Exploring Research Trends in Chronic Kidney Disease and Sedentary Behavior: A Text Mining Analysis of Article Titles Authors: Akihiro Yokoyama, Nobuhiro Nasu, Akihiko Katayama, Shuhei Hishii, Keiichi Namio, Kentaro Sugahara, Hiroaki Kataoka, Hiromi Suzuki, Nobuyuki Miyatake Published: 2026 ABSTRACT Sedentary behavior (SB) has recently gained attention across various diseases due to its relationship with disease onset, quality of life, and prognosis. In the present study, we searched for journal article titles associated with the relationship between chronic kidney disease (CKD) and SB and used an exploratory text mining analysis to conduct an exploratory examination of research trends.On January 1, 2026, we entered “chronic kidney disease and sedentary behavior” into the PubMed® article search engine and downloaded article titles, authors, journal names, and publication years. Using the text mining software KH Coder® (KH Coder 3.0, Koichi Higuchi, Japan), we extracted frequently occurring words from article titles and performed a correspondence analysis by time period (up to 2015, 2016-2020, and 2021-2025) and survey location (Japan and outside Japan). The search results found 178 articles. The total number of words obtained for article titles was 2,890. The top five most frequent words were “kidney” (91 words), “disease” (88 words), “physical” (78 words), “activity” (68 words), and “chronic” (68 words). In the correspondence analysis by time period, “randomize”, “intervention”, and “cohort” were characteristic. “Randomize” was characteristic from 2021 to 2025. In the analysis by survey location, only eight out of all articles conducted surveys in Japan, indicating that surveys in Japan were fewer, with most originating from outside Japan.Research investigating the relationship between CKD and SB has recently begun to see the implementation of high-level evidence studies, such as cohort studies and randomized controlled trials. Furthermore, many of these studies appear to have been conducted outside Japan. Introduction and background Patients with chronic kidney disease (CKD) are generally elderly, often have complications, and also have low levels of physical activity [1-3]. The latter characteristic may be described as not only “inactive”, but also “excessively sedentary” [2-8]. Sedentary behavior (SB) has recently gained attention as a health risk factor independent of physical activity levels [4-12]. It has also been associated with the onset of various diseases, quality of life (QOL), and prognosis [9-12]. Henson et al. reported that SB was related to an increased prevalence of multiple chronic diseases [9]. Xavier et al. also demonstrated that increased SB resulted in a decline in QOL among patients with Chagas disease [10]. Furthermore, a systematic review by de Rezende et al. showed that increased SB was associated with mortality, fatal and nonfatal cardiovascular diseases, type 2 diabetes, metabolic syndrome, waist circumference, and being overweight/obese [11,12]. We previously investigated the relationship between SB and QOL in patients with chronic hemodialysis (CHD) using cross-sectional and longitudinal studies [4-8]. Hishii et al., Namio et al., and Sugahara et al. reported correlations between SB and prolonged sedentary bouts, and between EuroQol 5-Dimensions 3-level scores, in CHD patients, suggesting a relationship with health-related QOL in cross-sectional studies [4-6]. In addition, Hishii et al. described the relationship between SB and all-cause mortality in patients with CHD using a longitudinal study [7]. Namio et al. also showed that prolonged sedentary bouts were closely associated with all-cause mortality in a prospective cohort study [8]. Text mining is a research method for qualitatively evaluating text and audio data. In recent years, it has been increasingly used in medical research fields [13-19]. Conceição and Couto introduced text mining research in the field of cancer [13]. We also used text mining analysis in studies investigating impressions of COVID-19 vaccines, human papillomavirus vaccines, AIDS, and syphilis [14-16]. In recent years, the use of text mining analysis has increased to efficiently extract information from large volumes of medical literature data [17-19]. However, studies have yet to examine the literature on the relationship between CKD and SB using text mining analysis. Since research on CKD and SB is insufficient, it is important to conduct an exploratory analysis of existing studies in order to identify areas for future research. The titles given to academic papers often convey their content efficiently and concisely. Therefore, the present study aimed to collect article titles related to CKD and SB and to clarify research trends through an exploratory text mining analysis. Review Methods On January 1, 2026, we entered “chronic kidney disease and sedentary behavior” into the PubMed® article search engine and downloaded PubMed ID, Title, Authors, Citation, First Author, Journal/Book, Publication Year, Create Date, PubMed Central ID, NIH Manuscript Submission System ID, and DOI. In the search's custom filter, the publication date was set to 2026 or earlier; “Text availability”, “Article attribute”, and “Article type” were left blank; “All results” was selected under “Save citations to file”; the format was set to CSV; and the data file was downloaded. As a preprocessing step, all uppercase letters in article titles were converted to lowercase. Furthermore, issuance years were categorized into three groups: “Up to 2015”, “2016-2020”, and “2021-2025”. Survey locations were also classified by verifying only the authors’ names and distinguishing between Japan and outside Japan. Article titles were then analyzed using the text mining software KH Coder® (KH Coder 3.0, Koichi Higuchi, Japan) [20,21]. Words were automatically extracted from the article titles collected by PubMed® using KH Coder® to create a list of the frequency of their occurrence. KH Coder® was performed with default settings to eliminate the bias of the analyst’s opinion as much as possible, and default settings were also used for the morphological analysis dictionary, which used the default Stanford Part-of-Speech Tagger, and no settings were configured for “force pick up” or “force ignore” terms. Furthermore, compound words were treated as separate words rather than a single word. We then performed a correspondence analysis [20,21] to visualize the relationships between words in a scatter plot. Classification criteria were the time period (“up to 2015”, “2016-2020”, and “2021-2025”) and survey location (“Japan” and “outside Japan”). The settings for the correspondence analysis were configured with a minimum frequency of five, a minimum number of documents of one, and the top 60 words. Additionally, in the word display for the correspondence analysis, “ckd” was converted into “CKD” and “nhanes” to “NHANES” and displayed in uppercase. Results The search results found 178 articles. All downloaded data are included in the Appendix. The total number of words obtained for article titles was 2,890. The top 10 most frequent words were “kidney” (91 words), “disease” (88 words), “physical” (78 words), “activity” (68 words), “chronic” (68 words), “sedentary” (52 words), “study” (44 words), “patient” (41 words), “adult” (33 words), and “behavior” (32 words). The top 20 most frequently occurring words are indicated in Table 1. Table 1Frequently occurring words (totally 2,890 words) S.no. Word Number of words Percentage 1 Kidney 91 3.15 2 Disease 88 3.04 3 Physical 78 2.70 4 Activity 68 2.35 5 Chronic 68 2.35 6 Sedentary 52 1.80 7 Study 44 1.52 8 Patient 41 1.42 9 Adult 33 1.14 10 Behavior 32 1.11 11 Risk 25 0.87 12 Association 24 0.83 13 Function 19 0.66 14 Factor 18 0.62 15 Older 18 0.62 16 Time 18 0.62 17 Cohort 16 0.55 18 Renal 16 0.55 19 Exercise 15 0.52 20 Health 15 0.52 Frequently occurring words (totally 2,890 words) The number of reports by time period was as follows: “up to 2015” (Japan: 0 reports, outside Japan: 47 reports), “2016-2020” (Japan: one report, outside Japan: 43 reports), and “2021-2025” (Japan: seven reports, outside Japan: 80 reports) (Table 2). Table 2Number of reports from Japan and outside Japan by time period Time period Japan Outside Japan Total Up to 2015 0 47 47 2016-2020 1 43 44 2021-2025 7 80 87 Total 8 170 178 Number of reports from Japan and outside Japan by time period In the correspondence analysis by time period, “randomize” was characteristic of the “2021-2025” period. “Intervention” and “cohort” were recognized in the intermediate between the “2016-2020” and “2021-2025” classifications. “Cross-sectional” and “study” were positioned in the middle of the three groups: “up to 2015”, “2016-2020”, and “2021-2025”. Furthermore, “chronic”, “kidney”, “disease”, and “CKD” were similarly positioned in the middle of the three groups. The results of the correspondence analysis by time period are indicated in Figure 1. Component 1 on the horizontal axis had an eigenvalue of 0.1303 and a contribution of 66.98%, while Component 2 on the vertical axis had an eigenvalue of 0.0642 and a contribution of 33.02%. Terms closer to the origin have higher centrality (the strength of their co-occurrence relationships with other terms). The origin is the point where the dotted lines connecting the 0-point on the vertical axis and the 0-point on the horizontal axis intersect. Correspondence analysis by time period The origin is the point where the dotted lines connecting the 0-point on the vertical axis and the 0-point on the horizontal axis intersect. Terms closer to the origin have higher centrality (the strength of their co-occurrence relationships with other terms) Image credit: This is an original image created by the author Akihiro Yokoyama using KH Coder (Koichi Higuchi, Tokyo, Japan) Furthermore, in the analysis by survey location, eight reports were conducted in Japan, while the remaining 170 originated from outside Japan. The results of the correspondence analysis by survey location are shown in Figure 2. In Figure 2, both the horizontal and vertical axes show an eigenvalue of 0.0531 and a contribution rate of 100%. Terms closer to the origin have higher centrality (the strength of their co-occurrence relationships with other terms). The origin is the point where the dotted lines connecting the 0-point on the vertical axis and the 0-point on the horizontal axis intersect. Correspondence analysis by survey location The origin is the point where the dotted lines connecting the 0-point on the vertical axis and the 0-point on the horizontal axis intersect. Terms closer to the origin have higher centrality (the strength of their co-occurrence relationships with other terms) Image credit: This is an original image created by the author Akihiro Yokoyama using KH Coder (Koichi Higuchi, Tokyo, Japan) Discussion In the present study, we entered “chronic kidney disease and sedentary behavior” into PubMed® and performed a text mining analysis of the downloaded article titles. The search results found 178 articles. In the correspondence analysis focusing on the terms “randomize”, “intervention”, and “cohort”, the word “randomize” was characteristic of the time period “2021-2025”. In the analysis by survey location, only eight of all articles conducted surveys in Japan, indicating that surveys in Japan were clearly fewer, with most reports originating from outside Japan. In recent years, the use of text mining analysis has increased to efficiently extract information from large volumes of medical literature data [17-19]. Forsgren et al. described programming methods that use text mining analysis to assist with screening and data extraction in systematic reviews [17]. Wang and Lo also proposed a method for visualizing and summarizing literature on COVID-19 [18]. Furthermore, Fukushima et al. analyzed and reported on literature regarding peritoneal dialysis published in PubMed® using artificial intelligence text mining [19]. In the present study, text mining analysis of journal article titles related to CKD and SB revealed that high-level evidence studies, including cohort studies and randomized controlled trials (RCTs), have been conducted only in recent years. This may be attributed to increased interest in lifestyle behaviors resulting from the aging of CKD patients and their longer survival [1]. Moreover, factors include the international recognition of SB as a health risk distinct from physical inactivity [4-7] and assessments of SB becoming more straightforward due to the widespread use of accelerometers and wearable devices [2,3,8-12]. In the present study, 178 articles were found. Since this is not a particularly large number, research on CKD and SB is still in its early stages. Furthermore, limited research has been conducted in Japan. One possible reason for this is that CKD research in Japan has generally focused on laboratory values, prognosis, and complications, with a minimal emphasis on behavioral epidemiology. Furthermore, the concept of SB itself is relatively new in Japan, which may have contributed to the delay in its adoption in research. In addition, there is a lack of large-scale behavioral data on CKD patients in the conservative management phase, and there is a need to accumulate more research data in Japan in the future. Limitations This study does not suggest a causal relationship between CKD and SB. Moreover, this study is not intended to demonstrate the effectiveness of intervention strategies; rather, it is a preliminary study aimed at identifying trends in the field through a text mining analysis and identifying priority areas for future research. Therefore, this study has several limitations that need to be addressed. This study was based on an analysis of titles; since the full text of each article was not read, there are limits to how well the details of their content were grasped. Furthermore, the choice of terminology may have had an impact. The term “sedentary behavior” was selected for this study, which itself may have introduced a selection bias in the search results. Another limitation is that the search engine used was limited to PubMed®, and the scope was restricted to English-language literature. Therefore, it did not fully cover all medical literature, including that written in Japanese or other languages. The increasing appearance of terms related to randomized and interventional designs suggests a possible shift toward more rigorous study designs, although confirmation through abstract or full-text review is needed. Nevertheless, this method is extremely useful for gaining an understanding of research trends and general directions [19]. Conclusions We conducted an exploratory investigation of article titles on CKD and SB using text mining analysis. High-level evidence studies, such as cohort studies and RCTs, have been conducted on the relationship between CKD and SB in recent years, and many of these studies appear to have been performed outside Japan. Therefore, further accumulation of cohort studies and RCTs in Japan is needed in the future. These findings suggest that research on CKD and SB is still developing, and the field has not yet reached maturity. The observed patterns help clarify how research interests have shifted over time and highlight areas where further investigation is needed. Even though our analysis focused only on titles, it provides a useful overview of emerging themes and the current direction of the field. Continued efforts, especially within Japan, will be important for advancing understanding and improving patient care.