Our interdisciplinary research group is known for applying maching learning to address problems arising in healthcare practice. We support clinicians in decision making process. We mindfully attend to problem understanding, data quality, and data visualization to facilitate cross-disciplinary collaborations. Active projects are:
- Predicting Hospital Readmissions, Mortality and Utilization
- Personalized Recommendation of Clinical Interventions
- Health Information Search Behavior on the Web
In the past we partnered with Nursing and Healthcare Leadership Program (http://www.tacoma.uw.edu/healthcare-leadership/healthcare-leadership), to offer two courses on campus to advance health informatics education -- Health Informatics I: Fundamentals (THLEAD 405 ) and Health Informatics II: Databases and Data Analysis (THLEAD 406).
30-Day Readmission Risk Management for Congestive Heart Failure
CHF is a leading cause of hospital admission and many of these hospitalizations are readmission within a small window. The Center of Medicare and Medicaid Services have considered 30-day readmission rate as a quality of metric. Many such readmissions could be prevented, if appropriate interventions are designed and administered on time. The objective of this project is to design appropriate statistical analyses or sequence mining techniques to suggest appropriate interventions to reduce 30-day readmission risk for CHF patients. This project would use Multicare Health System dataset for experimentation and validation. The results would be validated using data mining techniques, such as, precision, accuracy, recall, AUC, as well as by conducting case studies involving physicians, care managers, and data consultants.