Detecting and measuring respiratory events in horses during exercise with a microphone deep learning vs standard signal processing

Jul 16, 2026

Upload by Himbo

Monitoring respiration parameters such as respiratory rate supports understanding how training affects equine health, performance, and welfare. The study compares deep learning approaches with an adapted signal-processing method to automatically detect cyclic respiratory events and derive dynamic respiratory rate from microphone recordings collected during high-intensity exercise in Standardbred trotters. Results show robust detection of exhalation sounds in noisy signals and promising performance on unlabeled data at lower intensity.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages44

Reading Time1 h 51 min

Background

Respiration parameters and equine respiratory physiology

Upper respiratory tract disorders and current diagnostic approaches

Rationale for automatic detection from microphone signals

Detecting and measuring respiratory events in horses during exercise with a microphone: deep learning vs. standard signal processing Jeanne I. M. Parmentiera, b,c, Rhana M. Aartsa, Elin Hernlundd, Marie Rhodind, Berend Jan van der Zwaagb,c a Department of Clinical Sciences, Faculty of Veterinary Medicine, Utrecht University, Yalelaan 112-114, 3584 CM, Utrecht, The Netherlands b Pervasive Systems Research Group, Edge Research Centre, EEMCS, University of Twente, Enschede, The Netherlands c Inertia Technology B.V., Hengelosestraat 583, 7521 AG, Enschede, The Netherlands d Department of Animal Biosciences, Swedish University of Agricultural Sciences, Uppsala, Sweden Corresponding Author: Jeanne I.M. Parmentier [j.i.m.parmentier@uu.nl](j.i.m.parmentier@uu.nl) [Accepted at](Accepted at: BMC Veterinary Research)[: BMC Veterinary Research](Accepted at: BMC Veterinary Research) doi: [https://doi.org/10.1186/s12917-026-05614-5](https://doi.org/10.1186/s12917-026-05614-5) Abstract Background: Monitoring respiration parameters such as respiratory rate could be beneficial to understand the impact of training on equine health and performance and ultimately improve equine welfare. In this work, we compare deep learning-based methods to an adapted signal processing method to automatically detect cyclic respiratory events and extract the dynamic respiratory rate from microphone recordings during high-intensity exercise in Standardbred trotters. Results: Our deep learning models are able to detect exhalation sounds (median F1 score of 0.94) in noisy microphone signals and show promising results on unlabelled signals at lower exercise intensity, where the exhalation sounds are less recognisable. Temporal convolutional networks were better at detecting exhalation events and estimating dynamic respiratory rates (median F1: 0.94, Mean Absolute Error (MAE) ± Confidence Intervals (CI): 1.44±1 .04 bpm, Limits Of Agreements (LOA): 0.63±7 .06 bpm) than long short-term memory networks (median F1: 0.90, MAE±CI: 3. 11±1 .58 bpm) and signal processing methods (MAE±CI: 2.36±1 .11 bpm) . Conclusion: This work is the first to automatically detect equine respiratory sounds and automatically compute dynamic respiratory rates in exercising horses. In the future, our models will be validated on lower exercise intensity sounds and different microphone placements will be evaluated in order to find the best combination for regular monitoring. Keywords Artificial intelligence, automated detection, horse, respiration, respiratory rate, microphone, exercise physiology Abbreviations bpm c1 c2 CI DL Fs GNSS IQR LOA LOO LSTM MAE breath per minute channel 1 channel 2 confidence intervals deep learning sampling frequency global navigation satellite system interquartile range limits of agreements leave-one-out long short-term memory mean absolute error MOD mean of differences MvAvg moving average MvVar moving variance RR respiratory rate s.d. standard deviation SP signal processing TCN temporal convolutional network URT upper respiratory tract 1. Background Respiration parameters, including breathing patterns, respiratory rate (RR), inhalationexhalation ratio, respiratory sounds and locomotor-respiratory coupling represent critical physiological aspects in exercising horses. Quantitative and qualitative analyses of these parameters during exercise may contribute to a better understanding of performance mechanisms and respiratory function in horses (1) . Horses are obligate nose breathers meaning that they cannot switch to oronasal breathing like humans. The equine upper respiratory tract (URT) includes the nostrils, nasal passages, pharynx, and larynx and is essential for moving air in and out during respiration, with inspiratory and expiratory airway flows reaching 65-75 L/s and 60-80 L/s respectively during exercise (2) . In comparison, human literature reported maximum nasal airflo

FAQs

What is the document’s main objective?

To automatically detect cyclic respiratory events and compute dynamic respiratory rate in exercising horses using microphone recordings, comparing deep learning with an adapted signal processing approach.

How do the deep learning models perform for detecting respiratory sounds?

Deep learning models detect exhalation sounds with a median F1 score of 0.94 in noisy microphone signals and show promising results on unlabeled signals at lower exercise intensity.

Which model type and baseline methods are reported as outperforming others?

Temporal convolutional networks detect exhalation events and estimate dynamic respiratory rates better than long short-term memory networks and traditional signal processing methods, based on reported median F1 and MAE (with confidence intervals).

Jul 16, 2026

Upload by Himbo