Toward automatic recognition of high quality clinical evidence

AMIA Annu Symp Proc. 2008 Nov 6;2008:368.

Abstract

Automatic methods for recognizing topically relevant documents supported by high quality research can assist clinicians in practicing evidence-based medicine. We approach the challenge of identifying articles with high quality clinical evidence as a binary classification problem. Combining predictions from supervised machine learning methods and using deep semantic features, we achieve 73.5% precision and 67% recall.

MeSH terms

  • Algorithms*
  • Artificial Intelligence*
  • Canada
  • Evidence-Based Medicine*
  • MEDLINE*
  • Natural Language Processing*
  • Pattern Recognition, Automated / methods*
  • Periodicals as Topic*
  • Quality Control
  • Semantics
  • Vocabulary, Controlled