This $717,712 Project Grant from the National Science Foundation's Division of Information and Intelligent Systems, under the Computer and Information Science and Engineering program (CFDA 47.070), will support the development and evaluation of a knowledge-enhanced and interpretable radiology report generation framework using longitudinal, multimodal electronic health record data and domain knowledge.
The awardee, Joan & Sanford I Weill Medical College of Cornell University, will build a memory-enhanced reporting system to generate chest x-ray reports using historical images and reports. A radiology-specific knowledge graph will be constructed from heterogeneous EHR data and integrated into the framework. A rationale-based model supporting interpretability will also be developed. Finally, a prototype user-centered reporting system with a graphic user interface will be built and evaluated to enhance communication between radiologists and referring physicians. The research aims to employ data science techniques to support next-generation medical diagnostic reasoning from structured EHR data.
Generated 1/7/24, 3:21 AM