Machine learning and confirmatory factor analysis show that buprenorphine alters motor and anxiety-like behaviors in male, female, and obese C57BL/6J mice.
Journal:
Journal of neurophysiology
PMID:
39852951
Abstract
Buprenorphine is an opioid approved for medication-assisted treatment of opioid use disorder. Used off-label, buprenorphine has been reported to contribute to the clinical management of anxiety. Although human anxiety is a highly prevalent disorder, anxiety is a latent construct that cannot be directly measured. The present study combined machine learning techniques and artificial intelligence with confirmatory factor analysis to evaluate the hypothesis that buprenorphine alters motor and anxiety-like behavior in C57BL/6J (B6) mice ( = 30) as a function of dose, sex, and body mass. After administration of saline (control) or buprenorphine, mice were placed on an elevated zero maze (EZM) for 5 min. Digital video of mouse behavior was uploaded to the cloud, and mouse position on the maze was tracked and analyzed with supervised machine learning and artificial intelligence. ANOVA and post hoc test showed that buprenorphine significantly altered five motor behaviors. Confirmatory factor analysis revealed that the latent construct of anxiety-like behavior accounted for a statistically significant amount of variance in all five motor behaviors. Machine learning and pose estimation using a convolutional neural network accurately detected and objectively scored buprenorphine-induced changes in locomotor behaviors of mice on an elevated zero maze (EZM). Confirmatory factor analysis supports the interpretation that the anxiety-like construct accounted for the buprenorphine-induced changes in motor behavior. The results have noteworthy implications for the relationship between Darwin's story model of mammalian emotions and computational models of anxiety-like behavior in mice.