White Paper: Respiratory Rate Estimation in Single-Housed Mice using TrackPaw
Introduction
Respiratory rate is a sensitive physiological readout in preclinical research across respiratory and cardiovascular safety pharmacology, disease models, and general health and welfare assessment. The hard part is not deciding whether breathing matters. It is measuring it without changing the breathing being measured. Established methods usually force a trade-off. Whole-body or head-out plethysmography can provide controlled measurements (Hoymann, 2007), but the animal is removed from its home cage and placed into a short experimental session. Telemetry can extend observation, but it requires surgery and an attached or implanted device; recovery and welfare can be shaped by the housing context itself (Van Loo et al., 2007). In both cases, the method can introduce handling, restraint, novel-environment, or recovery effects that are physiologically meaningful in their own right. TrackPaw takes a different route: the home cage becomes the instrument. It uses the sensorised cage floor to recover respiratory rate during inactivity while the animal remains in its normal cage, with bedding and enrichment, without restraint or a respiratory sensor attached to it. This follows the broader move toward automated, longitudinal home-cage monitoring (Bains et al., 2018; Kahnau et al., 2023). This paper defines the resulting metric, describes the measurement principle, and validates the single-housed implementation against a simultaneous radar reference.
Summary
• What TrackPaw measures. TrackPaw estimates respiratory rate during inactivity from the sensorised floor of the home cage, without restraint or a respiratory sensor attached to the animal. When a mouse is inactive, breathing produces a small periodic force on the cage floor. TrackPaw recovers that rate and reports one value per animal per minute.
• How it was validated. TrackPaw output was compared minute by minute with a simultaneous radar respiration reference. The two systems recorded simultaneously for 413 hours across 10 animals and 4 devices. Both methods reported a rate in 1712 minutes, which form the scored set. Agreement with the radar was strong: a near-zero bias of −1.2 BPM with 95 % limits of agreement of −7.7–5.2 BPM across the observed rate range (reference mean 197 BPM). In plain terms, 99 % of compared minutes agree within 6 % of the radar (Pearson r = 0.99).
• How to read the output. The metric is opportunistic by design. It reports when the animal is still enough for a clean breathing signal and stays silent during movement rather than filling gaps with guesses. A missing minute is no measurement, not a respiratory rate of zero.
• Scope. All claims in this paper are for single-housed mice. Group-housed operation is active work and remains outside the scope of this paper.