Skip to content
October 18 @ 11:30 AM - 1:00 PM - Mann Hall, Room 203
Machine learning techniques have been used for decades in a variety of quantitative domains in an attempt to make sense of the information surrounding us. In this talk, I will discuss an experimental system that I developed while working at the Microsoft Health Solutions Group. The system, called a prediction disposition system, is capable of predicting admission and discharge disposition situations for a patient visiting the emergency room and, depending on the availability of the data associated with the patient, of presenting the updated probability of a certain situation in real time. One objective of the system is to help medical staff to be better prepared logistically in terms of labor and equipment resources. Another objective is clinical: to notify a physician of the need to revise his or her diagnostics in those situations in which the clinician and the decision system disagree about whether the patient needs to be discharged or not. The paradigm for how to handle decision making in critical scenarios (high risk/high stake) that involve expert systems and humans is also the main area of interest of Stockato, LLC, a company that helps people make better investment decisions. The main developments to come out of Stockato’s product line are a machine learning method called signal composition, which classifies time series regardless of length, type, and quantity; and a supervised-learning enhancement called self-labeling supervised learning. In my talk, I will also show how apparently disparate areas such as finance and health care share similar types of problems and how they can benefit from the technologies we developed at Stockato.
Dr. Uri Kartoun, a member of IEEE, holds a PhD from the Department of Industrial Engineering and Management, Intelligent Systems program, at Ben-Gurion University, Israel. In 2005, Dr. Kartoun completed a one-year informatics fellowship at the Washington Hospital Center, where he worked in health-care robotics. From 2008 to 2012, Dr. Kartoun worked at the Microsoft Health Solutions Group in Washington, D.C., in the area of health-care informatics, developing technologies in machine-learning, natural language processing, and virtual reality to enhance current patient health records (PHR) based systems. Currently, Dr. Kartoun is the cofounder of a start-up company called Stockato, LLC, which helps people make better investment decisions. Dr. Kartoun’s primary research interests include health-care informatics, decision making, large-scale medical data mining, and health-care robotics. Dr. Kartoun was the recipient of multiple awards from Microsoft including the Microsoft Research Technology Transfer Award and the Microsoft Dynamics Innovation Lab Challenge Award. Dr. Kartoun has published in leading journals in the field of human-robot collaboration and holds seven patents related to machine learning.
March 10 @ 9:30 AM - 10:30 AM ABE 204
March 11 @ 9:00 AM - 1:15 PM MRGN, Room 121
March 11 @ 3:30 PM - 4:30 PM Pfendler Hall, Room 241 Deans Auditorium
Office of the Vice President for Research
610 Purdue Mall
West Lafayette, IN 47907-2040