Endocrine Abstracts

Introduction/Aim: The diagnosis and management of hypothyroidism in pregnancy is a domain of shifting paradigms, with changes in guidelines, which may have led clinicians to confusion or angst (Endocrine Today; October 2019). Since sentiment analysis (SA) of medical texts can be implemented (Artificial Intelligence in Medicine 2015; 64: 17–27) we sought to use SA regarding this clinical issue.

Materials/Methods: We collected the English language abstracts of review articles in PubMed from 2011 to November 2021, using the Medical Subject Headings (MeSH) terms "hypothyroidism", "pregnancy" and "human" and performed SA with an online artificial intelligence tool (courtesy of Prof. Daniel Soper, California State University, Fullerton, CA, USA; https://www.danielsoper.com/sentimentanalysis/default.aspx). The results were evaluated by year with the Kruskal-Wallis and Chi square tests.

Results: From 2011 to 2015 a slight trend from negative to positive sentiment (P = 0.08) was noted in the literature studied, while from 2018 wide and significant (P = 0.04) sentiment fluctuations were noted by year.

Discussion: Regarding hypothyroidism and pregnancy uncertainties remain vis-à-vis screening and management bearing in mind the need to optimize perinatal outcomes. Researchers continue to debate the very definition of subclinical hypothyroidism, which can differ between nonpregnant or pregnant states. This situation is reflected in the sentiment of the published literature, especially in review articles, usually by experts in the field worldwide, which try to provide a critical evaluation of the data that is available from existing studies.