Machine learning from wristband sensor data for wearable, noninvasive seizure forecasting.

Collectively, these results show that statistically significant seizure risk assessments are feasible from easy-to-use, noninvasive wearable devices without the need of patient-specific training or parameter optimization.