Fighting Sepsis with Machine Learning

Dr. Bart Abban, VigiLanz

Sepsis is a major cause of in-hospital mortality; however, early diagnosis can reduce the risk of death considerably and improve patient outcomes. Early diagnosis of sepsis continues to be a challenge as most of the existing diagnostic criteria suffer from high false positive rates. Other diagnostic criteria deliver better true positive rates at the expense of timeliness. Machine learning-based algorithms offer potential improvements in the early diagnosis of sepsis but their adoption in clinical care have been slow.

VigiLanz is a leading clinical decision support company serving over 400 hospitals across the US. The company has adopted RapidMiner to integrate machine learning and advanced analytics into its top-ranked clinical decision support suite. Drawing on our experience with sepsis, we discuss the issues that are critical to building accurate clinical machine learning models and ensuring their adoption. Join us to learn how we are accelerating the development and deployment of customized predictive models at multiple hospitals.