Heart Failure Treatment Analysis

Project Overview

Congestive heart failure (CHF) is a chronic condition with substantial morbidity and mortality. CHF currently affects more than 6.6 million people in the United States, at a yearly medical care cost of approximately $25 billion. Current estimates project a 25% increase in incidence and a 300% increase in cost by 2030. Therapeutic options for CHF are complex and evolving, and evidence-based clinical practice guidelines have been developed to optimize care. Unfortunately, guideline compliance varies substantially among sites and patients, leading to variable care with suboptimal cost and outcomes. Measuring and tracking this variation could be an important method of optimizing CHF care and promoting guideline use, but reliable and objective metrics that express the extent of care variation and deviation from guidelines in patient populations are lacking. The PTL is investigating the development and evaluation of such metrics. [1][2]



  1. [vedomske2013scalable] Vedomske, M., M. Gerber, D. Brown, and J. Harrison, "Scalable and Locally Applicable Measures of Treatment Variation that Use Hospital Billing Data", 12th International Conference on Machine Learning and Appli- cations (ICMLA): IEEE Press, 12/2013.
  2. [vedomske2013random] Vedomske, M., D. Brown, and J. Harrison, "Random Forests on Ubiquitous Data for Heart Failure 30-Day Readmissions Prediction", 12th International Conference on Machine Learning and Applications (ICMLA), Miami, Florida, IEEE, 12/2013.