Behavioral analysis of mouse social behavior using artificial intelligence

Abstract

For the last decade, social interaction has been continuously improved for the measurement of sociability and social preference of mice. Although it is one of the promising behavioral criteria for analyzing interaction, the parameters used merely included the factors such as chamber time and sniffing time. Due to their unique behavioral patterns which comprise of building blocks of specific poses, artificial intelligence and computer vision techniques were integrated to find out forthcoming parameters. Depending on quantification methods, their behavioral motifs show a wide spectrum from actions and movemes with unique sequences. In this thesis, we study various behavioral paradigm of mouse social interaction, including postures, motifs and sequences according to their genotypical characteristics, based on t-test, two-way ANOVA test and computational models that represent each genotypical behavioral patterns.