Abstract

The increased use of electronic health records has made possible theautomated extraction of medical policies from patient records to aid in thedevelopment of clinical decision support systems. We adapted a boostedStatistical Relational Learning (SRL) framework to learn probabilistic rulesfrom clinical hospital records for the management of physiologic parameters ofchildren with severe cardiac or respiratory failure who were managed withextracorporeal membrane oxygenation. In this preliminary study, the resultswere promising. In particular, the algorithm returned logic rules for medicalactions that are consistent with medical reasoning.

 

Quick Read (beta)