot observational study. Disability and rehabilitation 2024; 46: 2640–2649. 2023/06/29. https://doi.org/10.1080/09638288.2023.2228690

12.

Feitosa

Fernandes

Casseb

, et al. Effects of virtual reality-based motor rehabilitation: a systematic review of fMRI studies. Journal of neural engineering 2022; 19: 1–18. 2021/12/22. https://doi.org/10.1088/1741-2552/ac456e

13.

Meng

Yan

, et al. Transcranial direct current stimulation with virtual reality versus virtual reality alone for upper extremity rehabilitation in stroke: A meta-analysis. Heliyon 2022; 9: 1–10. 2022/12/29. https://doi.org/10.1016/j.heliyon.2022.e12695

14.

Khan

Imam

Muneer

, et al. Virtual reality in stroke recovery: a meta-review of systematic reviews. Bioelectronic Medicine 2024; 10: 1–15. 2024/10/05. https://doi.org/10.1186/s42234-024-00150-9

15.

Donthu

Kumar

Mukherjee

, et al. How to conduct a bibliometric analysis: An overview and guidelines. Journal of Business Research 2021; 133: 285–296. 2021/09/01. https://doi.org/10.1016/j.jbusres.2021.04.070

16.

Ding

Yang

. Knowledge mapping of platform research: a visual analysis using VOSviewer and CiteSpace. Electronic Commerce Research 2022; 22: 787–809. 2022/09. https://doi.org/10.1007/s10660-020-09410-7

17.

Thelwall

. Bibliometrics to webometrics. Journal of Information Science 2008; 34: 605–621. 2008/06/13. https://doi.org/10.1177/0165551507087238

18.

Merigo

Yang

J-B

. A Bibliometric Analysis of Operations Research and Management Science. Omega 2016; 73: 37–48. 2017/12/16. https://doi.org/10.1016/j.omega.2016.12.004

19.

Lei

Wang

Liu

, et al. Global research hotspots and trends of the Notch signaling pathway in the field of cancer: a bibliometric study. American journal of translational research 2022; 14: 4918–4930. 2022/08/13.

20.

Jack

Boian

Merians

, et al. Virtual reality-enhanced stroke rehabilitation. IEEE transactions on neural systems and rehabilitation engineering: a publication of the IEEE Engineering in Medicine and Biology Society 2001; 9: 308–318. 2001/09/20. https://doi.org/10.1109/7333.948460

21.

Wang

Jiang

Sun

, et al. Prevalence, Incidence, and Mortality of Stroke in China: Results from a Nationwide Population-Based Survey of 480 687 Adults. Circulation 2017; 135: 759–771. 2017/01/06. https://doi.org/10.1161/circulationaha.116.025250

22.

Gao

Jiang

Sun

, et al. The burden of stroke in China: Results from a nationwide population-based epidemiological survey. PloS one 2018; 13: 1–11. 2018/12/07. https://doi.org/10.1371/journal.pone.0208398

23.

Wang

, et al. China Stroke Statistics 2019: A Report From the National Center for Healthcare Quality Management in Neurological Diseases, China National Clinical Research Center for Neurological Diseases, the Chinese Stroke Association, National Center for Chronic and Non-communicable Disease Control and Prevention, Chinese Center for Disease Control and Prevention and Institute for Global Neuroscience and Stroke Collaborations. Stroke and vascular neurology 2020; 5: 211–239. 2020/08/23. https://doi.org/10.1136/svn-2020-000457

24.

Budget

. https://www.nih.gov/about-nih/what-we-do/budget

25.

Calabrò

Naro

Russo

, et al. The role of virtual reality in improving motor performance as revealed by EEG: a randomized clinical trial. J Neuroeng Rehabil 2017; 14: 1–16. 2017/06/09. https://doi.org/10.1186/s12984-017-0268-4

26.

Saposnik

Teasell

Mamdani

, et al. Effectiveness of virtual reality using Wii gaming technology in stroke rehabilitation: a pilot randomized clinical trial and proof of principle. Stroke 2010; 41: 1477–1484. 2010/05/29. https://doi.org/10.1161/strokeaha.110.584979

27.

Laver

Lange

George

, et al. Virtual reality for stroke rehabilitation. The Cochrane database of systematic reviews 2017; 11: 1–161. 2017/11/21. https://doi.org/10.1002/14651858.CD008349.pub4

28.

Langhorne

Bernhardt

Kwakkel

. Stroke rehabilitation. Lancet (London, England) 2011; 377: 1693–1702. 2011/05/17. https://doi.org/10.1016/s0140-6736(11)60325-5

29.

Chen

. Effectiveness of Using Virtual Reality-Supported Exercise Therapy for Upper Extremity Motor Rehabilitation in Patients With Stroke: Systematic Review and Meta-analysis of Randomized Controlled Trials. Journal of medical Internet research 2022; 24: 1–26. 2022/06/21. https://doi.org/10.2196/24111

30.

Soleimani

Ghazisaeedi

Heydari

. The efficacy of virtual reality for upper limb rehabilitation in stroke patients: a systematic review and meta-analysis. BMC medical informatics and decision making 2024; 24: 1–25. 2024/05/25. https://doi.org/10.1186/s12911-024-02534-y

31.

Aderinto

Olatunji

Abdulbasit

, et al. Exploring the efficacy of virtual reality-based rehabilitation in stroke: a narrative review of current evidence. Annals of medicine 2023; 55: 1–18. 2023/11/27. https://doi.org/10.1080/07853890.2023.2285907

32.

Han

S-H

Jang

Lee

, et al. The effect of virtual reality-based treadmill gait training on functional mobility and balance in chronic stroke patients: a randomized controlled trial. Front Neurol 2025; 16: 1–11. 2025/07/21. https://doi.org/10.3389/fneur.2025.1603233

33.

Kenea

Abessa

Lamba

, et al. Immersive Virtual Reality in Stroke Rehabilitation: A Systematic Review and Meta-Analysis of Its Efficacy in Upper Limb Recovery. Journal of Clinical Medicine 2025; 14: 1–24. 2025/03/07. https://doi.org/10.3390/jcm14061783

34.

Sadeghi

Jehu

Daneshjoo

, et al. Effects of 8 Weeks of Balance Training, Virtual Reality Training, and Combined Exercise on Lower Limb Muscle Strength, Balance, and Functional Mobility Among Older Men: A Randomized Controlled Trial. Sports health 2021; 13: 606–612. 2021/02/16. https://doi.org/10.1177/1941738120986803

35.

Senthil Kumar

Rambabu G

Manasa K

, et al. Virtual Reality Applications in Healthcare Rehabilitation Therapy and Medical Training Innovations. ITM Web of Conferences 2025; 76: 1–10. 2025/03/25. https://doi.org/10.1051/itmconf/20257604001

36.

Kerem Gunel

Kaya Kara

Turker

, et al. Virtual Reality in Rehabilitation of Children with Cerebral Palsy. In: Švraka

(ed). Cerebral Palsy - Challenges for the Future. IntechOpen, 2014.

37.

Martin

Malpica

Gutierrez

, et al. Multimodality in VR: A Survey. ACM Comput Surv 2022; 54: 1–36. 2022/09/13. https://doi.org/10.1145/3508361

38.

Topini

Sansom

Secciani

, et al. Variable Admittance Control of a Hand Exoskeleton for Virtual Reality-Based Rehabilitation Tasks. Frontiers in Neurorobotics 2022; 15: 1–18. 2022/01/12. https://doi.org/10.3389/fnbot.2021.789743

39.

Tamayo-Serrano

Garbaya

Blazevic

. Gamified In-Home Rehabilitation for Stroke Survivors: Analytical Revie w. IJSG 2018; 5: 1–26. 2018/3/26. https://doi.org/10.17083/ijsg.v5i1.224

40.

Sgubin

Deodato

Murena

. Gamification in Rehabilitation: The Role of Subjective Experience in a Multisensory Learning Context – A Narrative Review. Gestalt Theory 2023; 45: 121–137. 2023/08/30. https://doi.org/10.2478/gth-2023-0012

41.

Sánchez-Gil

Sáez-Manzano

López-Luque

, et al. Gamified devices for stroke rehabilitation: A systematic review. Computer methods and programs in biomedicine 2025; 258: 108476. 2024/11/13. https://doi.org/10.1016/j.cmpb.2024.108476

42.

. Machine Learning in Rehabilitation Training: Traditional and Deep Learning Approaches. ITM Web of Conferences 2025; 78: 1–9. 2025/09/08. https://doi.org/10.1051/itmconf/20257802026

43.

Marks

. Artificial intelligence and rehabilitation: what’s new and promising. IPMRJ 2023; 8: 135–140. 2023/6/29. https://doi.org/10.15406/ipmrj.2023.08.00348

44.

Senadheera

Hettiarachchi

Haslam

, et al. AI Applications in Adult Stroke Recovery and Rehabilitation: A Scoping Review Using AI. Sensors (Basel, Switzerland) 2024; 24: 1–32. 2024/10/26. https://doi.org/10.3390/s24206585

45.

Hussain

Munir

Mamun

, et al. A Transparent House Price Prediction Framework Using Ensemble Learning, Genetic Algorithm-Based Tuning, and ANOVA-Based Feature Analysis. FinTech 2025; 4: 1–26. 2025/07/18. https://doi.org/10.3390/fintech4030033

46.

Chowdhury

Mamun

Shaikat

, et al. An Ensemble Approach for Artificial Neural Network-Based Liver Disease Identification from Optimal Features Through Hybrid Modeling Integrated with Advanced Explainable AI. Medinformatics 2025; 2: 107–119. 2025/03/20. https://doi.org/10.47852/bonviewMEDIN52024744

47.

Mamun

Chowdhury

Faruq

, et al. Identification of Maternal Health Risk From Optimal Features Using Explainable Machine Learning. Engineering Reports 2025; 7: 1–20. 2025/11/12. https://doi.org/10.1002/eng2.70491

48.

Shaikat

MTA

Chowdhury

Shovon

, et al. Explainability Elevated Obstructive Pulmonary Disease Care: Severity Classification, Quality of Life Prediction, and Treatment Impact Assessment. In: 2025 International Conference on Electrical, Computer and Communication Engineering (ECCE), Chittagong, Bangladesh, 13/02/2025, pp. 1–6.

49.

Chowdhury

Mamun

. Hrnet-Xgb: A Comprehensive Hybrid Framework for Automated and Interpretable Skin Cancer Detection with Multi-Scale Feature Fusion. SSRN, 2025.

50.

Mamun

Ali

Chowdhury

MSA

, et al. A Differential Privacy and TOPSIS Enhanced Explainable Machine Learning Framework for Diabetes Risk Diagnosis. In: 2025 International Conference on Electrical, Computer and Communication Engineering (ECCE), Chittagong, Bangladesh, 13/02/2025, pp. 1–6.

Research hotspots and trends in virtual reality combined exercise therapy for stroke rehabilitation: A bibliometric analysis

Abstract

Background

Objective

Methods

Results

Conclusions

Keywords

1. Introduction

2. Materials and methods

2.1. Search strategy

2.2. Inclusion and exclusion criteria

2.3. Bibliometric and visual analysis

2.4. Statistical analysis

2.5. Ethical considerations

3. Result

3.1. Trends in the number of publications

3.2. Distribution of countries

3.3. Analysis of institutions

3.4. Analysis of authors and co-cited authors

3.5. Analysis of journals

3.6. Cited reference analysis

3.7. Analysis of keywords

4. Discussion

5. Conclusion

Footnotes

ORCID iD

Consent to participate

Funding

Declaration of conflicting interests

Appendix

References