Patient Similarity Learning through Distance Metric Learning and Interactive Visualization
Department of Computer Science Special Seminar
Monday, March 25, 2013
10:00 AM Central Time
Room 2405 Siebel Center
Abstract: Heterogeneous and large volume of Electronic Health Records (EHR) data are becoming available in many healthcare institutes. Many healthcare applications such as clinical decision support and population management require robust and intuitive data mining algorithms to analyze these data. Patient similarity is a suite of such algorithms that quantitatively measures how similar patients are to each other based on their EHR data in a given clinical context.
I will present my research in learning patient similarity measures that address the following challenges:
- How to leverage physician feedback into the similarity computation?
- How to integrate multiple patient similarity measures into a single consistent similarity measure?
- How to incrementally update the existing patient similarity functions as new data or feedback arrive?
- How to present the similarity results in an intuitive and interactive way to users?
I will present patient similarity learning as a core component of a large-scale healthcare analytic research platform that we are building. The core of the patient similarity is the combination of supervised distance metric learning algorithms and visualization techniques. I will illustrate the effectiveness of our proposed algorithms for patient similarity learning in several different healthcare scenarios. Finally, I will demonstrate an interactive visual analytic system that allows users to efficiently cluster data and to refine the underlying patient similarity metric.
Bio: Jimeng Sun is a research staff member at Healthcare Analytic Department of IBM TJ Watson Research Center. He leads research projects of medical informatics, especially in developing large-scale predictive and similarity analytics on healthcare applications. Sun has extensive research track records on data mining research: specialized in healthcare analytics, big data analytics, similarity metric learning, social network analysis, predictive modeling and visual analytics. He has published over 70 papers, filed over 20 patents (4 granted). He has received ICDM best research paper in 2007, SDM best research paper in 2007, and KDD Dissertation runner-up award in 2008. Sun received his B.S. and M.Phil. in Computer Science from Hong Kong University of Science and Technology in 2002 and 2003, and PhD in Computer Science in Carnegie Mellon University in 2007, specialized on data mining on stream, graphs and tensor data. His advisor was Prof. Christos Faloutsos.