BACKGROUNDScholarly biomedical publications report on the findings of a analysis investigation. Scientists use a well-established discourse construction to narrate their work to the state of the artwork, categorical their very own motivation and hypotheses and report on their strategies, outcomes and conclusions.
In earlier work, we’ve proposed methods to explicitly annotate the construction of scientific investigations in scholarly publications. Here we current the means to facilitate computerized entry to the scientific discourse of articles by automating the recognition of 11 classes on the sentence stage, which we name Core Scientific Concepts (CoreSCs).
These embrace: Hypothesis, Motivation, Goal, Object, Background, Method, Experiment, Model, Observation, Result and Conclusion. CoreSCs present the construction and context to all statements and relations inside an article and their computerized recognition can enormously facilitate biomedical info extraction by characterizing the different sorts of info, hypotheses and proof out there in a scientific publication.
RESULTSWe have skilled and in contrast machine studying classifiers (help vector machines and conditional random fields) on a corpus of 265 full articles in biochemistry and chemistry to routinely acknowledge CoreSCs. We have evaluated our computerized classifications in opposition to a manually annotated gold customary, and have achieved promising accuracies with ‘Experiment’, ‘Background’ and ‘Model’ being the classes with the very best F1-scores (76%, 62% and 53%, respectively).
We have analysed the duty of CoreSC annotation each from a sentence classification in addition to sequence labelling perspective and we current an in depth characteristic analysis. The most discriminative options are native sentence options reminiscent of unigrams, bigrams and grammatical dependencies whereas options encoding the doc construction, reminiscent of part headings, additionally play an necessary function for some of the classes. We focus on the usefulness of routinely generated CoreSCs in two biomedical functions in addition to work in progress.
Technology-induced errors. The present use of frameworks and fashions from the biomedical and life sciences literatures.
OBJECTIVEThe goal of this paper is to look at the extent, vary and scope to which frameworks, fashions and theories coping with technology-induced error have arisen in the biomedical and life sciences literature as listed by Medline®.
METHODSTo higher perceive the state of work in the realm of technology-induced error involving frameworks, fashions and theories, the authors performed a search of Medline® utilizing chosen key phrases recognized from seminal articles in this analysis space.
Articles have been reviewed and these pertaining to frameworks, fashions or theories coping with technology-induced error have been additional reviewed by two researchers.RESULTSAll articles from Medline® from its inception to April of 2011 have been searched utilizing the above outlined technique. 239 citations have been returned. Each of the abstracts for the 239 citations have been reviewed by two researchers. Eleven articles met the factors primarily based on summary evaluation.
These 11 articles have been downloaded for additional in-depth evaluation. The majority of the articles obtained describe frameworks and fashions as regards to theories developed in different literatures outdoors of healthcare.
The papers have been grouped into a number of areas. It was discovered that articles drew primarily from three literatures: 1) the human elements literature (together with human-computer interplay and cognition), 2) the organizational habits/sociotechnical literature, and 3) the software program engineering literature.
CONCLUSIONSA selection of frameworks and fashions have been discovered in the biomedical and life sciences literatures. These frameworks and fashions drew upon and prolonged frameworks, fashions and theoretical views which have emerged in different literatures. These frameworks and fashions are informing an rising line of analysis in well being and biomedical informatics involving technology-induced errors in healthcare.