The observation problem
Separation distress typically peaks within the first thirty minutes after departure and then partially subsides. That window is, by definition, unobservable by the owner.
Owners therefore reason backwards from evidence: a chewed door frame, a complaint from a neighbour, a puddle in the hallway. All of these are late-stage outcomes, and none of them tell you about severity or trajectory.
What anxiety actually looks like in data
Physiologically, anxiety is straightforward to measure. Heart rate rises, heart rate variability collapses, and neither recovers in the way they would after exercise.
Behaviourally, distress produces a recognisable signature: pacing along consistent routes, repeated door checking, displacement behaviours such as yawning and lip licking, and vocalisation with a distinct acoustic profile from an alert bark.
Building a severity score
The clinically useful output is not a binary anxious or not anxious label. It is a graded score that combines peak arousal, time to settle, and total distress duration.
That number is what makes treatment measurable. A dog whose settle time falls from forty minutes to twelve is improving, even if the owner still comes home to occasional evidence of distress.
- Peak physiological arousal in the first thirty minutes.
- Time to settle, meaning return to baseline heart rate.
- Total minutes of classified distress behaviour.
- Recovery consistency across repeated departures.
Privacy is not optional here
Detecting distress vocalisation requires listening, which raises an obvious concern about household audio. Our position is that audio should never leave the device.
Classification runs locally on the sensor. The audio stream is analysed and discarded in memory, and only an event label such as distress vocalisation, four minutes, leaves your home. Human speech is never stored or transmitted.
Turning measurement into improvement
Once severity is measurable, graduated departure protocols become straightforward to run properly. You extend absence duration only when the previous duration produces a settled response, which is the step most owners skip.
Where scores remain high after eight to twelve weeks of consistent work, that is the point to involve a veterinary behaviourist. Arriving with a measured history makes that appointment dramatically more productive.