A 2.67GΩ 454nVrms 14.9μW Dry-Electrode Enabled ECG-on-Chip with Arrhythmia Detection

Abstract

Atrial fibrillation (AF) is a common type of cardiac arrhythmia and a silent killer which will affect 12.1 million people in USA in 2030 according to CDC. It is possible to identify AF at early stage. An ECGon-chip for wearable AF detector needs to possess following features: (1) $\mathrm{G} \Omega$ input impedance to deal with dry-electrodes that have contact impedance at $M \Omega$ level; (2) Good noise performance to capture small ECG details; (3) Embedded arrhythmia detection to detect sporadic AF events and minimize wireless transmission. The prior art of $\mathrm{G} \Omega$ analog frontends (AFEs) [1], [2] can handle $M \Omega$ electrode and mismatch, but they require manual tuning for AFE stability and better noise performance. On-chip machine learning arrhythmia classifications [3] is able to classify AAMI recommended 5 types of arrhythmia at the cost of power. The analog-based heartrate detection [4] reduces power but is limited to detect heart rate only. ECG-on-Chip solutions [5], [6] still miss one or more key features listed above.