bims-librar Biomed News
on Biomedical librarianship
Issue of 2021‒11‒21
twenty-one papers selected by
Thomas Krichel
Open Library Society

  1. Health Info Libr J. 2021 Nov 18.
      The article discusses the experiences of Italian librarians taking part in an institutional project to produce a new general-public-oriented health web portal. The web portal was set up to provide verified and easily understandable health information, and to debunk health-related fake news circulating on the internet. The different roles, knowledge and skills acquired in during the project this are discussed and show how the librarian's knowledge and skills were of fundamental importance for the success of the Web Portal. By collaborating with other professions new skills such as social media management, video creation and Search Engine Optimization were gained, which enhanced the information literacy role of the service.
    Keywords:  embedded librarian; health education; health information literacy; librarian skills; professional development
  2. Health Info Libr J. 2021 Nov 18.
      BACKGROUND: Artificial intelligence (AI) offers a promising solution to expedite various phases of the systematic review process such as screening.OBJECTIVE: We aimed to assess the accuracy of an AI tool in identifying eligible references for a systematic review compared to identification by human assessors.
    METHODS: For the case study (a systematic review of knowledge translation interventions), we used a diagnostic accuracy design and independently assessed for eligibility a set of articles (n = 300) using human raters and the AI system DistillerAI (Evidence Partners, Ottawa, Canada). We analysed a series of 64 possible confidence levels for the AI's decisions and calculated several standard parameters of diagnostic accuracy for each.
    RESULTS: When set to a lower AI confidence threshold of 0.1 or greater and an upper threshold of 0.9 or lower, DistillerAI made article selection decisions very similarly to human assessors. Within this range, DistillerAI made a decision on the majority of articles (93-100%), with a sensitivity of 1.0 and specificity ranging from 0.9 to 1.0.
    CONCLUSION: DistillerAI appears to be accurate in its assessment of articles in a case study of 300 articles. Further experimentation with DistillerAI will establish its performance among other subject areas.
    Keywords:  artificial intelligence; precision; recall; review, systematic
  3. J Public Health Manag Pract. 2022 Jan-Feb 01;28(1):28(1): E9-E15
      CONTEXT: Shortly after the first COVID-19 case in the United States was identified in Washington, the Washington State Department of Health (WA DOH) determined that real-time knowledge of scientific findings related to SARS-CoV-2 was critical for an effective response. Epidemiologists at the WA DOH established the Daily Literature Situation Report (Lit Rep), within the agency's incident management team, to support public health and state leaders in evidence-based decision making. However, from January to May, the scale of the pandemic response and daily volume of emerging information grew beyond the capacity of the WA DOH epidemiology team tasked with gathering, reviewing, summarizing, and disseminating it daily.OBJECTIVE: To ensure public health leaders maintained awareness of the rapidly evolving scientific literature during the pandemic to support evidence-based practice.
    DESIGN: The WA DOH contracted the University of Washington (UW) Alliance for Pandemic Preparedness to assemble a team of faculty and students to continue producing the Lit Rep.
    MAIN OUTCOME: In addition to the daily Lit Rep, the UW team developed in-depth reports addressing questions from public health leadership and further evolved the methodology for the daily reports to support long-term sustainability and broader accessibility.
    RESULTS: Throughout its existence, the Lit Rep had summarized more than 4300 articles from more than 150 000 citations and had more than 5600 subscribers from public health practice, academia, and the general public, both domestic and international.
    CONCLUSIONS: The flexible Lit Rep model sets a standard for responding to emerging public health threats and communicating complex scientific information to government leaders, public health staff, and other interested parties. The WA DOH and the UW have exemplified how a mutually beneficial partnership can be established to support more effective public health practice based on real-time evidence both during a crisis and potentially for future public health challenges.
  4. Acad Med. 2021 Nov 16.
      PROBLEM: Understanding and communicating medical advances driven by basic research, and acquiring foundational skills in critically appraising and communicating translational basic research literature that affects patient care, are challenging for medical students to develop.APPROACH: The authors developed a mandatory course from 2012 to 2018 at Texas A&M University College of Medicine to address this problem. Medical Student Grand Rounds (MSGR) trains first-year students to find, critically assess, and present primary research literature about self-selected medically relevant topics. With basic science faculty mentoring, students completed milestones culminating in oral presentations. Students learned to search literature databases then choose a clinical subject using these skills. They outlined the clinical subject area background and a mechanistic research topic into a clinical problem based on deeper evaluation of primary research literature. "Mechanistic" was defined in this context as experimental evidence that explained the "how" and "why" underlying clinical manifestations of a disease. Students received evaluations and feedback from mentors about discerning the quality of information and synthesizing information on their topics. Finally, students prepared and gave oral presentations, emphasizing the primary literature on their topics.
    OUTCOMES: In the early stages of the course development, students had difficulty critically assessing and evaluating research literature. Mentored training by research-oriented faculty, however, dramatically improved student perceptions of the MSGR experience. Mentoring helped students develop skills to synthesize ideas from basic research literature. According to grades and self-evaluations, students increased proficiency in finding and interpreting research articles, preparing and delivering presentations, and understanding links among basic and translational research and clinical applications.
    NEXT STEPS: The authors plan to survey fourth-year students who have completed MSGR about their perceptions of the course in the context of clinical experiences in medical school to guide future refinements.
  5. Afr Health Sci. 2021 Jun;21(2): 904-911
      Background: The use of point-of-care, evidence-based tools is becoming increasingly popular. They can provide easy-touse, high-quality information which is regularly updated and has been shown to improve clinical outcomes. Integrating such tools into clinical practice is an important component of improving the quality of health care. However, because such tools are rarely used in resource-limited settings, there is limited research on uptake especially among medical students.Objective: This paper explores the uptake of one such tool, Up-To-Date, when provided free of cost at a medical school in Africa.
    Methods: In partnership with the Better Evidence at Ariadne Labs free access to UpToDate was granted through the MakCHS IP address. On-site librarians facilitated training sessions and spread awareness of the tool. Usage data was aggregated, based on log ins and content views, presented and analyzed using Excel tables and graphs.
    Results: The data shows evidence of meaningful usage, with 43,043 log ins and 15,591 registrations between August 2019 and August 2020. The most common topics viewed were in obstetrics and gynecology, pediatrics, drug information, and infectious diseases. Access occurred mainly through the mobile phone app.
    Conclusion: Findings show usage by various user categories, but with inconsistent uptake and low usage. Librarians can draw upon these results to encourage institutions to support uptake of point-of-care tools in clinical practice.
    Keywords:  Makerere University College of Health Sciences, Uganda; UpToDate clinical decision support tool
  6. Health Info Libr J. 2021 Nov 14.
      In this article, Ms Yihang Chen with her supervisor, Prof Lihong Zhou, reports on her MA in Library Science study aimed to identify user requirements of library research support services (RSS) at the universities of Traditional Chinese Medicine (TCM) in China. This study adopted an inductive qualitative approach, employed as a case study and 14 TCM researchers and academic librarians using semi-structured interviews. The research findings point to 28 RSS requirements in five main themes: mastering, planning, project, publication and electronic preservation stages. Although this research is situated in China, it has implications for libraries worldwide in supporting research into holistic and indigenous medicine. LZ.
    Keywords:  China; case studies; library services; research support; research, qualitative
  7. IEEE Access. 2020 ;8 165201-165215
      This article addresses the problem of detecting misleading information related to COVID-19. We propose a misleading-information detection model that relies on the World Health Organization, UNICEF, and the United Nations as sources of information, as well as epidemiological material collected from a range of fact-checking websites. Obtaining data from reliable sources should assure their validity. We use this collected ground-truth data to build a detection system that uses machine learning to identify misleading information. Ten machine learning algorithms, with seven feature extraction techniques, are used to construct a voting ensemble machine learning classifier. We perform 5-fold cross-validation to check the validity of the collected data and report the evaluation of twelve performance metrics. The evaluation results indicate the quality and validity of the collected ground-truth data and their effectiveness in constructing models to detect misleading information.
    Keywords:  COVID-19; Coronavirus; SARS-CoV-2; WHO; fake news detection; infodemic; misleading information; pandemic; social media; social networks; text classification; text mining; web mining
  8. J Gastrointest Oncol. 2021 Oct;12(5): 2045-2051
      Background: The number and complexity of clinical trials is growing, making finding and accessing trials increasingly challenging. A key barrier to increasing recruitment efficiency is patients' low awareness of trials as an option. They rarely hear about trials from oncologists, unless these are based at a research center, which introduces trial population biases. Even if patients decide to search for trial information on their own, it is difficult for laypeople to understand. Trial search tools exist, but have multiple shortcomings: complex navigation, limited search functionality, data too complex for non-specialists, and next steps unclear.Methods: Access to trials via a novel trial search tool using technology, including artificial intelligence (AI), to restructure trial information and match patients to trials was the focus of this study. We focused on 6 gastrointestinal cancers for evaluating the search tool's AI model in which their trial information was restructured based on key eligibility differentiators identified. Research participants were recruited with the support of patient associations leveraging social media. They were asked to complete a survey evaluating an established tool and the novel tool to find trials relevant for them. They assessed aspects of the tools' usability on 5-level Likert scales.
    Results: Survey respondents had a base level of awareness of trials, with patient associations and the internet as leading information sources. The novel tool made it easier to find trials, led to an improvement in ease of understanding information presented and provided more clarity on what the next steps towards enrollment would be. Overall, this led to higher patient satisfaction.
    Conclusions: Novel patient-focused tools can unlock trial access for all patients. Democratizing trial information will not only increase trial accrual, but also patient satisfaction while reducing disparities. This tool could also be of value for academic or community oncologists who face similar issues when navigating trials for patients.
    Keywords:  Artificial intelligence (AI); cancer disparities; clinical trials; health equity; patient access
  9. J Endourol. 2021 Nov 15.
      BACKGROUND: Patients suffering from progressive symptoms of benign prostatic hyperplasia (BPH) increasingly turn to the internet for information on invasive and minimally invasive surgical therapies (MIST). This study aimed to identify online articles accessed by patients seeking guidance on said options and to evaluate the quality and readability of their content.METHODS: Social media analytics tool Buzzumo was used to identify the most shared articles on Facebook, Twitter, Reddit, and Pinterest related to BPH surgical treatments from 2016 to 2021. The content quality was graded using the DISCERN tool; a questionnaire that assesses the quality of written information on treatment choices for a health problem. An online Automatic Readability Checker was used to evaluate the readability of the written transcripts. Google Trends data was used to evaluate search term popularity for BPH procedures from 2016 to 2021. Google trends data and DISCERN ratings were analyzed with ANOVA and paired t-test, respectively.
    RESULTS: Thirty-nine percent of identified articles were rated as low quality, 39% as moderate quality, and 22% as high quality using the DISCERN tool, with no significant difference among reviewer ratings (p = 0.0561). The median reading level of all articles was 12th grade. Google Trends data illustrated a rising popularity of MIST such as Rezūm and Urolift, exceeding interest in more invasive procedures like Holmium laser enucleation of the prostate (HoLEP) and laser surgery. Transurethral resection of the prostate (TURP) remained popular with no significant difference between MIST and TURP on Google Trends (p = 3.46).
    CONCLUSIONS: The most accessed articles on BPH treatments have important shortcomings including risks of treatment and other available options for treatment. Article transcripts exceed the reading level of the average adult patient. TURP has remained a popular procedure online, but there is a notable rising interest in MIST, especially Urolift.
  10. BMJ Open Sport Exerc Med. 2021 ;7(4): e001203
      Patients increasingly access the internet to learn about their orthopaedic conditions. Despite this, online information may be unregulated, of questionable quality and difficulty to read.Objectives: Therefore, this study aimed to evaluate the readability and quality of the online information concerning subacromial impingement syndrome.
    Methods: A search using Australia's three most popular online search engines was undertaken using the search terms 'subacromial impingement syndrome' and 'shoulder impingement'. The first 15 websites for each term were evaluated. Duplicates, advertisements and sponsored links were removed.The quality and readability of each website were calculated using the DISCERN and Flesch-Kincaid Reading Ease (FKRE) tools, respectively. The differences in quality and readability between each website type (healthcare/academic, commercial, news outlet, charitable/not-for-profit, layperson, government) was assessed using analysis of variance. The correlation between quality and readability was assessed using the Pearson correlation coefficient.
    Results: The majority of 35 unique websites analysed were of 'poor'/'fair' quality (determined via the DISCERN instrument) and 'difficult' readability (per the FKRE tool), with no correlation established between the scores. There was no statistically significant difference in quality across website types, however layperson, news outlet and government websites were found to be significantly more readable than alternate website categories (p<0.05).
    Conclusions: We determined that much of the online information concerning subacromial impingement syndrome may be difficult to read and/or of poor quality. By recognising the shortcomings of information accessed by patients online, it is hoped clinicians may be prompted to better educate their patients.
    Keywords:  injuries; orthopaedics; shoulder
  11. Blood Coagul Fibrinolysis. 2021 Nov 18.
      Access to health-related internet information has increased dramatically during the past decade. A significant proportion of this information has been demonstrated to be set at too high a level for general comprehension. The goal of this study was to evaluate the readability and quality of information available on the internet relating to deep vein thrombosis (DVT). A search for 'DVT' and 'Deep Vein Thrombosis' was performed on five most commonly accessed search engines. Top hundred websites were reviewed. Website authorship was classified. Each site was assessed using the recognized quality and readability scoring systems. The presence of the health on the net foundation code of conduct (HON-code), a reported quality-assurance marker, was noted. The majority of web sites (45%) were academic or physician compiled. Only 36% of the websites were HON-code certified. HON-code certified sites achieved significantly greater DISCERN and Journal of the American Medical Association (JAMA) scores. Seventy-three percent of the websites were above the recommended sixth-grade readability level. Flesch--Kincaid readability test (FRES) of more than half of the websites (56%) was below 60, which makes them fairly difficult to read by general public. Internet information relating to DVT is of variable quality and largely set at an inappropriate readability level. Given this variability in quality, healthcare providers should direct patients to known sources of reliable, readable online information. Identification of reliable sources may be aided by known markers of quality such as HONcode certification and to educate them to use academic and physician-provided sites which have been shown to contain better-quality information. Moreover, academic and physician-compiled websites should be written in a fashion appropriate for general public consumption.
  12. Braz J Anesthesiol. 2021 Nov 15. pii: S0104-0014(21)00389-4. [Epub ahead of print]
      INTRODUCTION: There are many possible sources of medical information; however, the quality of the information varies. Poor quality or inaccurate resources may be harmful if they are trusted by providers. This study aimed to analyze the quality of coronavirus disease 2019 (COVID-19)-related intubation videos on YouTube.METHODS: The term "COVID-19 intubation" was searched on YouTube. The top 100 videos retrieved were sorted by relevance and the 37 videos were included. The video demographics were recorded. The quality of the videos was analyzed using an 18-point checklist, which was designed for evaluating COVID-19 intubation. Videos were also evaluated using general video quality scores and the modified Journal of the American Medical Association score.
    RESULTS: The educational quality was graded as good for eight (21.6%) videos, moderate for 13 (35.1%) videos, and poor for 16 (43.2%) videos. The median safe COVID-19 intubation score (SCIS) was 11 (IQR = 5-13). The SCISs indicated that videos prepared in an intensive care unit were higher in quality than videos from other sources (p < 0.05). The length of the video was predictive of quality (area under the curve = 0.802, 95% CI = 0.658-0.945, p = 0.10).
    CONCLUSIONS: The quality of YouTube videos for COVID-19 intubation is substandard. Poor quality videos may provide inaccurate knowledge to viewers and potentially cause harm.
    Keywords:  Airway management; COVID-19; Coronavirus; Hand washing; Intubation
  13. JMIR Cancer. 2021 Nov 18. 7(4): e25783
      BACKGROUND: In China, the internet has become one of the most important ways to obtain information about breast cancer. However, quantitative evaluations of the quality of Chinese health websites and the breast cancer treatment information they publish are lacking.OBJECTIVE: This study aimed to evaluate the quality of Chinese breast cancer websites and the value, suitability, and accuracy of the breast cancer treatment information they publish.
    METHODS: Chinese breast cancer health websites were searched and manually screened according to their Alexa and Baidu search engine rankings. For each website included in the survey, which was conducted on April 8, 2019, the three most recently published papers on the website that met the inclusion criteria were included for evaluation. Three raters assessed all materials using the LIDA, DISCERN, and Suitability Assessment of Materials (SAM) tools and the Michigan Checklist. Data analysis was completed with the Statistical Package for Social Sciences (SPSS) version 20.0 and Microsoft Excel 2010.
    RESULTS: This survey included 20 Chinese breast cancer websites and 60 papers on breast cancer treatment. The LIDA tool was used to evaluate the quality of the 20 websites. The LIDA's scores of the websites (mean=54.85, SD 3.498; total possible score=81) were low. In terms of the layout, color scheme, search facility, browsing facility, integration of nontextual media, submission of comments, declaration of objectives, content production method, and robust method, more than half of the websites scored 0 (never) or 1 (sometimes). For the online breast cancer treatment papers, the scores were generally low. Regarding suitability, 32 (53.33%) papers were evaluated as presenting unsuitable material. Regarding accuracy, the problems were that the papers were largely not original (44/60, 73%) and lacked references (46/60, 77%).
    CONCLUSIONS: The quality of Chinese breast cancer websites is poor. The color schemes, text settings, user comment submission functions, and language designs should be improved. The quality of Chinese online breast cancer treatment information is poor; the information has little value to users, and pictorial information is scarcely used. The online breast cancer treatment information is accurate but lacks originality and references. Website developers, governments, and medical professionals should play a full role in the design of health websites, the regulation of online health information, and the use of online health information.
    Keywords:  Chinese language websites; breast cancer; online health information; quantitative evaluations
  14. Front Sports Act Living. 2021 ;3 692613
      Social media platforms are an accessible and increasingly used way for the public to gather healthcare-related information, including on sports injuries. "TikTok" is currently one of the fastest-growing social media platforms worldwide, and it is especially popular amongst adolescents and young adults. The widespread use and popularity of TikTok suggests that this platform has potential to be a source for healthcare information for younger individuals. The aim of this study was to gain a preliminary understanding of the concussion/head injury-related information on TikTok, and to gauge if TikTok could serve as a platform for concussion education. This exploratory study used a systematic search strategy to understand more about how concussion is being portrayed through TikTok videos. Using the keywords "concussion" and "head injury," 200 videos were downloaded from TikTok and 43 videos were excluded. Of the 92 videos retrieved using the keyword "concussion," 95% (n = 88) had more than 100,000 views and 6% (n = 10) had been viewed more than 10 million times. Over half, 54% (n = 50) of the "concussion" videos depicted individuals "playing around" and getting hit in the head, whilst only 1% (n = 1) of the TikTok videos were categorized as "explaining concussion facts." The large numbers of views of concussion-related TikTok videos demonstrates the popularity of this platform and indicates that healthcare organizations should consider TikTok as a potential means for concussion education amongst younger individuals.
    Keywords:  concussion; head injury; healthcare education; public health; social media
  15. Rev Esp Enferm Dig. 2021 Nov 16.
      Introduction Health literacy assumes an increasing relevance as the disseminated access to online contents may impact community health. This study aims to evaluate the suitability of online content on benign proctological disorders. Methods The platforms Google© and YouTube™ were searched using the keywords hemorrhoidal disease, anal fistula, anal fissure, anal pruritus, and fecal incontinence. Each page was analyzed for: interactivity, credibility, readability, and content adequacy. For each video the content, reactions, and comments of the users were analyzed. Three validated scales were applied - Simple Measure of Gobbledygook (SMOG); Suitability Assessment of Materials (SAM), and quality criteria for consumer health information on treatment choices (DISCERN). Results A total of 127 webpages and 113 videos were evaluated. Forty webpages (31.50%) included images and 10 (37.04%) had informative videos. The SMOG scale presented an average value of 9.91±0.94 which estimates the need of 10 years of schooling for comprehension and 64 webpages (52.07%) presented a reading level higher than the recommended B1. The mean SAM score was 39.46±11.2% and the DISCERN scale averaged 40.33±5.69, corresponding to a reasonable content. Only 20 pages (15.75%) included interactivity tools and 44 (34.64%) contained bibliographic references. The videos posted by patients (n=67, 59.2%) had the highest number of views and positive comments whereas health professionals' videos (n=21; 18.75%) had the highest proportion of requests for additional information. Discussion Online content regarding benign proctological disorders is globally inappropriate for the health literacy level of the Portuguese population, at the writing and reading levels.
  16. J Craniofac Surg. 2021 Nov 12.
      OBJECTIVE: The purpose of this study was to determine the content quality and accuracy of the YouTube videos about the Botox application during gummy smile treatment and to analyze the efficacy of these videos in informing the healthcare professionals and the patients.MATERIALS AND METHODS: The first 200 videos resulting from the search on YouTube by using the keywords "gummy smile/Botox," which were determined by Google Trends, were evaluated in the study. One hundred fourteen out of these 200 videos were included in the study. The videos were calculated using these descriptive features. A content scale consisting of 8 categories was used to determine the content levels of the videos and the video information and quality index scale was used to determine the quality level.
    RESULTS: The most mentioned subjects in the videos were "Description" of the gummy smile and Botox treatment (71.9%), "Advantage" of Botox treatment (51.8%), "Procedure" (48.2%), and "Prognosis" (47.4%). The "Contraindication" was mentioned rarely (2.6%). A statistically significant difference was found between the videos with different content levels in terms of Description, Advantage, Indication, Procedure, Complication, Prognosis, and Cost. Likewise, there was a statistically significant difference between videos in distinct content level groups in terms of video information and quality index total score.
    CONCLUSIONS: The quality of the YouTube videos about Botox applications in gummy smile treatment was found to be low in general. It is important for the physician who will perform the procedure to explain the procedure to the patients in detail. In addition, YouTube was determined to be insufficient for accessing high-quality information for students, assistants, and specialist physicians.
  17. J Cancer Educ. 2021 Nov 16.
      Cancer information services (CISs) can play an important role within the pathway of cancer information seeking, but so far, this role is not well understood. Callers (n = 6,255) who contacted the largest provider of cancer information in Germany participated in a survey in which they reported their information sources, information level, and needs leading to the call. Persons with prior information from a physician (n = 1,507) were compared to people with prior online information (n = 901) and people with prior information from both sources (n = 2,776). Nearly all callers (96.7%) stated prior sources, while physicians and the Internet were the most frequently reported sources. People, who only talked to a doctor before, are more likely to be a patient and in the disease stages during/after the first treatment or with recurrence than prior Internet users. The two groups do not differ in their prior information level but did differ in their information needs. CISs are an important supplement to other sources, while the information repertoire depends on patients' stages during the cancer journey. Specific characteristics and needs of callers with different prior information sources help to individualize the service of CISs and similar providers.
    Keywords:  Cancer; Cancer information seeking; Cancer information service; Information sources; Pathway
  18. Lang Resour Eval. 2021 ;55(2): 525-549
      Medieval documents are a rich source of historical data. Performing named-entity recognition (NER) on this genre of texts can provide us with valuable historical evidence. However, traditional NER categories and schemes are usually designed with modern documents in mind (i.e. journalistic text) and the general-domain NER annotation schemes fail to capture the nature of medieval entities. In this paper we explore the challenges of performing named-entity annotation on a corpus of Spanish medieval documents: we discuss the mismatches that arise when applying traditional NER categories to a corpus of Spanish medieval documents and we propose a novel humanist-friendly TEI-compliant annotation scheme and guidelines intended to capture the particular nature of medieval entities.
    Keywords:  Annotation scheme; Historical NER; Medieval Spanish corpus; Medieval named entities; Named-entity annotation
  19. Comput Biol Med. 2021 Nov 06. pii: S0010-4825(21)00798-8. [Epub ahead of print] 105004
      In the last years, the rise of digital technologies has enormously augmented the possibility for people to access health information and consult online versions of Patient Information Leaflets (PILs), enabling them to improve their knowledge about medication and adherence to therapies. However, health information may often be difficult to consult and comprehend due to an excessively lengthy and undersized text, coupled with the presence of many incomprehensible medical terms. To face these issues, this paper proposes a conversational agent as a valuable solution to simplify health information retrieval and improve health literacy in Italian by codifying PILs and making them query-able in natural language. In particular, the system has been devised to: i) comprehend natural language questions on medicines of interest; ii) proactively ask the user or automatically infer from the dialog state all the missing information necessary to generate an answer; iii) extract the answer from a structured knowledge base built from PILs of registered drugs. An experimental study has been carried out to evaluate both the performance and usability of the proposed system. Results showed an adequate ability of the system to handle most of the dialogues started by participants correctly, good users satisfaction, and, thus, proved its feasibility and usefulness.
    Keywords:  Chatbots; Conversational agent; Health literacy; Medical information; Natural language interaction
  20. BMC Med Inform Decis Mak. 2021 Nov 16. 21(Suppl 9): 317
      BACKGROUND: A lot of medical mentions can be extracted from a huge amount of medical texts. In order to make use of these medical mentions, a prerequisite step is to link those medical mentions to a medical domain knowledge base (KB). This linkage of mention to a well-defined, unambiguous KB is a necessary part of the downstream application such as disease diagnosis and prescription of drugs. Such demand becomes more urgent in colloquial and informal situations like online medical consultation, where the medical language is more casual and vaguer. In this article, we propose an unsupervised method to link the Chinese medical symptom mentions to the ICD10 classification in a colloquial background.METHODS: We propose an unsupervised entity linking model using multi-instance learning (MIL). Our approach builds on a basic unsupervised entity linking method (named BEL), which is an embedding similarity-based EL model in this paper, and uses MIL training paradigm to boost the performance of BEL. First, we construct a dataset from an unlabeled large-scale Chinese medical consultation corpus with the help of BEL. Subsequently, we use a variety of encoders to obtain the representations of mention-context and the ICD10 entities. Then the representations are fed into a ranking network to score candidate entities.
    RESULTS: We evaluate the proposed model on the test dataset annotated by professional doctors. The evaluation results show that our method achieves 60.34% accuracy, exceeding the fundamental BEL by 1.72%.
    CONCLUSIONS: We propose an unsupervised entity linking method to the entity linking in the medical domain, using MIL training manner. We annotate a test set for evaluation. The experimental results show that our model behaves better than the fundamental model BEL, and provides an insight for future research.
    Keywords:  Medical entity linking; Multiple instance learning; Unsupervised learning
  21. Bioinformatics. 2021 Nov 12. pii: btab740. [Epub ahead of print]
      MOTIVATION: Significant effort has been spent by curators to create coding systems for phenotypes such as the Human Phenotype Ontology (HPO), as well as disease-phenotype annotations. We aim to support the discovery of literature-based phenotypes and integrate them into the knowledge discovery process.RESULTS: PheneBank is a Web-portal for retrieving human phenotype-disease associations that have been text-mined from the whole of Medline. Our approach exploits state-of-the-art machine learning for concept identification by utilising an expert annotated rare disease corpus from the PMC Text Mining subset. Evaluation of the system for entities is conducted on a gold-standard corpus of rare disease sentences and for associations against the Monarch initiative data.
    AVAILABILITY: The PheneBank Web-portal freely available at Annotated Medline data is available from Zenodo at DOI: 10.5281/zenodo.1408800. Semantic annotation software is freely available for non-commercial use at GitHub:
    SUPPLEMENTARY INFORMATION: Supplementary data is available at Bioinformatics online.