A Hardware-Aware MLP Inference Framework for Bradyarrhythmia Classification

Abstract

This paper presents a hardware-accelerated model for bradyarrhythmia classification, which addresses the shortcomings of traditional techniques by the integration of machine learning with hardware to enable power-efficient and accurate classification of bradyarrhythmia, using Butterworth bandpass filter. Common software algorithms for ECG classification find it difficult to maintain performance and efficiency, especially in wearable or portable medical devices. The architecture which is designed helps in improving the accuracy of classification of bradyarrhythmia with MLP and bridging the gap between accuracy and hardware implementation which helps in the development of biomedical signal processing and machine learning.