Artificial Intelligence

Biography

Biography: Fuchiang (Rich) Tsui

Abstract

Now, more than ever, Electronic Health Records (EHRs) are generated in large quantities and in diverse contents. Th is
explosion of information has naturally enabled powerful patient data analyses to potentially improve healthcare. Th is talk
will review several on-going research projects focusing on predictive modeling from structured and unstructured EHR data
and real-time production systems deployed at hospitals that we have developed in the Department of Biomedical Informatics,
University of Pittsburgh (Pitt). We have developed Bayesian networks and utilized machine learning methods in conjunction
with natural language processing to predict 30-day hospital readmissions, detect infectious diseases from emergency
department visits, classify the severity of psychiatric reports; we will also report our pilot study on infant mortality predictive
modeling based on various EHR data and non-EHR information. We have developed several production systems deployed at
the University of Pittsburgh Medical Center (UPMC) that provide daily infl uenza surveillance reports, real-time laboratory
reporting and event-driven based 30-day readmission risk prediction; we also developed a national retail data monitor system
at Pitt, that monitors over-the-counter medicine sales on a daily basis from 30,000+ retail stores in the US. I will demonstrate
one of our currently deployed production systems, the System for Hospital Adaptive Readmission Prediction and Management
(SHARP). Th is is integrated into the EHR system in place at the Children's Hospital of Pittsburgh of UPMC and demonstrates
how the research we have developed can translate directly into practice.