ABS 007: A Computer-Vision-Based Automated Continuous Behavioral Monitoring (ACBM) System Characterizing Neuromotor Behavior in Male 5XFAD Mice in Light and Dark Conditions

Siddhant Karmali ¹ ; Akash Nagaraj ² ; Patrick Gravelle, PhD ³ ; Alex Von Eckartsberg1 ⁴ ; Thomas Serre, PhD ² ; Roee Gutman, PhD ³ ; Justin Fallon, PhD ¹

¹ Department of Neuroscience, Brown University
² Carney Institute for Brain Science, Brown University
³ Department of Biostatistics, Brown University
⁴ Georgetown University McDonough School of Business

Van Wickle (2025) Volume 1, ABS 007

Introduction: Alzheimer's disease (AD) is a progressive neurodegenerative disorder characterized by cognitive decline and memory loss. Early and accurate diagnosis is crucial for effective treatment. Animal models, particularly the 5xFAD mouse model, play a vital role in understanding disease mechanisms and developing therapies. The 5xFAD mouse model is a widely used transgenic model of AD that overexpresses human APP and PSEN1 genes with five familial AD mutations. As early as 2-4 months of age, this animal exhibits amyloid-β plaque formation, gliosis, and cognitive impairments, which quickly progress into severe amyloid pathology. In this project, we report on neuromotor behaviors of the 5xFAD mouse as captured by computer vision-based ACBM.

Methods: A total of 20 mice (10 WT and 10 5xFAD) were recorded at 2, 4, and 8 month time points. Behavioral assessment is then performed using a Bi-LSTM RNN to generate behavior labels for human verification at 81% accuracy, relieving the inefficiency of manual annotation.

Results: We believe that there may be a connection between apathetic behavior, i.e. decreased grooming, and the severity of amyloid-β plaque formation. This may be an explanation for the decreased grooming seen in the 8-month 5xFAD mouse.
However, we see that 4-month 5xFAD mice groom more than their WT counterparts. This result, albeit needing further testing, highlights the need for data on 6-month 5xFAD mice.

Discussion: ACBM is an unbiased and robust tool for automated disease assessment:
Simultaneously characterized multiple mouse behaviors over a 5-day period
Captured typical mouse behavior, such as greater activity in the night and lesser activity during the day, and atypical mouse behavior.