X. Conclusions After reviewing the available closed-loop neurostimulators used to treat refractory epilepsy (RE), we concluded that a closed-loop vagus nerve stimulator (VNS) constructed in such a way that it senses the scalp EEG via a wearable EEG recorder and stimulates the vagus nerve via an implantable VNS (together constituting the HybridVNS system) is feasible to treat RE. We firstly discussed the architecture and system control of the HybridVNS. Next, discrete component implementations of its key building blocks including the multi-channel preamplifier, pulse generator and wireless power link were presented. Moreover, a 4-channel preamplifier amplified the recorded scalp EEG signals and each channel contained a capacitively coupled instrumentation amplifier together with a 2nd-order low-pass filter. We also introduced a switched-resistor array pulse generator for 200 150 100 50 -50 -100 -200 10 Time (s) 10 Figure 18. Recorded rodent LFP in the inter-ictal period, induced epileptic seizure and 15 minutes after stimulation. 1 constant-current stimulation, and a distance-frequency adaptive inductive coupling link to tune the resonant frequency and improve the power efficiency of the wireless power link. Finally, a simulation experiment on the closed-loop operation of the HybridVNS was conducted. In the experiment, the recording feasibility of the HybridVNS was evaluated in a group of patients (where a thresholding epileptic seizure detector was performed to detect seizure onset by the thresholds of both coastline length and slopes in a unit time segment), and the stimulation feasibility of the HybridVNS was tested with an acute rodent epilepsy model. Acknowledgement This work was supported by the National Natural Science Foundation of China (NSFC) under Grant No. 61671221 and Grant No. 61804068. Inter-Ictal Period PTX Induced Epileptic Spikes 15 Minutes After Stimulation Table IV. Performance comparison of the state-of-the-art epileptic seizure detection algorithms in the CLNS. Refs [40] [41] [42] [43] [44] This work Method MinMaxHist Computation Complexity Accuracy (%) Sensitivity (%) High Amplitudes + correlation Low DWT DWT + ANN EWT TESD THIRD QUARTER 2022 86.27 67.9 High High High Low 93.24 95.3 97.9 88.2 80.32 86.9 83.34 97.2 99.6 80.7 Specificity (%) 92.22 89.5 93.53 93.5 99.4 95.3 IEEE CIRCUITS AND SYSTEMS MAGAZINE 37 ERV (µV)