Our early work focused on (1) knowledge representation of patient safety events using biomedical ontology and (2) automatic detection of critical information from patient safety events (e.g., error types, severity, and contributing factors) using clinical Natural Language Processing and machine learning. These studies are among the first that leveraged informatics methods for analyzing large-scale patient safety events to improve data interoperability and clinical workflow.
Current research is centered on multimodal health data integration, predictive modeling, and EHR-based data mining to be used for augmenting Clinical Decision Support, diagnosis, screening, comorbidity detection, prognosis prediction, and evidence-based intervention. Since 2020, our research was directed to the development of biomedical informatics methods and applications tailored for combating the COVID-19 pandemic. Much of the recent work was done with the Big Data Health Science Center at the University of South Carolina.