The purpose of this study was to determine risk factors and create a machine learning model to predict in-hospital DVT after total laparoscopic or vaginal hysterectomy in patients with endometrial, ovarian, and cervical cancer. This study analyzed 134,649 admissions using the Nationwide Inpatient Sample. Multivariable regression revealed that increased age and comorbidity were associated with increased DVT risk, while obesity and race were not. The random forest predicted in-hospital DVT with AUROC of 0.83.

Learning Objective 1: Understand the application and efficacy of machine learning models to predict adverse surgical outcomes in a large, national database.


Justin Kleiner, Brown University
Michael Cohen, Brown University
Dilum Aluthge, Brown University
Ishan Sinha, Brown University
Indra Sarkar (Presenter)
Brown University

Elizabeth Chen, Brown University
Dario Roque, Brown University

Presentation Materials: