Researchers report that they’ve used a mobile, brain-inspired processor to analyze brain signals from retrospective patient data and successfully predict an average of 69 percent of seizures across all patients with artificial intelligence.
The research could help pave the way for personalized seizure prediction for patients with epilepsy.
With a third of epilepsy patients worldwide currently living with unpredictable seizures that are not adequately controlled through medication or otherwise. This research could dramatically improve the lives of 250,000 Australians and 65 million people worldwide, says Mark Cook, director of the University of Melbourne’s Graeme Clark Institute for Biomedical Engineering and director of neurology at St. Vincent’s Hospital in Melbourne.
“Epilepsy is a neurologic condition that can be incredibly debilitating,” says Cook.
“It prevents some patients from doing simple activities such as getting a driver’s license or swimming. This technology has the potential to improve millions of lives and reduce the physical, emotional, and financial costs of one of the world’s most common, yet intractable chronic disorders,” he explains.
Have you read?
Using the world’s most comprehensive epilepsy patient EEG dataset collected from electrodes inside patients’ skulls, the technology has the ability to adapt to individual patient’s needs, according to David Grayden, head of the university’s biomedical engineering department.
“By collecting data from inside the patient’s skull and combining this with deep learning and AI, we’re able to develop a system that can self-train, based on learning the brain states and signs that preempt seizures unique to an individual,” says Grayden.
“Our algorithm also allows for instantaneous and easy adjustment, giving patients the flexibility to control how sensitive and in advance the warning is,” he says.
While previous epilepsy prediction research has only been possible on high powered computers, Stefan Harrer, IBM Research-Australia’s Brain-Inspired Computing Manager, says that by using IBM’s brain-inspired computing chip, there is the potential to create a wearable, real-time patient warning system.
“By deploying the technology on a computing chip that is the size of a postage stamp and runs on the same power use of a hearing aid, we’re able to simulate how such systems could one day operate in real life,” Harrer says.
“The hope is that one day this research could help inform the development of assistive technologies that could not only warn people with epilepsy of imminent seizures, but constantly adapt to how their brains change over time,” he says.
Cook says developing a reliable means of predicting epileptic seizures for individual patients was an incredibly complex area of research.
“This is in large part due to how epilepsy manifests itself uniquely in each patient, as well as individual long-term changes in brain signals,” he says.
“While we still need to continue to build on this research before we can confidently say that we can identify any seizure before it occurs, these results have proven incredibly promising,” says Cook.