A team of Australian researchers from IBM Research and the University of Melbourne developed an "epileptic seizure prediction system," according to study results published in EBioMedicine.
For the study, the researchers obtained the intracranial electroencephalography data of ten patients from a seizure advisory system. They used this data to train a deep learning model to identify brain activity signals that took place before a seizure, which they applied to create a predictive system that alerts clinicians and patients of potential seizure onset.
The researchers deployed the prediction system onto a neuromorphic chip in a wearable device that assess an individual patient's data. The system achieved a mean sensitivity of 69 percent, significantly surpassing a random predictor for all patients, according to the study authors.
The researchers' goal for the system is to enable preventive treatment for patients with epilepsy.
"This study demonstrates that deep learning in combination with neuromorphic hardware can provide the basis for a wearable, real-time, always-on, patient-specific seizure warning system with low power consumption and reliable long-term performance," the study authors concluded.