Ventricular Ectopic Beats Detection Using MEMS-Resonator-Based Reservoir Computer

In this study, we designed and evaluated microelectromechanical systems (MEMS) resonator-based reservoir computer for ventricular ectopic beats (VEB) detection. The proposed system exploits the dynamic properties of an MEMS resonator to enhance classification performance. Compared with the learning outcomes obtained without the reservoir system, our system demonstrated substantial improvements: the sensitivity for detecting VEB increased from 71.3% to 93.3% and the overall classification accuracy rose from 83.6% to 92.3%. In addition, t-distributed stochastic neighbor embedding visualized the enhancement in data separation, reflecting the system's enhanced recognition capability. These findings suggest that MEMS-resonator-based reservoir computers hold great potential for enhancing arrhythmia detection and offer a promising solution for wearable diagnostic systems in healthcare.