Trends and hotspots related to traditional and modern approaches on acupuncture for stroke: A bibliometric and visualization analysis

Abstract

Background:

Acupuncture role in stroke treatment and post-stroke rehabilitation has garnered significant attention. However, there is a noticeable gap in bibliometric studies on this topic. Additionally, the precision and comprehensive methodology of cluster analysis remain underexplored. This research sought to introduce an innovative cluster analysis technique (called follower-leading clustering algorithm, FLCA) to evaluate global publications and trends related to acupuncture for stroke in the recent decade.

Methods:

Publications pertaining to acupuncture for stroke from 2013 to 2022 were sourced from the Web of Science Core Collection. For the assessment of publication attributes—including contributing countries/regions (e.g., US states, provinces, and major cities in China) in comparison to others, institutions, departments, authors, journals, and keywords—we employed bibliometric visualization tools combined with the FLCA algorithm. The analysis findings, inclusive of present research status, prospective trends, and 3 influential articles, were presented through bibliometrics with visualizations.

Results:

We identified 1050 publications from 92 countries/regions. An initial gradual rise in publication numbers was observed until 2019, marking a pivotal juncture. Prominent contributors in research, based on criteria such as regions, institutions, departments, and authors, were Beijing (China), Beijing Univ Chinese Med (China), the Department of Rehabilitation Medicine, and Lidian Chen (Fujian). The journal “Evid.-based Complement Altern” emerged as the most productive. The FLCA algorithm was effectively employed for co-word and author collaboration analyses. Furthermore, we detail the prevailing research status, anticipated trends, and 3 standout articles via bibliometrics.

Conclusion:

Acupuncture for stroke presents a vast research avenue. It is imperative for scholars from various global regions and institutions to transcend academic boundaries to foster dialogue and cooperation. For forthcoming bibliometric investigations, the application of the FLCA algorithm for cluster analysis is advocated.

Key Points:

• 1.This study aimed to develop a cluster algorithm for network analysis to evaluate research achievements (RAs) in research on acupuncture for stroke.
• 2.This study illustrated 2 core components of cluster analysis, namely, coword and author collaboration, which have rarely been discussed in the literature.
• 3.The developed cluster algorithm was provided in detail on an R platform and recommended for future relevant bibliometric studies using cluster analysis.

1. Introduction

Cerebral stroke, a common cerebrovascular disorder, arises from disruptions in blood circulation due to vascular rupture or blockages in the brain blood vessels. This condition encompasses both ischemic and hemorrhagic strokes.^[1,2] Based on earlier research,^[3,4] East Asia recorded the highest age-standardized incidence rate (ASIR) for ischemic stroke in 2019, witnessing a substantial rise from 1990 to 2019.

1.1. The incidence of ischemic stroke is rising

Ischemic stroke, resulting from a clot or other impediment restricting blood flow to the brain, is a prevalent stroke type.^[5] As a primary contributor to global disability and mortality, it accounts for approximately 13.7 million cases annually.^[6] The considerable socio-economic burden imposed by stroke underscores the significance of its prevention and management. Though the incidence fluctuates based on age, gender, and location, it typically rises with age and is more prevalent in men.^[7] Hypertension, diabetes, smoking, and elevated cholesterol levels are among the notable risk factors for ischemic stroke. Prompt identification and intervention can substantially curtail the risk.^[8–12]

Worldwide, ischemic stroke incidence is on the upswing, especially among women between the ages of 50 and 69.^[13] In 2019, China recorded 394 million new cases of stroke, marking an 86.0% surge compared to 1990.^[14] Although developed nations are witnessing a decline, stroke incidence in China has seen a notable spike, continuing to be a leading cause of death.^[15,16] Consequently, prioritizing the prevention and management of cerebral infarction is imperative.

Acupuncture encompasses a variety of procedures aimed at stimulating specific body points using diverse techniques to derive therapeutic benefits.^[17] For millennia, acupuncture has been a cornerstone of Traditional Chinese Medicine (TCM), especially in China.^[18] Over recent years, acupuncture's global popularity has surged, with even top-tier academic medical institutions adopting its practice.^[19–21] Nonetheless, its widespread application does not come without skepticism, as there remains a persistent debate within the medical fraternity regarding its efficacy and specific therapeutic uses.^[17,22–26] This discourse is significantly shaped by the volume, quality, origin, and character of existing research on acupuncture.^[18] Therefore, the study was conceived to investigate research hotspots and emerging trends regarding acupuncture for strokes based on bibliometric analysis.

1.2. Acupuncture utilized for patients with stroke

Acupuncture has demonstrated efficacy in improving neurobehavioral functions and diminishing brain infarct size in animals, positioning it as a prevalent treatment for stroke and its subsequent complications like motor and cognitive impairments, and mental disorders. A retrospective study from Taiwan has suggested that acupuncture could potentially reduce the recurrence rate of strokes, even in patients already on preventive medications.^[27] Investigations into the underlying mechanisms of acupuncture benefits have highlighted its role in fostering autophagy, inhibiting apoptosis, countering inflammation, reducing oxidative stress, and enhancing cerebral circulation.^[28–32]

In recent times, a myriad of systematic reviews and meta-analyses have been undertaken to assess the efficacy and safety of acupuncture across diverse ailments, including back and neck pain,^[33–35] osteoarthritis,^[36,37] nausea,^[38,39] dental pain,^[40] menstrual discomfort,^[41] and infertility.^[42,43] Yet, there remains a discernible gap in understanding the overarching research trajectories associated with acupuncture. Hence, there is an imperative to impartially delineate publication tendencies and juxtapose them with other pertinent fields^[44–46] through a series of visualization using bibliometric analysis.

1.3. Trends and hotspots using bibliometric analyses

Bibliometrics serves as a quantitative technique for analyzing scholarly literature.^[47] This method facilitates the assessment of productivity, quality (or “performance”), and evolving patterns of researchers, institutions, or specific academic domains.^[18] Acting as a simplified big data analytical tool, it can chronicle historical progress,^[48] quantify prevailing trends,^[45] and forecast prospective trajectories within a research area.^[49]

Concurrently, bibliometric approaches have found extensive application in medical research,^[50–54] encompassing studies in complementary and alternative medicine.^[55–57] The results derived from such investigations can offer insights into the strengths and deficiencies in a domain, guiding the formulation of a holistic strategy to further elevate the field. However, there are some drawbacks associated with traditional approaches to bibliometrics that should be considered.

1.4. Research questions of trends and hotspots over years

To bibliometrics, a comprehensive review encapsulating global trends in acupuncture research remains conspicuously missing. Moreover, there a distinct gap in exploring the accuracy and nuanced methodologies inherent in cluster analysis, a pivotal aspect of bibliographic studies on social network analysis. While some research has leveraged bibliometric software tools like CiteSpace^[58,59] and VOSviewer^[60] to decipher emerging patterns and key areas over time, the intricate algorithms underpinning their methodologies often lack thorough elucidation. This underscores an urgent need to devise a robust cluster analysis technique apt for both coword and author collaboration evaluations.

In addition, 2 notable voids were observed in the literature: an absence of exhaustive overviews of worldwide acupuncture future trends, and a restricted comprehension of cluster analysis techniques used for detecting research hotspots. This study seeks to bridge these gaps, with the ambition of enhancing the domain, steering subsequent research, and potentially fostering advancements in bibliographic studies by introducing a modern cluster analysis approach (called follower-leading clustering algorithm, FLCA). Consequently, the potential influence of this research is expected to resonate beyond its direct context, catalyzing wider progress within the academic realm.

1.5. Study objectives

This study had 3 main objectives: to create a new FLCA algorithm, to demonstrate the trends and hotspots in acupuncture research for stroke, and to employ the FLCA algorithm to analyze author collaborations and keyword occurrences in the field of acupuncture for stroke.

2. Methods

2.1. Data source

A total of 1050 articles related to acupuncture and stroke research between 2013 and 2022 were collected using the Web of Science Core Collection. This collection only consisted of articles and review articles indexed in the Science Citation Index Expanded (SCI-E). Others, such as those marked as “Published Erratum, Editorial, or conference proceedings,” were excluded. Article citations were extracted and matched to the articles with the following search criteria: TS = (acupuncture or Electroacupuncture or dry need or acupoint injection) AND TS = (apoplexy or stroke or cerebral infarction).

A comprehensive range of metadata was obtained, including author names, research institutes, departments, countries of origin, and other relevant information. A complete list of metadata is available in Supplemental Digital Content 1, http://links.lww.com/MD/K741.

All data presented in this study are publicly available on Web of Science Core Collection, and no participant identification information was collected or disclosed. As such, ethical approval was waived for this study.

2.2. Features of the FLCA algorithm

2.2.1. Features in traditional and FLCA approaches.

CiteSpace^[58,59] often generates 2 graphs displaying keyword hotspots in Figure 1, panels A and B, in a traditional manner. However, the first graph shows too many connected links, making it difficult to distinguish the relationship between keywords and years by color. The second graph (i.e., panel B of Fig. 1), which does not cluster keywords, displays the strength of keyword hotspots over time. The heatmap with the dendrogram in panel C of Figure 1 has complex colors that may confuse readers when identifying the vertical and horizontal clusters.

Figure 1.

Comparison of trends and hotspots between traditional (A, B, and C) and innovative (D, E, and F) approaches.

In contrast, the network chart in panel F provides a clear visualization of the evolution of keywords from cluster #1 to #6 with bubble sizes and colors, which was achieved through data arrangement and cluster analysis using FLCA in panels D and E of Figure 1.

2.2.2. FLCA algorithm with visual interpretations.

FLCA involves 2 major types of networks for coword and author-collaboration analyses: 2-mode and 1-mode.^[61]

2.2.2.1. Steps for the 2-mode network.

For the 2-mode network, there are 3 steps involved.

Step 1: The dataset should include keywords and years defined as followers and leaders in panel D of Figure 1.

Step 2: All keywords are sorted in descending order based on their total observed counts (also known as weighted centrality degree^[62]), allowing the cluster leader to be vertically ranked at the top.

Step 3: keywords are horizontally assigned to the year with the highest observed count number (for example, the counts of 13 and 10 are assigned to 2013 for expression and stroke patients, respectively). Through the heatmap with a dendrogram in R^[63] and the network chart shown in panels E and F of Figure 1, it is anticipated that the followers of keywords can be led by cluster leaders (i.e., from 2013 to 2022).

Additional graphs for interpret the FLCA in 2-mode data are shown in Figure 2.

Figure 2.

FLCA algorithm in 2-mode data. FLCA = follower-leading clustering algorithm.

2.2.2.2. Steps for 1-mode network.

For the 1-mode network, there are 5 steps involved in FLCA.

Step 1: Data for vertices (such as keywords or countries of origin) are arranged in descending order based on the weighted centrality degree (WCD)^[62] in network analysis, with provisional cluster numbers assigned to each vertex from top to bottom (e.g., 1 to n).

Step 2: Followers with lower WCD (e.g., the weighted article number) in relatively right columns are connected to potential leaders with higher WCD in relatively left columns, based on a single connection, with the maximum, linked to each other (e.g., in panel D of Fig. 1). In a 1-mode network, labels in columns are identical to those in rows.

Step 3: Leaders recruit their followers as members in the designated cluster by assigning the leader cluster number to members until the stop criterion is met. Note that any vertex with at least 1 follower becomes a new leader in the designated cluster.

Step 4: Cluster numbers are sorted in a consecutive series of numbers.

Step 5: Clusters are visualized using a network chart.

The FLCA algorithm of a 1-mode network can be expressed mathematically in Equations 1 and 2.^[64-66]

$$\{\begin{array}{l}{L}_i=\underset{i=1}{\stackrel{k}{{{\displaystyle \sum }}^{\text{}}}}\underset{j=1}{\stackrel{n}{{{\displaystyle \sum }}^{\text{}}}}\left(F_j\in L_i\right),\left(1\right)\\ F_i=\underset{i=1}{\stackrel{n}{{{\displaystyle \sum }}^{\text{}}}}\underset{j=1}{\stackrel{n}{{{\displaystyle \sum }}^{\text{}}}}\left(F_j\in L_i\right),\left(2\right)\\ F_j\in L_{i}withtheoriginalclusternumberof{L}_{i}if{F}_{j}becomesanew{L}_i\end{array}$$

In the FLC algorithm via Equation 1, F is the follower assigned to a leader, while a follower appears based on the stop criterion k (where k<=n = sample size). A smaller k leads to fewer clusters. Followers are determined by the link to the leader based on Eq 1, where k = n. It is worth noting that if $L_i$ becomes a new leader from a member in the previous clustering process, the followers are assigned the newly (instead of the original series of numbers) assigned cluster number of $L_i$.

2.3. Four major parts in this section

2.3.1. Descriptive analytics.

To analyze the trend of publications on acupuncture for stroke, an inflection point^[67–69] was used to create a line chart. In addition, a 4-quadrant plot^[70] was employed to present the dominant entities in acupuncture for stroke research based on the category, journal, authorship, L-index core,^[70–72] which takes into account the category, journal impact factor, and authorship (CJA) score^[71] as well as the L-index.^[73] The publications of the first and corresponding authors FP and RP^[74] were merely considered.

To display the top 20 prestigious journals in the field of acupuncture for stroke, a circle packing chart in R^[66] was utilized. Furthermore, a 2-axes stacked bar and line chart in R^[66] was used to compare trends in publications for the top 3 regions and institutes. Further details about the method used to draw charts in R can be found in Supplemental Digital Content 2, http://links.lww.com/MD/K742.

2.3.2. Cluster analysis for cowords with a 2-mode network.

To enhance comprehension of the clustering process based on the 2-mode keywords using FLCA,^[64–66] 3 visualizations were employed to interpret the FLCA algorithm: a heatmap with a dendrogram, a circle bar plot,^[75,76] and a network chart referred to the software Pajek [in Koeln; PajekMan in Osoje (Ossiach, Austria)].^[77]

2.3.3. Cluster analysis for author collaborations with a 1-mode network.

Chord diagrams^[75,78,79] were used to present the results of FLCA algorithm for the top 20 productive entities based on region, institute, department, and author.

2.3.4. The most noteworthy articles to read.

A temporal heatmap (THM)^[80] was utilized to exhibit the top 20 most cited articles and the most significant articles to read based on yearly article citations. By analyzing their citation trend, 4 types of articles were identified: consecutively increasing, reading to rise, slowing down, and reading to decline.

2.4. The assumptions of the FLCA algorithm

There are multiple techniques and software tools available for conducting cluster analysis, such as Bibexcel,^[81] CiteSpace,^[58,59] CoPalRed,^[82,83] IN-SPIRE,^[84] Leydesdorff Software, Network Workbench Tool,^[85,86] Science of Science (Sci2) Tool,^[87] VantagePoint,^[88] and VOSViewer.^[63] To classify article topics or themes, coword analysis^[76] can be used, such as through exploratory factor analysis.^[89] In this study, we assume that coword analysis is akin to author collaborations, as their co-occurrences can be observed in a network if a 1-mode network is utilized.

2.5. Special terms used in this study

Cluster: In bibliometric analysis, a cluster is a group of related scientific papers that share common themes or topics based on their bibliographic information, such as authors, institutions, keywords, or citation patterns. The resulting clusters can help inform research policy and funding decisions.

Topic spot: A topic spot is identified through trend analysis of keywords, which involves tracking the frequency of specific keywords over time and identifying patterns or changes in their usage. Topic spots can be used to identify new research trends and emerging areas of interest.

Inflection point: A mathematical model can be used to estimate the infection point, which is the time at which the rate of infection or spread of the disease changes significantly due to public health measures or changes in population behavior.

3. Results

3.1. Descriptive analytics

Figure 3 indicates that a total of 1050 publications from 92 countries/regions were analyzed, and the growth rate in the number of publications was slow until it reached an inflection point in 2019.

Figure 3.

Trend analysis of publications in research on acupuncture for stroke.

According to Figure 4, the top 3 leading regions were Beijing (China), Guangdong (China), and South Korea. The top 3 ranking institutions were Beijing University of Chinese Medicine (China), Guangzhou University of Chinese Medicine (China), and Fujian University of Chinese Medicine (China). The most productive department and author in acupuncture for stroke were Rehabilitation Medicine and Lidian Chen (Fujian), respectively.

Figure 4.

Dominant entities in comparison using a 4-quadrant plot to display.

Figure 5 with the circle packing chart highlights the most prestigious journals in this field, which were Evid.-based Complement Altern. Med., Medicine (Baltimore), and Acupuncture in Medicine.

Figure 5.

Top 20 journals in research on acupuncture for stroke.

Finally, Figure 6 illustrates that publications by 2 types of authors from Guangzhou Univ Chinese Med (China), Beijing Univ Chinese Med (China), Beijing, and Guangdong province exhibited a sharp increase since 2019. Publications in 3 regions from Beijing, Guangdong, and South Korea were increasing.

Figure 6.

Comparison of trends in publications for the top 3 regions and institutes was made using the 2-axes stacked bar and line chart in R.

3.2. Cluster analysis for cowords

Figures 7 to 9 display the outcomes of cluster analysis employing FLCA algorithm.^[64–66] The 10 clusters with keywords are presented in all 3 visualizations: the heatmap (Fig. 6), where the cluster leaders are located at the top with vertical followers shown on the right-hand side; the circle bar plot with observed counts; and the network chart depicting trends and hotspots over the years.

Figure 7.

Cluster analysis of keywords in trends with the top 21 keywords using heatmap.

3.3. Cluster analysis for author collaborations

To display the cluster results for the top 20 productive entities based on region, institute, department, and author, chord diagrams were utilized in Figures 10 to 13. The clusters are color-coded outside the chord diagrams, while the members with links are colored inside the circle. Compared to network charts in bibliometrics, chord diagrams are more effective in presenting the relationships between and within cluster members. There is a premise that clusters have been classified in advance.

Figure 10.

Clusters related to regions for author collaborations with a 1-mode network.

3.4. The most noteworthy articles to read

Figure 14 shows the temporal heatmap (THM)^[80] for displaying the top 20 most cited articles^[90] and the most significant articles to read. Readers are provided with 3 articles^[91–93] with a high citation trend over the last 4 years.

Figure 14.

The top 20 most cited articles and the most noteworthy articles to read using the temporal heatmap.

Figure 8.

Cluster analysis of keywords in trends with the top 21 keywords using a circle bar plot (note. bar heights mean the observed counts for keywords).

Figure 11.

Clusters related to institutes for author collaborations with a 1-mode network.

Figure 12.

Clusters related to departments for author collaborations with a 1-mode network.

Figure 13.

Clusters related to authors for author collaborations with a 1-mode network.

4. Discussion

4.1. Principal finding

In our study, we scrutinized 1050 articles originating from 92 countries/regions. There was a consistent uptick in publication numbers until 2019, at which point a noticeable shift was evident. The leading regions were Beijing (China), Guangdong (China), and South Korea. The foremost institutions comprised Beijing University Chinese Med (China), Guangzhou University Chinese Med (China), and Fujian University Chinese Med (China). The department of Rehabilitation Medicine and Lidian Chen (Fujian) emerged as the most prolific in their respective categories concerning acupuncture for stroke. The journals with the highest output were Evid.-based Complement Altern. Med., Medicine (Baltimore), and Acupuncture in Medicine. Leveraging the FLCA algorithm,^[64–67] our study delved into co-word and author collaboration analyses. Additionally, we outlined the prevailing research landscape, prospective trends, and spotlighted 3 standout articles through bibliometric evaluation.

The study effectively achieved 3 objectives: introducing the FLCA algorithm, illustrating trends and hotspots in acupuncture research for stroke, and utilizing the FLCA algorithm to analyze author collaborations and keyword occurrences in the field of acupuncture for stroke.

4.2. Additional Information

Recent years have seen a surge in acupuncture research for stroke,^[94,95] particularly since 2010. While initial studies yielded negative results,^[96] subsequent research has validated acupuncture efficacy for stroke,^[97] probing into its underlying mechanisms using both clinical and animal models.

This study offers a comprehensive overview of the past decade trends in this realm. China emerges as a leader, contributing the most publications (n = 1494), given acupuncture roots in Chinese tradition and robust policy support.^[98,99] Taiwan, South Korea, and the US follow China.

Key research themes encompass clinical studies on stroke complications like insomnia,^[100] depression,^[101] and hemiplegia.^[102] Experimental studies gravitate towards the mechanism of action, spotlighting cerebral ischemia, ischemic stroke, and cognitive impairments, especially in 2022 (Fig. 9).

Figure 9.

Cluster analysis of keywords in trends with top 21 keywords using network (note. the larger bubble means more observed counts for keywords).

4.3. Research on acupuncture for stroke rehabilitation

Acupuncture efficacy in post-stroke rehabilitation has been extensively explored, with over 829 related articles indexed in PubMed.^[103]

Meta-analysis^[104] from 88 studies indicated acupuncture positive impacts on motor function and daily activities in stroke patients. Li et al^[105] reported mixed results after a 2-week acupuncture treatment, with some improvements seen in NIHSS scores and BI scores. Kuang et al^[106] conducted a comprehensive database search, revealing 28 high-quality trials suggesting acupuncture potential benefits in cognitive function assessments, such as MoCA and MMSE scores. Zhong et al^[107] analyzed 35 studies and found that acupuncture, in conjunction with other treatments, showed improvements in various swallowing assessments post-stroke.

While current research underscores acupuncture potential advantages in stroke rehabilitation, there is a need for more rigorous clinical trials to determine optimal treatment protocols and assess long-term outcomes.

4.4. Articles worth reading on acupuncture for stroke

In a 2017 study by Chavez et al,^[91] which garnered 122 citations, the World Health Organization endorsed acupuncture as an alternative for stroke treatment and enhancement of stroke care. The study delves into the mechanisms of ischemic stroke rehabilitation through acupuncture and electroacupuncture.

Another significant research by Lan et al in 2013^[92] with 92 citations found that electroacupuncture at specific points improved neurological deficits, reduced cerebral infarction, and mitigated inflammatory responses in ischemic stroke rats. This process regulated key molecules in the TLR4/NF-B signaling pathway.

The article^[93] authored by Yang et al in 2016, cited 88 times, highlighted that acupuncture decreased the risk of death or dependency for subacute and chronic stroke patients, ameliorating neurological deficits and enhancing life quality. In 31 trials involving 2257 participants, those in post-stroke recovery exhibited notable improvements after receiving either actual or sham acupuncture, with minimal serious side effects.

4.5. Implications and changes

This research introduces the FLCA algorithm for enhancing cluster analyses of bibliometric cooccurrence patterns in coword and author collaborations. Using modern visualization tools such as chord diagrams, circle bar plots, heatmaps, and network charts, the study offers a more insightful understanding of these patterns than traditional methods based on similarity and Euclidean distance.

Notably, the study uses R code for its visualizations, available in Supplemental Digital Content 2, http://links.lww.com/MD/K742. In a departure from many other works, the FLCA is promoted as a method even implementable in MS Excel, demonstrated in Supplemental Digital Content 3, http://links.lww.com/MD/K743.

In addition, the study distinctive features include: the development of the FLCA algorithm, the FLCA with traditional heatmap approaches can be applied beyond bibliometrics, unique chord diagrams for network visualization with accompanying R code, and guidance on producing plots and an MP4 video, detailed in Supplemental Digital Contents and reference.^[108]

4.6. Limitations and suggestions

Future studies should address several considerations:

1. The FLCA algorithm^[54–67] is straightforward but needs refinement. Improvements could include giving precedence to individuals with higher WCD for maximum connection and incorporating a stop criterion for relationships within the AC network.

2. The current study focuses on dominant entities in acupuncture research for stroke from 2013 to 2022. A broader perspective could be achieved by examining varying time frames using indicators like category, journal, authorship, L-index.^[70]

3. Although R was used for visual displays in this study, multiple software options exist for network analysis. The priority should be on user-friendly, clear visual displays for readers via accessible online tools.^[109]

4. The primary and corresponding authors were only taken into account in author collaborations. This method may yield different outcomes from tools like CiteSpace^[58,59] and VOSviewer,^[60] which consider all coauthors during collaboration analysis.

5. The heatmap and dendrogram efficacy for clustering relies heavily on the input data. Proper positioning of variable and sample definitions is vital to avoid cluster analysis interpretation challenges.

5. Conclusion

This study provides a comprehensive bibliometric analysis of acupuncture role in stroke treatment and rehabilitation over the past decade, addressing the existing gap in the literature. Through the innovative FLCA algorithm, we have effectively categorized and visualized key contributors, research trends, and influential publications in the field. The findings underscore Beijing and Beijing Univ Chinese Med as significant contributors, and the journal “Evid.-based Complement Altern” as a primary publication platform. The successful application of the FLCA algorithm in this research not only illuminates the current state and potential trajectory of acupuncture research related to stroke but also demonstrates the algorithm utility in enhancing bibliometric studies. This research is pivotal for researchers seeking a holistic overview and deeper insights into the domain of acupuncture for stroke treatment and post-stroke rehabilitation.

Acknowledgments

We thank Enago (www.enago.tw) for the English language review of this manuscript.

Author contributions

Conceptualization: Chao-Yu Chuang.

Data curation: Willy Chou.

Formal analysis: Tung-Hui Jen.

Methodology: Tsair-Wei Chien.