Decoding Social Interactions in an Autism Mouse Model through Machine Learning

Abstract

Social interaction is essential for human development and well-being, influencing cognition, decision-making, and overall quality of life. Autism spectrum disorder (ASD) disrupts social cognition, posing challenges in interpreting social cues, which can have an effect on interpersonal interactions and relationships. Autism mouse models, such as the L7-Tsc1 strain with a cerebellum-specific mutation, offer insights into ASD-like behaviors and allow researchers to study the nuances of social behavior. To examine the influence of context on behavior, pairs of mice were recorded in an open field arena and their behavior analyzed. Machine learning methods were implemented to improve the efficiency and accuracy of animal pose tracking, which allowed for isolation of context-dependent effects on behavior. Combining positional data of animals within the arena with an unsupervised analysis of their behaviors demonstrated the complex interplay between behavior and temporal, spatial, and genetic factors. Altering the size of the arena not only affected interactions between wildtype mouse pairs, but also reversed behavioral patterns exhibited by wildtype mice when interacting with mutants. The results highlight the intricate dynamics of social interaction, informing not only ASD research but also broader neurodevelopmental and psychiatric disorders affecting social function.