Editor’s Choice Articles for May

Abstract

This electronic health record study used data from over 27,000 admissions (1). The mortality rate was 1.8%, and ICU length of stay was less than 120 hours for three quarters of the admissions. In this work, the authors propose a modeling methodology for monitoring patient improvement and deterioration in real-time. The accompanying editorial gives us a context for such work and provides a helpful summary of characteristics that readers need to focus on when evaluating machine learning based clinical prediction models (2).