Tyler Nowak
Pronouns:
Research Mentor(s): Matthew Hodgman
Research Mentor School/College/Department: Department of Computational Medicine and Bioinformatics / Medicine
Program:
Authors: Tyler Nowak, Yufeng Zhang, Matthew Hodgman, Emily Wittrup, Kayvan Najarian
Session: Session 3: 11:00 am – 11: 50 am
Poster: 13
Abstract
The increasing number of cardiovascular diseases globally necessitated the development of more efficient and effective diagnostic tools. The early detection of pulmonary embolism remains a challenging yet vital step in improving patient treatment and management. In response to this challenge, our project employs rule-based interpretable machine learning methods applied to electronic health records data from Michigan Medicine. The aim is to develop an early-warning model for identifying patients at risk of pulmonary embolism. This project involves comparing these interpretable models with traditional machine learning counterparts, using a suite of evaluation metrics including accuracy, specificity, sensitivity, recall, Area Under the Receiver Operating Characteristic curve (AUROC), and Area Under the Precision-Recall Curve (AUPRC). Additionally, the project will extract and analyze clinical rules learned by the models, thereby outlining the risk profile of patients and enabling more focused intervention strategies.