The School of Mathematical and Computational Sciences welcomes the campus community to a public research presentation by Catherine Inibhunu, candidate for tenure track position in Data Analytics, on Monday, March 11th at 3:00 pm in Room 101 of the Cass Science Hall.

Presentation Title: Temporal Pattern Recognition in Time Series Data Streams and its Application in Critical Care

Abstract: In critical care, complex systems and sensors continuously monitor patients physiological features resulting in significant amounts of data generated every second. It’s an immense challenge for anyone trying to utilize this time oriented data when making critical decisions about care of the patients. Temporal abstraction and data mining are two research fields that have tried to synthesis time oriented data and bring out an understanding on hidden relationships that may exist between time oriented data. In clinical settings, having the ability to understand hidden relationships in time series patient data as they unfold could help save a life by aiding in detection of conditions that are not obvious to clinicians and healthcare workers. Various researchers have looked at techniques for generating abstractions from clinical data, however, the variety and speed of data streams generated in healthcare often overwhelms the current systems which are not designed to handle such data. Another attempt has been to understand the complexity in time-series data utilizing data mining techniques, however, existing data mining models are not designed to handle temporal relationships that might exist in time series data. To address this challenge, this research proposes innovative techniques and methodologies in temporal pattern recognition augmented with frequent pattern mining to understand temporal behaviors and relationships in time series data. The research premise is that discovery of any hidden relationships and patterns in underlying time series data would be valuable in building a classification system that can automatically characterize physiological data streams as they are generated. Such a characterization could help in detection of normal and abnormal behaviors in patients that might be associated with life threatening conditions.

All are welcome.