A 1.83µJ/classification nonlinear support-vector-machine-based patient-specific seizure classification SoC

To mitigate seizure-affected patients, SoCs [1-3] have been developed 1) to detect electrical onset of seizure seconds before the clinical onset, and 2) to combine the SoC with neurostimulation. In particular, having detection delay of <;2s (for real-time suppression) while maintaining high detection rate is challenging [4]. However, [2] had a long latency (13.5s) and [3] suffered from a low detection rate (84.4%) with a high false alarm (max. 14.7%) due to an intermittent limit of the Linear Support Vector Machine (LSVM). In this paper, we present a Non-Linear SVM (NLSVM)-based seizure detection SoC which ensures a >95% detection accuracy, <;1% false alarm and <;2s latency.