fairfolio
Saved Projects

Automatic Annotation of Social Behavior of Voles

AJAS · 2022 Behavioral Science

A L

Thumbnail supplied by the source for Automatic Annotation of Social Behavior of Voles

Overview

The study of social attachment behaviors and the factors behind them is integral for developing a deeper understanding of many neuropsychiatric conditions with social deficits, such as autism and schizophrenia. The main animal model used in these studies is Microtus ochrogaster, commonly known as the prairie vole, due to its tendency to form lifelong socially monogamous pair bonds. However, studies of naturalistic social behaviors require time- and labor-intensive manual annotation, which is also limited by imperfect inter-rater reliability. Therefore, developing a system for automatic annotation of these behaviors can make these studies more accessible. This study aimed to develop a set of features from computationally defined vole contours, and also build a classification model for huddling, an attachment behavior specific to prairie voles. The average F1 scores for the presence and absence of the behavior were 97% and 98%, respectively, with an accuracy of 98%. The features that were deemed to be most important for the huddling classifier were the area, minor axis length, and eccentricity of the center vole. Future directions for this study are focused on optimizing the current model and developing classifiers for other vole behaviors. Developing a classification model for huddling behaviors in voles is an important step in increasing the accessibility and accuracy of social behavior studies.

Video

From the student

Two summers ago, I participated in a summer research program known as the Summer Research Institute. I was matched with a mentor based on my research interests. During that summer and the year after, I worked on this research project with my mentors.

From the student

The study of social attachment behaviors and the factors behind them is integral for developing a deeper understanding of many neuropsychiatric conditions with social deficits, such as autism and schizophrenia. The main animal model used in these studies is Microtus ochrogaster, commonly known as the prairie vole, due to its tendency to form lifelong socially monogamous pair bonds. However, studies of naturalistic social behaviors require time- and labor-intensive manual annotation, which is also limited by imperfect inter-rater reliability. Therefore, developing a system for automatic annotation of these behaviors can make these studies more accessible. This study aimed to develop a set of features from computationally defined vole contours, and also build a classification model for huddling, an attachment behavior specific to prairie voles. The average F1 scores for the presence and absence of the behavior were to 97% and 98%, respectively, with an accuracy of 98%. The features that were deemed to be most important for the huddling classifier were the area, minor axis length and eccentricity of the center vole. Future directions for this study are focused on optimizing the current model and developing classifiers for other vole behaviors. Developing a classification model for huddling behaviors in voles is an important step in increasing the accessibility and accuracy of social behavior studies.

Images (16)

Project image Project image Project image Project image Project image Project image Project image Project image Project image Project image Project image Project image Project image Project image Project image Project image

Awards (1)

  • AJAS Fellows Badge

Competition history

  • AJAS 2022 Behavioral Science

Related projects

Closest projects by meaning, across every fair and year in the corpus.

Browse more like this

Source: ProjectBoard / American Junior Academy of Science

Save projects to your library

Sign in with Google to keep track of projects you find interesting, organized into folders. An account also raises your daily allowance for “Has this been done?”, and lets you create a key for the MCP server with a much higher limit than anonymous use. Browsing stays public.

Continue with Google