Automated Data Acquisition for Heart Failure

This study was realized in the context of the ADAHF (Automated Data Acquisition for Heart Failure) project, a U.S. Department of Veterans Administration (VA) project that aims to automatically extract data for heart failure treatment performance measures from clinical notes (PI: Jennifer Garvin). These performance measures include left ventricular ejection fraction (LVEF) assessments (their mention and measured values), medications (ACEIs and ARBs), or reasons not to administer these medications. They are part of the Joint Commission National Hospital Quality Measures Heart Failure Core Measure Set the VA has adopted. To automate the extraction of these performance measures and automatically detect patients not treated according to recommendations, we developed the CHIEF (Congestive Heart failure Information Extraction Framework), a NLP application that allowed detecting patients with unmet heart failure treatment performance measures with 98.9% sensitivity and 98.7% positive predictive value.

ADAHF

To extend its applicability for heart failure treatment quality improvement, CHIEF was then adapted to detecting more medication types and more general left ventricular systolic function information. This adaptation was realized in the context of a CREATE VA project in collaboration with the Houston, Nashville, Madison, Chicago, and Salt Lake City VA Medical Centers. It also included use of the automatically extracted clinical information to offer clinicians patient-specific treatment recommendation messages.

Finally, CHIEF was adapted for operational use in the context of the VA Health Management Platform (HMP) project, in collaboration with FirstView Consultants, and with important user-centered design efforts for the extracted information presentation.

Publications:

  • Kim, Y., Garvin, J., Heavirland, J., & Meystre, S. M. (2015). Improving Detection of Reasons Not to Take a Medication by Leveraging Medication Prescription Status (p. 1528). AMIA Annu Symp Proc.
  • Meystre, S. M., Kim, Y., Heavirland, J., Williams, J., Bray, B. E., & Garvin, J. H. (2015). Heart Failure Medications Detection and Prescription Status Classification in Clinical Narrative Documents. Studies in Health Technology and Informatics, 216, 609–613. 
  • Kim, Y., Garvin, J., Goldstein, M. K., & Meystre, S. M. (2015). Classification of Contextual Use of Left Ventricular Ejection Fraction Assessments. Studies in Health Technology and Informatics, 216, 599–603. 
  • Meystre, S., Kim, Y., Redd, A., & Garvin, J. H. (2014). Congestive Heart Failure Information Extraction Framework (CHIEF) Evaluation (p. 86). AMIA Annu Symp Proc.
  • Kim, Y., Garvin, J., Heavirland, J., & Meystre, S. M. (2014). Automatic Clinical Note Type Classification for Heart Failure Patients (p. 182). AMIA Summits Transl Sci Proc, CRI.
  • Kim, Y., Garvin, J. H., Heavirland, J., & Meystre, S. M. (2013). Relatedness Analysis of LVEF Qualitative Assessments and Quantitative Values (p. 123). AMIA Summits Transl Sci Proc, CRI.
  • Kim, Y., Garvin, J. H., Heavirland, J., & Meystre, S. M. (2013). Improving heart failure information extraction by domain adaptation., 185–189.
  • Garvin, J., Heavirland, J., Weaver, A., Kim, Y., Bray, B. E., Bolton, D., et al. (2012). Determining Section Types to Capture Key Clinical Data. Proc AMIA Symp, 1745.
  • Meystre, S. M., Kim, J., & Garvin, J. (2012). Comparing Methods for Left Ventricular Ejection Fraction Clinical Information Extraction (p. 138). AMIA Summits Transl Sci Proc, CRI.