Identifying risk factors for metabolic syndrome in biomedical text.

Identifying risk factors and biomarkers for diseases is an important aspect of biomedical research. However, much of the underlying information resides in the research literature and is not available in executable form. We propose a methodology based on automatic semantic interpretation (using SemRep) to capture risk factors and biomarkers for diseases asserted in MEDLINE citations. In this initial study, we focus on metabolic syndrome. The performance of SemRep in identifying risk factors and biomarkers for this disorder was 53% recall (CI, 44% to 62%) and 67% precision (CI, 62% to 72%). We discuss how the information captured could assist clinicians in finding current and new risk factors for metabolic syndrome as well as diseases predisposed by this disorder. The availability of this information in executable form can support guideline development and the timely translation of biomedical research into improvements in quality of patient care.