Using sentiment analysis to identify similarities and differences in research topics and medical subject headings (MeSH terms) between Medicine (Baltimore) and the Journal of the Formosan Medical Association (JFMA) in 2020: A bibliometric study. Issue 11 (18th March 2022)

• Record Type: Journal Article
• Authors: Lin, Ju-Kuo; Chien, Tsair-Wei; Yeh, Yu-Tsen; Ho, Sam Yu-Chieh; Chou, Willy
• Other Names: Gupta Poonam section editor.
• Is Part Of: Medicine. Volume 101:Issue 11(2022)
• Publication Date: 2022-03-18
• Subjects: cluster analysis -- forest plot -- latent topic -- MeSH term -- sentiment analysis -- social network analysis; Medicine -- Periodicals; Medicine -- Periodicals; Médecine -- Périodiques; Geneeskunde; Medicine; Periodicals; Periodicals; 610.5
• DOI: 10.1097/MD.0000000000029029
• Languages: English
• ISSNs: 0025-7974

Abstract

Abstract: Background: Little systematic information has been collected about the nature and types of articles published in 2 journals by identifying the latent topics and analyzing the extracted research themes and sentiments using text mining and machine learning within the 2020 time frame. The goals of this study were to conduct a content analysis of articles published in 2 journals, describe the research type, identify possible gaps, and propose future agendas for readers. Methods: We downloaded 5610 abstracts in the journals of Medicine (Baltimore) and the Journal of the Formosan Medical Association (JFMA) from the PubMed library in 2020. Sentiment analysis (ie, opinion mining using a natural language processing technique) was performed to determine whether the article abstract was positive or negative toward sentiment to help readers capture article characteristics from journals. Cluster analysis was used to identify article topics based on medical subject headings (MeSH terms) using social network analysis (SNA). Forest plots were applied to distinguish the similarities and differences in article mood and MeSH terms between these 2 journals. The Q statistic and I 2 index were used to evaluate the difference in proportions of MeSH terms in journals. Results: The comparison of research topics between the 2 journals using the 737 cited articles was made and found that most authors are from mainland China and Taiwan in Medicine and JFMA, respectively, similarity is supported by observing the abstract mood ( Q = 8.3, I 2 = 0, P = .68; Z = 0.46, P = .65), 2 journals are in a common cluster (named latent topic of patient and treatment) using SNA, and difference in overall effect was found by the odds ratios of MeSH terms ( Q = 185.5 I 2 = 89.8, P < .001; Z = 5.93, P < .001) and a greater proportion of COVID-19 articles in JFMA. Conclusions: SNA and forest plots were provided to readers with deep insight into the relationships between journals in research topics using MeSH terms. The results of this research provide readers with a concept diagram for future submissions to a given journal. Highlights: The main approaches frequently used in Meta-analysis for drawing forest plots contributed to the following: Comparing abstract mood in 2 journals, which is modern and innovative in the literature. Extracting article topics from MeSH terms using SNA, drawing visual representations by using SNA, choropleth map, and forest plots that can inspire other relevant research to replicate the approaches for the other 2 paired journals in comparison of differences in research topics in the future.