In clinical domains, various types of free texts appear including generic clinical text, discharge summary, insurance claims, clinical notes, lab reports, radiology reports, pathology reports, and clinical trial reports. We can extract information from such text data using natural language processing (NLP) tools and deep learning.

Venous thromboembolism (VTE), which includes deep vein thrombosis (DVT) and pulmonary embolism (PE), is one of the top 10 causes of preventable hospital death.

In this project, we aim to develop an AI algorithm that provides precise and reliable estimates of VTE risk, which can be coupled with an evidence-based threshold for VTE prophylaxis recommendations. We have developed and demonstrated NLP and Deep Learning tools to identify postoperative VTE among surgical and non-surgical patients treated in VA hospitals, based on free text radiology reports.

We are collaborating with vascular surgeon experts at the Center for Vascular Research, University of Maryland School of Medicine.

 

Other Projects

Artificial Intelligence in Biomedical Imaging

FDA AI Testbed

NIH AIM-AHEAD Program

AI/ML for Marine Atmospheric Boundary Layer Profiles via Satellite Remote Sensing Data Fusion

Intelligently Secured Distributed Solution

Networking with Human in the Loop

Cyberphysical Security Research for Biopharma Production and Personalized Medicine