Automatic discovery and processing of EEG cohorts



The medical records of one individual may contain several different types of information such as image files and written lab results. Multiply the information available in one file by thousands of patients and the result is a vast quantity of data that potentially contains valuable information. For example, researchers could compare treatment options for a particular diagnosis in order to determine their effectiveness. However, varying formats of the information poses a challenge in comparing data.

Faculty from Temple University and the University of Texas-Dallas aim to resolve some of these issues. Dr. Joseph Picone, professor, Department of Electrical and Computer Engineering at Temple University, Dr. Iyad Obeid, associate professor, Department of Electrical and Computer Engineering at Temple University, and Dr. Sanda Maria Harabagiu, professor, Department of Computer Science at the University of Texas - Dallas, will be examining approximately 25,000 EEG sessions, within a twelve year period, from patients at Temple University Hospital. Drs. Picone, Obeid and Harabagiu aim to automatically annotate the EEG events and process the text using clinical language processing techniques. Ultimately, the information will be organized in the Qualified Medical Knowledge Graph (QMKG), which will be built using BigData solutions through MapReduce. The overall goal is for medical professionals and biomedical researchers to be able to access EEG information for the purpose of conducting comparative research.

The research is supported by a grant through the National Institutes of Health, Big Data to Knowledge (BD2K) program.