Informatics Research Seminar: Using Relational Learning for Preventing Heart Attacks

November 9 @ 4:00 – 5:00 pm

 

Speaker: Sriraam Natarajan, PhD
Presented from Duke University


Abstract:

Coronary heart disease(CHD) is a major cause of death and illness worldwide. In the U.S. CHD is responsible for approximately 1 in every 6 deaths with a coronary event occurring every 25 seconds and about 1 death every minute based on data current to 2007. In this talk, Dr. Natarajan will present the results of applying novel machine learning algorithms in two problems: First is the task of identifying young adults at high risk for developing CHD in middle and later life using the data collected from CARDIA study. Second is the task of predicting the occurrences of Myocardial Infarction (MI) in adults from real Electronic Health Records(EHR).

Since EHRs are inherently relational and noisy, there is a need to employ methods that go beyond traditional learning algorithms. Most traditional Artificial Intelligence (AI) methods are based on one of two approaches: first-order logic, which excels at capturing the rich relationships among many objects, and statistical representations, which handle uncertain environments and noisy observations.  Statistical relational learning (SRL), an area of growing interest, seeks to unify these approaches in order to handle problems that are both complex and involve uncertainty. Dr. Natarajan will present one such scalable learning algorithm and its application to the above two problems.

Biosketch:

Dr. Sriraam Natarajan is currently an Assistant Professor in the Translational Science Institute and School of Bio-Medical Engineering and Sciences of  Wake Forest University School of Medicine.  He was previously a Post-Doctoral Research Associate at the Department of Computer Science at University of Wisconsin-Madison. He graduated with his PhD from Oregon State University working with Dr. Prasad Tadepalli. His research interests lie in the field of Artificial Intelligence, with emphasis on Machine Learning, Statistical Relational Learning, Reinforcement  Learning, Graphical Models and Bio-Medical Applications.