Evaluation of Real-Time Preprocessing Methods in AI-Based ECG Signal Analysis

TL;DR

The paper addresses real-time ECG preprocessing on edge devices to enable privacy-preserving, low-latency analytics. It surveys filtering, QRS detection, and signal-quality assessment approaches, highlighting adaptive filters (ANF), MaMeMi, ICA+LDA, and optimized Pan-Tompkins as key candidates. It finds fixed bandpass filters are energy-efficient but struggle with time-varying signals, while adaptive filters and fast SQIs offer real-time viability; the artifact-detection layer can reduce cloud transmission at some cost. Future work will benchmark these techniques on a common reference dataset and standardized hardware such as the NVIDIA Jetson AGX Orin to enable objective comparisons and practical edge deployment.

Abstract

The increasing popularity of portable ECG systems and the growing demand for privacy-compliant, energy-efficient real-time analysis require new approaches to signal processing at the point of data acquisition. In this context, the edge domain is acquiring increasing importance, as it not only reduces latency times, but also enables an increased level of data security. The FACE project aims to develop an innovative machine learning solution for analysing long-term electrocardiograms that synergistically combines the strengths of edge and cloud computing. In this thesis, various pre-processing steps of ECG signals are analysed with regard to their applicability in the project. The selection of suitable methods in the edge area is based in particular on criteria such as energy efficiency, processing capability and real-time capability.