Comparative Analysis of Topical Evolution Patterns and Temporal Trends of Hypertension Research

Exploring the topical evolution patterns and temporal trends of hypertension can promote knowledge communication among experts, and is of great significance to understand the profile and frontiers of chronic disease. Current popular topic detection mainly focuses on two directions: one is based on social network analysis (SNA), the other is based on the topic models. Aiming at distinguishing their similarities and differences, this paper adopts the community detection method and expanded topic model Dirichlet-multinomial regression (DMR) respectively to detect the topic distribution and evolution trends of hypertension research. A total of 26,717 articles in the PubMed database were used as examples to construct the MeSH Terms co-occurrence matrix. It is found that hypertension literature is mainly concentrated on three communities and five research topics. MeSH Terms obtained from SNA are more specific and clearer, while the DMR has an advantage in exploring the evolution patterns of various themes.