Wrist-Worn Devices For People With Epilepsy May Provide 30-Minute Warning Before Seizure

Wrist-worn devices for people with epilepsy may provide an average seizure warning of 33-minutes through machine-learning algorithms (a long short-term memory (LSTM) recurrent neural network (RNN) algorithm). About 83% of those analyzed had seizure forecasts significantly more accurate than a random predictor. Seizure alerts in these individuals provided ample warning time to administer fast-acting medication, or to increase neuromodulation therapy in an attempt to prevent seizures.

Seizure forecasts are a result of specific patterns identified by a machine-learning algorithm for individuals with epilepsy who wear special wrist-worn recording device, who also have drug . . .