Automatic Assessment of Canine Trainability Using Heart Rate Responses to Positive and Negative Emotional Stimuli

Canine trainability is an important characteristic for working and companion dogs, but assessing it accurately is a challenging task. In this study, we propose a novel approach to evaluate canine trainability by measuring their heart rate responses during behavioral trials. We conducted behavioral tests on a group of 10 dogs of the Siberian Husky breed and recorded their heart rates using a wearable sensor Polar OH1. The behavioral trials consisted of eight tests that induced different emotional reactions. We analyzed the heart rate data using statistical and machine learning techniques to identify patterns and correlations with the dogs' performance in the trials. Our results indicate that heart rate responses during behavioral trials are significantly correlated with the dogs' trainability, as measured by the high correlation (0.77 Pearson's correlation index) and the small error (1.54 MAE on a rating scale from 1 to 10) showed by the trained regression model. We conclude that heart rate variability is a useful feature for distinguishing between highly trainable and less trainable dogs. Our approach provides a reliable and noninvasive method for assessing canine trainability, which could be useful for breeders, trainers, and owners in selecting and training dogs.