Long-term scratching analysis of mice using machine learning

Pruritus is a bothersome symptom that can occur at any time and is caused by a variety of physical and mental disorders. There is an urgent need to elucidate the detailed pathophysiology using animal models. Using image analysis and machine learning, we analyzed the scratching behavior of BALB/c mouse under both light and dark conditions and objectively quantified persistency, strength, and amount of pruritus. A 24-h analysis in naive mice revealed that mice scratched themselves more frequently and longer in the light period than in the dark period, and that scratching behavior in the light period was hard to extinguish. Administration of dinitrofluorobenzene (DNFB) induced persistent and biphasic scratching behavior while the vehicle only induced transient one. Of interest, DNFB-induced scratching behavior occurred even when the mice were resting. These features are revealed for the first time, highlighting the usefulness of automated long-term scratching behavior analysis.