Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Real-Time, Single-Ear, Wearable ECG Reconstruction, R-Peak Detection, and HR/HRV Monitoring

Carlos Santos, Sebastian Frey, Andrea Cossettini, Luca Benini, Victor Kartsch

TL;DR

This paper tackles unobtrusive, real-time ECG monitoring from a single-ear wearable by reconstructing arm-ECG signals and extracting HR and HRV entirely on-device. It introduces DeepMF-mini, a lightweight adaptation of the DeepMF model, deployed on BioGAP to perform ear-ECG reconstruction and R-peak detection with edge computing constraints. The approach achieves state-of-the-art R-peak detection accuracy (F1 up to 0.95 cross-ear and 0.82 single-ear) and low mean errors in HR (as low as 0.49 bpm) and HRV (as low as 25.82 ms), while delivering extremely low inference energy (36.7 µJ) and long battery life (up to 36 hours). This work demonstrates the feasibility of fully embedded cardiovascular monitoring in everyday head-worn devices, enabling continuous at-home screening for cardiovascular irregularities with minimal hardware complexity.

Abstract

Biosignal monitoring, in particular heart activity through heart rate (HR) and heart rate variability (HRV) tracking, is vital in enabling continuous, non-invasive tracking of physiological and cognitive states. Recent studies have explored compact, head-worn devices for HR and HRV monitoring to improve usability and reduce stigma. However, this approach is challenged by the current reliance on wet electrodes, which limits usability, the weakness of ear-derived signals, making HR/HRV extraction more complex, and the incompatibility of current algorithms for embedded deployment. This work introduces a single-ear wearable system for real-time ECG (Electrocardiogram) parameter estimation, which directly runs on BioGAP, an energy-efficient device for biosignal acquisition and processing. By combining SoA in-ear electrode technology, an optimized DeepMF algorithm, and BioGAP, our proposed subject-independent approach allows for robust extraction of HR/HRV parameters directly on the device with just 36.7 uJ/inference at comparable performance with respect to the current state-of-the-art architecture, achieving 0.49 bpm and 25.82 ms for HR/HRV mean errors, respectively and an estimated battery life of 36h with a total system power consumption of 7.6 mW. Clinical relevance: The ability to reconstruct ECG signals and extract HR and HRV paves the way for continuous, unobtrusive cardiovascular monitoring with head-worn devices. In particular, the integration of cardiovascular measurements in everyday-use devices (such as earbuds) has potential in continuous at-home monitoring to enable early detection of cardiovascular irregularities.

Real-Time, Single-Ear, Wearable ECG Reconstruction, R-Peak Detection, and HR/HRV Monitoring

TL;DR

This paper tackles unobtrusive, real-time ECG monitoring from a single-ear wearable by reconstructing arm-ECG signals and extracting HR and HRV entirely on-device. It introduces DeepMF-mini, a lightweight adaptation of the DeepMF model, deployed on BioGAP to perform ear-ECG reconstruction and R-peak detection with edge computing constraints. The approach achieves state-of-the-art R-peak detection accuracy (F1 up to 0.95 cross-ear and 0.82 single-ear) and low mean errors in HR (as low as 0.49 bpm) and HRV (as low as 25.82 ms), while delivering extremely low inference energy (36.7 µJ) and long battery life (up to 36 hours). This work demonstrates the feasibility of fully embedded cardiovascular monitoring in everyday head-worn devices, enabling continuous at-home screening for cardiovascular irregularities with minimal hardware complexity.

Abstract

Biosignal monitoring, in particular heart activity through heart rate (HR) and heart rate variability (HRV) tracking, is vital in enabling continuous, non-invasive tracking of physiological and cognitive states. Recent studies have explored compact, head-worn devices for HR and HRV monitoring to improve usability and reduce stigma. However, this approach is challenged by the current reliance on wet electrodes, which limits usability, the weakness of ear-derived signals, making HR/HRV extraction more complex, and the incompatibility of current algorithms for embedded deployment. This work introduces a single-ear wearable system for real-time ECG (Electrocardiogram) parameter estimation, which directly runs on BioGAP, an energy-efficient device for biosignal acquisition and processing. By combining SoA in-ear electrode technology, an optimized DeepMF algorithm, and BioGAP, our proposed subject-independent approach allows for robust extraction of HR/HRV parameters directly on the device with just 36.7 uJ/inference at comparable performance with respect to the current state-of-the-art architecture, achieving 0.49 bpm and 25.82 ms for HR/HRV mean errors, respectively and an estimated battery life of 36h with a total system power consumption of 7.6 mW. Clinical relevance: The ability to reconstruct ECG signals and extract HR and HRV paves the way for continuous, unobtrusive cardiovascular monitoring with head-worn devices. In particular, the integration of cardiovascular measurements in everyday-use devices (such as earbuds) has potential in continuous at-home monitoring to enable early detection of cardiovascular irregularities.
Paper Structure (18 sections, 2 equations, 7 figures, 2 tables, 1 algorithm)

This paper contains 18 sections, 2 equations, 7 figures, 2 tables, 1 algorithm.

Table of Contents

  1. Introduction
  2. Related Work
  3. Materials and Methods
  4. Data collection setup
  5. Data Preparation and Pre-processing
  6. Training: autoencoder
  7. Training: R-peak detection classifier
  8. Inference: real-time rolling-based R-peak postprocessing
  9. Inference: HR and HRV calculation
  10. Accuracy evaluation
  11. Edge Deployment
  12. Results
  13. Signal Characterization
  14. Inference examples
  15. Algorithm Evaluation
  16. ...and 3 more sections

Figures (7)

  • Figure 1: Overview of the measurement setup (left) and algorithmic flow: (1) autoencoder training, (2) R-peak detection classifier, (3) rolling-based postprocessing for improved R-peak detection, (4) Real-time HR/HRV calculation.
  • Figure 2: Rolling window R-peak refinement process.
  • Figure 3: Biopotential measurements aligned to the ground truth R-peaks and averaged over all recordings for the arm ECG (top), single-ear biopotential (middle), and cross-ear biopotential (bottom).
  • Figure 4: Inference example illustrating the input signals for single-ear and cross-ear configurations (top), the corresponding ECG reconstructions alongside the ground truth ECG (middle), and the predicted R-peak locations compared to the ground truth R-peaks (bottom).
  • Figure 5: HR and HRV prediction.
  • ...and 2 more figures