Respiratory Monitoring
Audio and video nearables for monitoring respiratory rate in sleeping dogs
This study validates smartphone-based audio and video 'nearable' methods for non-invasive respiratory rate (RR) monitoring in sleeping dogs. Four methods (two audio, two video) were tested on 27 sleeping dogs, comparing estimated RR to manual breath counting. All methods showed good agreement, with video-based methods (especially lateral view) achieving the lowest errors. These findings support the feasibility of accurate, non-invasive RR monitoring using common smartphones for continuous health assessment in pets.
Executive Impact for Your Enterprise
Leveraging existing smartphone technology for non-invasive health monitoring in pets offers significant benefits for veterinary and pet-care businesses. This AI-powered approach reduces the need for costly, invasive, and stress-inducing clinical visits, making continuous health tracking accessible and affordable for pet owners. Early detection of respiratory and cardiac conditions through passive monitoring can lead to improved animal welfare, proactive veterinary care, and increased client satisfaction, ultimately enhancing the value proposition for pet insurance, wellness programs, and veterinary services.
Lowest RR Error (Video-Lateral)
Highest Agreement (Video-TopDown/Lateral)
Tightest Limits of Agreement
Study Cohort Size
Deep Analysis & Enterprise Applications
Select a topic to dive deeper, then explore the specific findings from the research, rebuilt as interactive, enterprise-focused modules.
Respiratory Monitoring
Video Analysis
Audio Analysis
Data Collection
Non-Invasive
AI-powered smartphone methods offer a non-invasive alternative to traditional vital sign monitors, reducing animal stress and the need for sedation in clinical settings.
| Method Type | Advantages | Limitations for Home Use |
|---|---|---|
| Wearable Devices (e.g., collars) |
|
|
| UWB Radars |
|
|
| Smartphone Nearables (Audio/Video) |
|
|
Superior Performance of Video-Based Methods
The study found that video-based methods consistently outperformed audio-based methods, achieving the lowest RMSE (1.1 bpm) and highest R² (0.99). Method D (lateral view video) showed a bias of 0.00 bpm and the tightest limits of agreement ([-2.17, +2.17] bpm), indicating superior accuracy and reliability for detecting subtle thoracic movements during sleep. This suggests that visual monitoring of movement is more robust than audio detection of breathing sounds, especially in varying home environments.
Video analysis, particularly from a lateral perspective, provides the most accurate and robust respiratory rate monitoring using smartphones.
Enterprise Process Flow
Read Video Track
→
Manually Select ROI (e.g., Thorax/Abdomen)
→
Track & Average Coordinates (25 subpoints)
→
Calculate Net Displacement Z(t)
→
Signal Detrending
→
High-pass Filter (0.05 Hz)
→
Smoothing
→
Automated Peak Counting
→
Respiratory Rate (bpm)
Utility and Challenges of Audio-Based Monitoring
Audio-based methods, while less accurate than video, still demonstrated good agreement with manual counting (Method A R²=0.97, Method B R²=0.90). Method A, using an external earphone microphone, performed better than Method B, which used the smartphone's built-in microphone. This suggests that proximity to the muzzle and a higher signal-to-noise ratio (SNR) improve accuracy. However, challenges include differentiating inspiratory/expiratory sounds from environmental noise and detecting quiet breathing patterns. Custom parameter adjustments were often required for accurate peak detection, highlighting the need for further robustness.
Audio methods can be effective, especially with external microphones, but require careful noise management and signal processing.
Enterprise Process Flow
Read Audio Track
→
Normalize & Center Waveform
→
Downsampling (10 Hz)
→
RMSE-based Envelope Extraction
→
Smoothing
→
Band-pass Filter (0.075-1 Hz)
→
Automated Peak Counting
→
Respiratory Rate (bpm)
Home Environment
All data was collected in the pet's natural home environment, during sleep, without sedation, minimizing stress and capturing true resting respiration rates.
| Method | Setup | Key Takeaways |
|---|---|---|
| Method A (Audio) | External earphone microphone near muzzle |
|
| Method B (Audio) | Smartphone built-in microphone near muzzle |
|
| Method C (Video) | Smartphone camera, top-down view (thoracic/abdominal region) |
|
| Method D (Video) | Smartphone camera, lateral view (trunk area) |
|
Calculate Your Potential ROI
See how AI-powered respiratory monitoring can translate into tangible savings and efficiency gains for your organization.
Annual Savings
$0
Annual Hours Reclaimed
0
Your AI Implementation Roadmap
Our AI implementation roadmap focuses on iterative development, starting with foundational data processing and moving towards fully automated, user-friendly solutions. Each phase builds upon the last, ensuring a robust, scalable, and clinically relevant system for pet health monitoring.
Phase 1: ROI Automation & Adaptive Filtering
Develop AI models for automatic region-of-interest (ROI) selection in video recordings and implement adaptive filter/envelope windows for audio. This will enhance reproducibility and user-friendliness, reducing manual parameter adjustments.
Phase 2: Mobile App Development (MVP)
Create a Minimal Viable Product (MVP) mobile application for synchronized audio/video acquisition, on-device processing, RR computation with quality flags, and anonymized data upload. Integrate guided acquisitions and posture detection hints.
Phase 3: Expanded Cohort & Clinical Validation
Conduct large-scale studies with diverse dog cohorts (stratified by age, breed, size, disease status) and validate against clinical standards (e.g., impedance pneumography) to establish generalizability and clinical relevance.
Phase 4: Feature Expansion & Privacy-by-Design
Implement advanced features like audio/video fusion, outlier rejection, veterinarian data portal, and robust privacy controls (local processing, owner consent, encrypted storage) to ensure a secure and comprehensive solution.
Phase 5: Production-Grade Release & Regulatory Compliance
Finalize the production-grade release, covering usability testing, quality assurance, and compliance with relevant veterinary and data privacy regulations for widespread adoption.
Ready to Transform Pet Healthcare with AI?
Unlock the power of non-invasive, continuous respiratory monitoring for your veterinary practice or pet-tech business. Our AI solutions can enhance patient care, improve owner engagement, and drive new revenue streams. Let's discuss how we can tailor this technology to your specific needs.
Ready to Get Started?
Book Your Free Consultation.
Let's Discuss Your AI Strategy!
Lets Discuss Your Needs
Select a Date
Sun
Mon
Tue
Wed
Thu
Fri
Sat