Abstract: Postoperative bleeding (POB) following colorectal surgery (CRS) was predicted using data found in EHR. Several machine learning (ML) methods were trained and evaluated; demonstrating their utility compared to traditional logistic regression. All methods yielded area under the curves between 0.79 and 0.86. ML methods were able to identify novel risk factors which could lead to improved understanding of the causes of POB.

Learning Objective 1: predicting postoperative bleeding following colorectal surgery

Learning Objective 2 (Optional): Using electronic health data for predictive modeling


David Chen (Presenter)
Mayo Clinical

Naveed Afzal, Mayo Clinic
Sunghwan Sohn, Mayo Clinic
Elizabeth Habermann, Mayo Clinical
James Naessens, Mayo Clinical
David Larson, Mayo Clinic
Hongfang Liu, Mayo Clinic

Presentation Materials: