Slides and Video Available

page ccdDavid Page, PhD, Professor, Department of Biostatistics & Medical Informatics, School of Medicine and Public Health, University of Wisconsin Madison, Machine Learning for Predictive Phenotyping from EHR Data,” at 11:00 am on Friday, April 17, 2015, in Rooms 407A/B BAUM, 5607 Baum Blvd., The Offices at Baum.

Abstract: One view of a clinical data warehouse, holding a snapshot of EHR data, is as a relational database with patient information distributed over tables for diagnoses, labs, vitals, prescriptions, procedures, etc. Another view is of each patient as an irregularly-sampled timeline of events. Yet most machine learning algorithms expect each patient to be represented as a feature vector–a single row of a spreadsheet. All known methods for converting patient data into this restrictive format lose information or alter the frequencies in the data on which machine learning algorithms critically depend. This talk will discuss algorithms for machine learning directly from the data warehouse in one of its two native views.  Tasks addressed include learning phenotype definitions, learning to predict future phenotypes for a patient, and causal discovery in the particular context of finding adverse drug events.

Biography:  Dr. Page obtained a B.A. degree in Political Science from Clemson University, a M.S. degree in Computer Science from Clemson University, and a Ph.D. degree in Computer Science from the University of Illinois. He currently is a professor in the Department of Biostatistics and Medical Informatics at the University of Wisconsin at Madison. His current research is on algorithms for data mining and machine learning, and their applications to biomedical data, especially clinical and high-throughput genetic and other molecular data. Of particular interest are inductive logic programming (ILP) and other multi-relational learning techniques capable of dealing with complex data points (such as molecules or clinical histories) and producing logical rules (such as the rules in green to the side).