Consumers’ Use of UMLS Concepts on Social Media: Diabetes-Related Textual Data Analysis in Blog and Social Q&A Sites

Abstract

BackgroundThe widely known terminology gap between health professionals and health consumers hinders effective information seeking for consumers.ObjectiveThe aim of this study was to better understand consumers’ usage of medical concepts by evaluating the coverage of concepts and semantic types of the Unified Medical Language System (UMLS) on diabetes-related postings in 2 types of social media: blogs and social question and answer (Q&A).MethodsWe collected 2 types of social media data: (1) a total of 3711 blogs tagged with “diabetes” on Tumblr posted between February and October 2015; and (2) a total of 58,422 questions and associated answers posted between 2009 and 2014 in the diabetes category of Yahoo! Answers. We analyzed the datasets using a widely adopted biomedical text processing framework Apache cTAKES and its extension YTEX. First, we applied the named entity recognition (NER) method implemented in YTEX to identify UMLS concepts in the datasets. We then analyzed the coverage and the popularity of concepts in the UMLS source vocabularies across the 2 datasets (ie, blogs and social Q&A). Further, we conducted a concept-level comparative coverage analysis between SNOMED Clinical Terms (SNOMED CT) and Open-Access Collaborative Consumer Health Vocabulary (OAC CHV)—the top 2 UMLS source vocabularies that have the most coverage on our datasets. We also analyzed the UMLS semantic types that were frequently observed in our datasets.ResultsWe identified 2415 UMLS concepts from blog postings, 6452 UMLS concepts from social Q&A questions, and 10,378 UMLS concepts from the answers. The medical concepts identified in the blogs can be covered by 56 source vocabularies in the UMLS, while those in questions and answers can be covered by 58 source vocabularies. SNOMED CT was the dominant vocabulary in terms of coverage across all the datasets, ranging from 84.9% to 95.9%. It was followed by OAC CHV (between 73.5% and 80.0%) and Metathesaurus Names (MTH) (between 55.7% and 73.5%). All of the social media datasets shared frequent semantic types such as “Amino Acid, Peptide, or Protein,” “Body Part, Organ, or Organ Component,” and “Disease or Syndrome.”ConclusionsAlthough the 3 social media datasets vary greatly in size, they exhibited similar conceptual coverage among UMLS source vocabularies and the identified concepts showed similar semantic type distributions. As such, concepts that are both frequently used by consumers and also found in professional vocabularies such as SNOMED CT can be suggested to OAC CHV to improve its coverage.

Enriching consumer health vocabulary through mining a social Q&A site: A similarity-based approach

“…Answers , we collected a total of 58,422 questions and their corresponding best answers (one for each question) in the diabetes category, and 81,433 questions and answers in the cancer category. Park et al [23] recently assessed the UMLS terminology coverage in the same dataset but did not find a significant difference between the questions and the answers with respect to the frequently-used terms and semantic types. Therefore, we combined each question with its corresponding best answer as one document in the corpus.…”

“…According to a recent work of ours [23], we used the 12 most frequent semantic types to construct 12 semantic context sub-features that can cover more than 80% of the UMLS terms in our dataset: “Amino Acid, Peptide, or Protein,” “Body Part, Organ, or Organ Component,” “Disease or Syndrome,” “Finding,” “Medical Device,” “Organic Chemical,” “Pharmacologic Substance,” “Sign or Symptom,” “Therapeutic or Preventive Procedure.” “Finding,” “Pharmacologic Substance,” and “Disease or Syndrome.” The method for calculating these sub-features is similar to that for the feature POS context features. We give a concrete example of the semantic context feature in Table S2 in the Supplementary Material.…”

“…Also a previous study found that the diabetes-related questions and answers posted on Yahoo! Answers has a reasonably good coverage of UMLS concepts [23]. In short, findings from our study will provide a foundation for developing automated and domain-agnostic methods to identify new consumer health terms from consumer-generated textual corpora.…”

Analysis of temporal constraints in qualitative eligibility criteria of cancer clinical studies

“…In our previous work [18], we employed a well-known technique called n-gram to extract terms from a word sentence for UMLS concept matching. Specifically, after sentence detection, tokenization, and POS tagging, we applied a fuzzy matching method to match an n-gram to a UMLS term.…”

Plain-language medical vocabulary for precision diagnosis

“…Patients could contribute computable phenotype data themselves; however, the ‘terminology gap’ between medical professionals and patients is a hindrance. The terminology gap has also limited patient participation both in research studies and in clinical phenotyping 9 . Current patient vocabularies (such as the Consumer Health Vocabulary Initiative; see URLs) provide broad consumer equivalents for clinical findings, medical procedures and equipment, but are not well integrated with research terminologies; thus, they provide neither the structure nor the coverage required for translational research and diagnostic tools.…”