Automated and data-driven discovery of behavioral signatures in preclinical models of developmental disorders
Presenter(s)
Henry Salisbury
Files
Description
The primary mode of identifying developmental disorders in children involves behavioral analysis. However, behavioral datasets are mainly quantified and analyzed using manual methods that are time-consuming and cumbersome. Automated and data-driven quantification and analysis of neurobehavioral datasets is therefore a pressing need. By leveraging advanced computer vision algorithms, we aim to analyze and interpret the behaviors of mouse models of developmental disorders in specific experimental tasks. By harnessing cutting-edge computer vision algorithms such as automatic detection of body-parts and tracking using DeepLabCut, combined with data-driven identification of “behavioral signatures” using the Behavior Segmentation of Open field in DeepLabCut (B-SOiD) pipeline, we will train a system to objectively perform data analysis on data gathered from videos of mice performing diverse experimental tasks including the three-chamber social test and the Erasmus Ladder test for motor learning. Through this, we seek to find patterns between the neurobiological and behavioral data from these mouse models, to learn how developmental disorders including Down syndrome and premature brain injury impact locomotor learning and neurophysiology. Our methods provide a platform to identify potential treatment options for humans with these developmental disorders.
Publication Date
4-23-2025
Project Designation
Honors Thesis
Primary Advisor
Dorian Borbonus, Aaron S. Sathyanesan
Primary Advisor's Department
Biology
Keywords
Stander Symposium, College of Arts and Sciences
Institutional Learning Goals
Scholarship
Recommended Citation
"Automated and data-driven discovery of behavioral signatures in preclinical models of developmental disorders" (2025). Stander Symposium Projects. 3917.
https://ecommons.udayton.edu/stander_posters/3917
Comments
9:00-10:15, Kennedy Union Ballroom