PhD 1 Feichen Shen PhD 1 Shulan Tian 1 David Chen 1 Hongfang Liu PhD 1 Department of Health Sciences Research Mayo Clinic Rochester MN 1 Background Reference Mining for Improving Cohort Establishment Method Consistency ID: 796127
Download The PPT/PDF document "Yiqing Zhao, MS 1 , Yanshan Wang," is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
Slide1
Yiqing Zhao, MS1, Yanshan Wang, PhD1, Feichen Shen, PhD1, Shulan Tian1, David Chen1, Hongfang Liu, PhD1 Department of Health Sciences Research, Mayo Clinic, Rochester, MN1
Background
Reference Mining for Improving Cohort Establishment Method Consistency
63 REP 2016 publications available on PMC websites at the time of experiment (June, 2017). Four college level reviewers was instructed to identify references that are relevant to the cohort establishment methods mentioned in the original paper. Automatic extraction results were compared with gold standard from manual literature review results.
Evaluation
Inter-rater agreement is 92.5%. Precision of 96.3%, recall of 85.7% and F-score of 0.88. A user interactive interface was built using D3.js library to visualize the reference map (shown at bottom).
Results
Methods
© 2016 Mayo Foundation for Medical Education and Research
Using reference mining and semantic filtering to identify previous work cited in the reference that studied the same cohort.
Cohort study is a popular study design aimed to evaluate population health and identify risk factors. Mayo Clinic researchers have used Rochester Epidemiology Project (REP), a medical record linkage system, to establish several famous cohorts. However, the increasing amount of REP studies has made it more and more difficult to keep track of the Cohort Establishment Methods (include cohort definition, confirmation and attribute extraction) used in each REP research. This may lead to data discrepancies and research irreproducibility in cohort studies.
REP 2016 Publication“Methods” “Material and Methods”…
JSoup
UMLS Semantic tagger
Disease Related Entities26704438: {Stroke, Myocardial Infarction, Herpes Zoster}17976353: {Herpes Zoster, Vaccine}17408489: {CHD, Myocardial Infarction}23664666: {Hypercalcemia}…
This framework can be used to facilitate further extraction of historic cohort establishment methods and create a standardized cohort establishment methods for future references.
Conclusions
PMID
StrokeMyocardial InfarctionHerpes ZosterVaccineCHDHypercalcemia26704438111179763531117408489112366466611
26704438 :Ref #1 PMID: 17976353, Ref #2 PMID: 17408489,Ref #3 PMID: 23664666…
Bag-of-Word Similarity Filter
Final Reference Set