Fast Sample Entropy Atrial Fibrillation Analysis Towards Wearable Device

Atrial Fibrillation (AF) is a common and often debilitating heart rhythm disorder characterized by an irregular, rapid, and sometimes chaotic heart rate. This paper proposes to use fast entropy estimation algorithm to analyze AF on wearable processors for ECG R wave interval. The suggested fast entropy method requires less computation time to detect AF. Information entropy, typical as Sample Entropy (SampEn), is a popularly used complexity measurement metric for AF detection. However, the quadratic time complexity of conventional SampEn restricts its analysis ability. The fast entropy estimation map original signals by Merge Sort (MS) and use a lightweight (LW) similarity-checking method to avoid unnecessary operations on entropy analysis. The proposed algorithm is verified on the long-term recordings from MIT-BIH AF database. This fast entropy measurement, MS-LW, has the same complexity measurement result as SampEn but is three times faster, which is convenient for long-term data. Altogether, this lightweight SampEn is suitable for wearable applications.