Several machine learning classifiers were applied to electronic health records (EHR) for predictive modeling of post surgical complications. During the course of a hospital stay the number of variables accumulate allowing us to leverage information for early prediction of occurrences of complications. However missing values for some variables are a prevalent issue for EHR data.  Data collection standards, data quality, and interoperability will be a crucial element of successful data-driven probabilistic risk monitoring tools in hospital settings.  Leveraging electronic health records for predictive modeling of post-surgical complications

CSTAT collaborator: Marianne Huebner