Distributed word representations have become an essential foundation for biomedical natural language processing (BioNLP), text mining and information retrieval. Word embeddings are traditionally computed at the word level from a large corpus of unlabeled text, ignoring the information present in the internal structure of words or any information available in domain specific structured resources such as ontologies. However, such information holds potentials for greatly improving the quality of the word representation, as suggested in some recent studies in the general domain. Here we present BioWordVec: an open set of biomedical word vectors/embeddings that combines subword information from unlabeled biomedical text with a widely-used biomedical controlled vocabulary called Medical Subject Headings (MeSH). We assess both the validity and utility of our generated word embeddings over multiple NLP tasks in the biomedical domain. Our benchmarking results demonstrate that our word embeddings can result in significantly improved performance over the previous state of the art in those challenging tasks.

No abstract

Background Although mass vaccination against COVID-19 may prove to be the most efficacious end to this deadly pandemic, there remains concern and indecision among the public towards vaccination. As pregnant and reproductive-aged women account for a large proportion of the population with particular concerns regarding vaccination against COVID-19, this survey aims at investigating their current attitudes and beliefs within our own institution. Objective To understand vaccine acceptability among pregnant, non-pregnant and breastfeeding respondents and elucidate factors associated with COVID-19 vaccine acceptance. Methods We administered an anonymous online survey to all women (including patients, providers and staff) at our institution assessing rates of acceptance of COVID-19 vaccination. Respondents were contacted in one of three ways: by email, advertisement flyers and distribution of QR codes at virtual townhalls regarding the COVID-19 vaccine. Based on their responses, respondents were divided into three mutually exclusive groups: (1) non-pregnant respondents (2) pregnant respondents and (3) breastfeeding respondents. The primary outcome was acceptance of vaccination. Prevalence ratios were calculated to ascertain the independent effects of multiple patient-level factors on vaccine acceptability. Results The survey was administered from January 7, 2021 to January 29, 2021 with 1,012 respondents of whom 466 (46.9%) identified as Non-Hispanic White, 108 (10.9%) as Non-Hispanic Black, 286 (28.8%) as Hispanic, and 82 (8.2) as Non-Hispanic Asian. The median age was 36 (IQR 25-47) years. Of all the respondents, 656 respondents (64.8%) were non-pregnant, 216 (21.3%) were pregnant and 122 (12.1%) were breastfeeding. There was no difference in chronic comorbidities when evaluated as a composite variable (Table 1). 390 respondents (39.2%) reported working in healthcare. Non-pregnant respondents were most likely to accept vaccination (457 respondents, 76.2%, p<0.001) with breastfeeding respondents the second most likely (55.2%). Pregnant respondents had the lowest rate of vaccine acceptance (44.3%, p<0.001). Prevalence ratios revealed all non-White races except for non-Hispanic Asian respondents and Spanish speaking respondents were less likely to accept vaccination(Table 3). Working in healthcare was not found to be associated with vaccine acceptance among our cohort. Conclusions and Relevance In this survey study of only women at a single institution, pregnant respondents of non-White or non-asian races were more likely to decline vaccination compared to non-pregnant and breast-feeding respondents. Working in healthcare was not associated with vaccine acceptance.