By Jof Enriquez,
Follow me on Twitter @jofenriq
IBM is creating a computing system, called a neural network and powered by the company's TrueNorth chip, that can study the brain wave patterns that indicate a seizure. The company hopes that such a system could someday be built into mobile devices which can warn, and possibly even prevent, epileptic seizures.
Tech companies like Google and Facebook already are using neural networks to perform computing tasks like image and voice recognition, but they run them via the cloud and on large machine-learning data centers. In contrast, IBM's postage stamp-sized TrueNorth chip consumes over 1,000 times less power than a conventional processor of similar size, making it ideal to run on existing mobile devices. Moreover, it's built using a new architecture patterned after neurons of the human brain, making it more efficient.
The advanced machine intelligence software in the chip has myriad uses potentially, but researcher Stefan Harrer and his team at IBM Research Australia are focusing on how the chip can support a neural network that is capable of predicting seizures, based on what it learns from actual EEG readings from epilepsy patients, reports Wired.
"We’re trying to extract all the meaningful information from all the background noise. We want to be able to detect a specific seizure for a specific patient," Harrer told the online magazine.
He and his team plan to use the chip with an external computer, and eventually a wearable device, that will work in tandem with a brain implant. The implant would send EEG information to the TrueNorth device, which would then use the data to predict the possibility of an epileptic seizure, and warn the patient beforehand, according to Digital Trends.
"We want to do this on a wearable system that you put on a subject — on a patient — and have it do analysis in real-time, 24/7. That’s the only way this technology will have an impact beyond cool research papers," Harrer told Wired.
Later, such a device could be sophisticated enough to be capable not only of detecting an impending seizure attack, but ultimately preventing it by sending electrical impulses.
Existing device-based therapies to treat epileptic seizures also rely on electric stimulation.
FDA approved in 2013, the RNS Stimulator is implanted under the scalp and delivers electrical impulses to normalize brain activity. The device has been shown to decrease the frequency of seizures in epileptic patients. Companies such as Medtronic, St. Jude, NeuroPace, and others, are developing and marketing neurostimulators that address neurological disorders.
Meanwhile, other researchers are exploring other technologies to treat these disorders. A National Institutes of Health (NIH) study is looking at how biodegradable silk brain implants can prevent seizures, while researchers at the University of California Irvine are testing how optogenetic lasers targeted at specific areas of the brain can inhibit seizure activity.
The BRAIN (Brain Research through Advancing Innovative Neurotechnologies) Initiative is a public/private collaboration with the goal of devising technologies that will advance our understanding of the brain, including the development of intelligent devices to track and treat abnormal brain activity in people with epilepsy, Alzheimer’s, Parkinson’s, autism, and other neurological disorders.