Synapse
⌘+K
Synapse
PulseExploreClubsResearchersJournals
Instagram
HomeClubsExplore
September 5, 2026Technology and Health Care

Hybrid CWT framework matches CNN-LSTM heart rate accuracy while reducing latency to ~86 ms.

View Full Paper
Ask AI
Bookmark
Share

Why the study?

Ballistocardiography-based heart rate monitoring experiences accuracy degradation caused by motion artifacts, limiting practical deployment.

Does a hybrid system integrating adaptive filtering and enhanced CWT improve heart rate estimation accuracy and reduce processing latency from bed-based ballistocardiography signals with motion artifacts?

Population

6000 min of bed-based ballistocardiography signal data

Comparison

Hybrid adaptive filtering and CWT framework vs CNN-LSTM baseline and conventional methods

Key result

A hybrid framework integrating adaptive filtering and enhanced CWT achieved an MAE of 2.94 BPM for heart rate monitoring, comparable to a CNN-LSTM baseline, while reducing latency to 86.4 ms.

Authors

ZWZeya WangZFZhumu FuHWHua Wang

Discussion

Loading...

Member takes

Overview

May enable real-time BCG heart rate monitoring; leaves open prospective clinical validation.

Key Points

  • Enhance the accuracy and real-time computational performance of heart rate estimation from bed-based ballistocardiography signals corrupted by motion artifacts.
  • Developed a hybrid framework integrating adaptive filtering and enhanced continuous wavelet transform (CWT) with magnitude-frequency nullification for spectral reconstruction.
  • Evaluated performance on 6000 minutes of signal data and assessed processing latency on an embedded ARM platform.
  • The framework achieved a mean absolute error (MAE) of 2.94 BPM (comparable to a CNN-LSTM baseline of 2.85 BPM) with Bland-Altman limits of agreement within [-4.77, 5.23] BPM.
  • Processing latency decreased from 450.2 ms to 86.4 ms, reducing MAE by 53.8% and improving monitoring stability by 35.6% compared to conventional methods.

Structured PICO

Does a hybrid system integrating adaptive filtering and enhanced CWT improve heart rate estimation accuracy and reduce processing latency from bed-based ballistocardiography signals with motion artifacts?

P
Population
6000 minutes of bed-based ballistocardiography (BCG) signal data with motion artifacts
I
Intervention
Hybrid system integrating adaptive filtering and enhanced continuous wavelet transform (CWT)
C
Comparator
CNN-LSTM baseline and conventional methods
O
Outcome
Mean Absolute Error (MAE) of heart rate estimation and processing latencysurrogate

Main Result

Absolute Event Rate: 2.94% vs 2.85%

A novel hybrid framework integrating adaptive filtering and enhanced CWT provides a computationally efficient and accurate solution for non-contact heart rate monitoring in motion-prone clinical environments.

Cite This Study

Wang et al. (2026) studied Heart rate monitoring with motion artifacts. Hybrid system integrating adaptive filtering and enhanced continuous wavelet transform (CWT) vs. CNN-LSTM baseline was evaluated on Mean Absolute Error (MAE) of heart rate estimation. A hybrid framework integrating adaptive filtering and enhanced CWT achieved an MAE of 2.94 BPM for heart rate monitoring, comparable to a CNN-LSTM baseline, while reducing latency to 86.4 ms.

synapsesocial.com/papers/6a9bd4046b95aff0620eb519https://doi.org/10.1177/09287329261481871
View Full Paper
Ask AI
Bookmark
Share