Advanced Risk Stratification and Prediction of Venous Thromboembolism in Critically Ill Patients

Embargo until
Date
2021-05-11
Journal Title
Journal ISSN
Volume Title
Publisher
Johns Hopkins University
Abstract
Venous Thromboembolism (VTE) is a disease responsible for more than 100,000 deaths a year in the U.S. Thru early detection, this mortality rate can be decreased as the early administration of therapeutic heparin can prevent VTE from being fatal. However, this can be challenging in a surgical ICU setting where symptoms can hard to distinguish from common ICU patient symptoms. In this work, we developed a risk prediction model and a novel real time classification model to determine the risk of VTE. The risk prediction model analyzes patient demographic and history data to determine if they are at a high risk. Meanwhile, the real time classification model analyzes high frequency physiological time series data to determine if a patient is currently experiencing a VTE. The findings from this study and model will be implemented as a screening tool to assist clinicians in determining which patients require additional care.
Description
Keywords
Healthcare, Machine Learning, AI, Artificial Intelligence, Data Science, Biomedical Engineering, BME, Clinical Care Medicine
Citation