Sub-threshold Computational Circuits for High-order Data-driven Analysis of Physiological Signals

Thanks to emerging sensing and acquisition technologies, wearable and implantable systems are able to access physiologically‐indicative signals (e.g., ECG, EEG, ECoG, LFPs, etc.) at low power levels, making possible devices of potentially high scientific and clinical value. The signals, however, represent the superposition of many simultaneous physiological processes. Detecting targeted states therefore requires high‐order analysis. Sub‐threshold computational devices have been demonstrated for basic but low‐energy signal processing. Higher‐order signal models incur substantially higher computational power, challenging the viability of advanced analysis in wearable and implantable devices