Predicting First-Episode Venous Thromboembolism Risk Using a Supervised Regression Random Forest Model

Abstract

Venous thromboembolism (VTE), the formation of blood clots in deep veins, kills over 100,000 people in the United States. Many of these deaths occur due to the fact that VTE is not diagnosed until the patient is in critical condition. While AI models have been explored for predicting the risk of recurrent venous thromboembolism (rVTE), there is a paucity of research using these models to predict the risk of first-episode venous thromboembolism. This study compared several AI approaches to identify the superior method for VTE risk stratification. It was determined that a supervised regression random forest machine learning model would be the optimal choice for this task, given its numerous complex factors. While this model can be potentially used in clinical settings, further research must be done in order to determine its accuracy and applicability.