About
I'm a PhD candidate in my fifth year of studies in the department of Computer and Information Science at the University of Pennsylvania. I work with my advisor, Dr. Insup Lee, as a member of the PRECISE Center, specifically the Smart Alarm research group. In particular I'm interested in applying machine learning to high-frequency, multi-source physiologic data to improve clinical care. In particular, I'm working on designing and implementing smarter alarm systems and clinical decision support systems that will help doctors and nurses detect and prevent adverse medical effects in the ICU, and operate safety and effectively in the hospital environment.
Research Projects
Decision Support System Development: In the modern hospital, patients are often connected to a multitude of medical devices, each of which records some data about the patient. Much of this data is currently inaccessible, or is underutilized. As part of our efforts to improve hospital care, I work with doctors and nurses to identify specific use-cases in which better access to clinical data provides a promising possibility of increased care. Once use-cases are identified, I work with clinicians to determine what data is currently being collected, how it is being stored, and how we can gather that data and apply straightforward machine learning techniques develop clinical decision support systems (CDSS), software systems that provide as tools for use in clinical care.
Machine Learning on Physiologic Data: Utilizing the aformentioned large amount of patient data generated in hospitals requires the ability to extract patterns from and learn over physiologic data. These data take the form of high to medium-frequency, temporal waveforms, and are often influenced by external care events. My research incorporates investigations into novel mechanisms for utilizing these physiologic data streams in machine learned models to improve CDSS.
Generic Clinical Decision Support Systems In working with clinicians to develop clinical decision support sytems, we have identified design methodologies and implementation commonalities that we see as being key to the development of robust CDSSs. The Generic Smart Alarm and Generic Clinical Decision Support Systems represent the establishment of reconfigurable pipelines and component libraries to facilitate the development of smarter alarms and CDSSs, to allow them to be quickly customized and reused in creation of these systems.
Publications
Wandering Data: A Scalable, Durable System for Effective Visualization of Patient Health Data Alexander Roederer, Andrew King, Sanjian Chen, Margaret Mullen-Fortino, Soojin Park, Oleg Sokolsky, Insup Lee IEEE Computer Based Medical Systems 2014 Poster to Appear
Prediction of Significant Vasospasm in Aneurysmal Subarachnoid Hemorrhage Using Automated Data Alexander Roederer, John H. Holmes, Michelle J. Smith, Insup Lee, Soojin Park Neurocritical Care 2014
A Survey of Active Learning for Classification of Medical Signals Alexander Roederer University of Pennsylvania Written Preliminary Examination II Presented November 2012
Clinical Decision Support for Integrated Cyber-Physical Systems: A Mixed Methods Approach Alex Roederer, Andrew Hicks, Enny Oyeniran, Insup Lee and Soojin Park IHI 2012 Demo Presented January 2012
Challenges and Research Directions in Medical Cyber-Physical Systems. Insup Lee, Oleg Sokolsky, Sanjian Chen, John Hatcliff, Eunkyoung Jee, BaekGyu Kim, Andrew L. King, Margaret Mullen-Fortino, Soojin Park, Alexander Roederer, Krishna K. Venkatasubramanian Proceedings of the IEEE, 2012
Limitations of Threshold-Based Brain Oxygen Monitoring for Seizure Detection Soojin Park, Alexander Roederer, Ram Mani, Sarah Schmitt, Peter D. LeRoux, Lyle H. Ungar, Insup Lee and Scott E. Kasner Neurocritical Care, November 2011
GSA: a framework for rapid prototyping of smart alarm systems. Andrew L. King, Alex Roederer, David Arney, Sanjian Chen, Margaret Mullen-Fortino, Ana Giannareas, William Hanson III, Vanessa Kern, Nicholas Stevens, Jonathan Tannen, Adrian Viesca Trevino, Soojin Park, Oleg Sokolsky, Insup Lee IHI 2010
Demo of the Generic Smart Alarm: a framework for the design, analysis, and implementation of smart alarms and other clinical decision support systems. Andrew L. King, Alex Roederer, Sanjian Chen, Nicholas Stevens, Philip Asare, Oleg Sokolsky, Insup Lee, Margaret Mullen-Fortino, Soojin Park Wireless Health 2010
Divvy: An ATP Meta-system Based on Axiom Relevance Ordering. Alex Roederer, Yury Puzis, Geoff Sutcliffe CADE 2009
Teaching
TA for CIS 400 Taught by Insup Lee, Fall 2011 and Spring 2012
TA for CIS 160 Taught by Jean Gallier, Fall 2010 Was honored with the 2011Penn Prize for Excellence in Teaching by Graduate Students.
Professional
Undergraduate Student Researcher at NASA Ames Research Center Summer 2009
Intern, Software Testing for ACAS/Altimeter Groups at Rockwell Collins Summer 2007, Summer 2008
Education
University of Pennsylvania - PhD Candidate, Computer and Information Science
Research: Machine Learning, Decision Support, and Smart Alarms in Critical Healthcare scenarios, with Dr. Insup Lee, with the advise of Dr. Soojin Park.
2009-Present
University of Miami - B.S. Computer Science, B.S. Applied Mathematics
Research Area: Latent Semantic Analysis and Automated Theorem Provers, with Professor Geoff Sutcliffe.
2005-2009