Researchers from Mayo Clinic and the University of Pennsylvania announce the launch of the Seizure Detection Challenge, an international competition inviting the best minds in “machine learning” to improve devices to track and treat epilepsy. Researchers from Penn and Mayo Clinic have designed the challenge, which is hosted by Kaggle.com, an online community where data scientists come together to solve complex problems.
The unique data sets gathered during the challenge will be made freely available to researchers worldwide by the National Institutes of Health (NIH), University of Pennsylvania and Mayo Clinic at the International Epilepsy Electrophysiology Portal to advance epilepsy research and treatment.
“Predicting seizures, early and accurately is a first step toward controlling them and improving the quality of life for millions of patients with epilepsy worldwide”, said Mayo Clinic neurologist Gregory Worrell, M.D., Ph.D., a co-primary investigator of the project.
Contestants will analyze retrospective prolonged intracranial EEG data recorded from four dogs with naturally occurring epilepsy and from eight patients with medication-resistant seizures during evaluation for epilepsy surgery. The contestant or group that can identify the earliest EEG changes leading to seizures with the fewest false alarms wins.
“Leveraging the tremendous talent of scientists and the machine learning community worldwide has a great potential to help researchers and our patients. It’s an exciting new model for collaboration”, added co-primary investigator Brian Litt, MD, professor of Neurology and Bioengineering at the University of Pennsylvania and director of the new Penn Center for Neuroengineering and Therapeutics.
The contest runs until August 19, 2014, and offers $8,000 in prizes funded by the American Epilepsy Society and the National Institute of Neurological Disorders and Stroke. The challenge is also supported by the Neural Information Processing Society (NIPS).
The Mayo Clinic team also includes Ben Brinkman, PhD, from the Department of Neurology. In addition to Dr. Litt, the Penn team is co-led by Zack Ives, PhD, associate professor of Computer and Information Science, and includes Joost Wagenaar, PhD, from Neurology and Bioengineering and Charles Vite, DVM, PhD, associate professor of Veterinary Neurology, School of Veterinary Medicine. The dog data were provided by Neurovista and Ned Patterson DVM, PhD, University of Minnesota.
The work is supported by NINDS (NS063930) in addition to the American Epilepsy Society.