Key result
Wearable IMU-based respiration monitor accurately estimates breathing rates with under ~4% error.
Why the study?
Existing wearable systems for respiration monitoring during daily activities often prove complex, non-wearable, and expensive, creating a need for a more effective solution.
Does a wearable respiration monitoring system using an IMU sensor and ResPara-Net DCNN accurately estimate respiration parameters compared to a gold standard device?
Does a wearable respiration monitoring system using an IMU sensor and ResPara-Net DCNN accurately estimate respiration parameters compared to a gold standard device?
A novel wearable IMU-based system with a deep learning model accurately estimates respiration parameters across different breathing rates, offering a potential tool for continuous respiratory monitoring.
May support continuous ambulatory respiration monitoring; leaves open clinical validation in larger cohorts.
Respiration plays an important role in detecting and diagnosing cardiovascular diseases such as asthma, chronic obstructive pulmonary disease (COPD), and sleep apnea (SA). Despite the increasing interest in wearable devices for comfortable respiration recognition during daily activities, existing systems often prove complex, non-wearable, and expensive, emphasizing the need for a more effective solution. The presented research introduces a user-friendly, cost-effective, and wearable respiration monitoring system (WRMS) coupled with a novel deep convolutional neural network (ResPara-Net DCNN) for real-time respiration monitoring during daily activities. The developed WRMS continuously estimates respiration parameters using the DCNN and an inertial measurement unit (IMU) signal. Experiments encompassing three different respiration rates (RRs) were conducted to assess the system’s performance, with results compared to a laboratory-level gold standard device. The average root-mean-square-error (RMSE) values for normal, fast, and slow breathing rates were found to be 0.14, 0.12, and 0.13, respectively. Similarly, the average correlation coefficient (CC) values for normal, fast, and slow breathing rates were 64.47%, 67.48%, and 71.53%, indicating a robust level of accuracy. The average predicted breathing rates for normal, fast, and slow were 16, 28, and 12 breaths per minute (BPM), respectively. Furthermore, the average normalized mean absolute error (NMAE) between predicted and actual respiration signals for all breathing speeds was less than 4% across all subjects. The proposed DCNN-based WRMS system provides valuable insights into breathing dynamics during daily activities. Moreover, it also introduces a fresh perspective on respiration monitoring. The physical and physiological significance lies in its potential to offer a user-friendly, cost-effective, and continuous monitoring solution, thereby contributing to the advancement of cardiovascular health diagnostics and daily activity-based respiratory assessments.
No takes yet. Share an insight, caveat, or question.
Bhongade et al. (2024) studied Respiration monitoring. Wearable respiration monitoring system (WRMS) with ResPara-Net DCNN vs. Laboratory-level gold standard device was evaluated on Respiration parameter estimation accuracy (RMSE, CC, NMAE). A wearable respiration monitoring system using a deep convolutional neural network and IMU signal accurately estimated breathing rates with an average normalized mean absolute error of less than 4%.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: