Year: 2025
Type: article
Abstract: Continuous monitoring of respiratory rate (RR) in dogs is critical for early detection of respiratory and cardiac conditions. However, current solutions are often invasive, or impractical for at-home use. This study aims to evaluate four smartphone-based nearable methods (two audio-based and two video-based) for non-invasive RR monitoring in sleeping dogs. 27 dogs were recorded during natural sleep using four different setups: (A) audio with earphone microphone, (B) audio with smartphone microphone, (C) video from a top-down perspective, and (D) video from a lateral view. Signals were processed to estimate RR. Manual breath counting served as the reference. All methods demonstrated good agreement with the reference. RMSE ranged from 1.1 to 2.2 bpm, MAE from 0.7 to 1.5 bpm. Video-based methods performed better, with method D achieving the lowest errors (RMSE = 1.1 bpm, MAE = 0.7 bpm, bias = 0.00 bpm) and tightest limits of agreement in the Bland-Altman analysis ([-2.17, + 2.17] bpm). Pearson's R² was above 0.97 for all methods except B. Friedman tests revealed no significant differences among methods. Based on these results, smartphone-based nearable solutions are accurate for non-invasive respiratory monitoring in sleeping dogs. Video-based methods are more robust, while audio methods may be suitable for dogs with audible breathing patterns.
Source: Scientific Reports
Authors: Alessandra Angelucci, Aurora Badino, Valentina Bussolati, Sara Caccia, Cecilia Campanini, Francesco Birettoni, Andrea Aliverti
Institutions: Politecnico di Milano, University of Perugia
Language: English
Subfield: Biomedical Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: gold
Estimated APC paid: $2,690
Funder: European Commission
OpenAlex ID: W4416462126