How low can we go? Influence of sample rate on equine pelvic displacement calculated from inertial sensor data.
Authors: Pfau Thilo, Reilly Patrick
Journal: Equine veterinary journal
Summary
# Editorial Summary: Sample Rate Requirements for Equine Pelvic Motion Analysis Inertial measurement units (IMUs) are increasingly used to quantify pelvic movement asymmetry in lame horses, but many affordable or portable devices operate at lower sample rates than laboratory-grade equipment. Pfau and Reilly investigated how reducing sample rates from a validated 100 Hz reference standard affected the accuracy of pelvic displacement measurements, using the Nyquist theorem as their theoretical framework and systematically down-sampling data to rates between 8 and 50 Hz. At 50 Hz, root mean square errors remained below 2 mm with precision ±1 mm—clinically meaningful thresholds for detecting lameness—whilst errors escalated sharply at 20 Hz and below, reaching up to 20 mm RMS error. Since many low-cost sensors, smartphones and wireless devices can reliably maintain 50 Hz sampling, this work validates their use for clinical lameness assessment despite being substantially cheaper and more portable than traditional force plate or high-frequency IMU systems. Practitioners selecting IMU-based gait analysis tools should prioritise devices operating at minimum 50 Hz to ensure measurements of pelvic motion remain sufficiently precise for diagnostic confidence.
Read the full abstract on PubMed
Practical Takeaways
- •You can reliably use affordable wireless sensors, smartphones, or low-cost data loggers running at 50 Hz for lameness evaluations—measurements will be accurate to within 2 mm
- •Avoid sensors operating below 20 Hz as errors become unacceptably large (≥20 mm); verify your device specifications before purchase
- •When assessing pelvic asymmetry in lame horses, ensure your sensor equipment meets the 50 Hz minimum threshold for clinically useful data
Key Findings
- •RMS errors in pelvic displacement measurement are <2 mm at 50 Hz sample rate compared to reference 100 Hz data
- •Errors increase to 20 mm or more at 20 Hz and below sample rates
- •Accuracy for detecting maxima and minima peaks is generally below 1 mm across tested sample rates
- •Low-cost sensors capable of sustaining 50 Hz sample rates can validly quantify pelvic movement parameters for lameness assessment