LOW COST, SCALABLE HOME CAGE MONITORING SYSTEM FOR PRE-WEANING MICE

Standard rodent behavioral assays often require invasive, out-of-cage procedures that are particularly disruptive during early postnatal development. This represents a significant methodological gap, as the early-life period is vital for understanding the ontogeny of neurodevelopmental disorders like schizophrenia and autism. Our research addresses this by utilizing computer vision and deep learning to enable continuous, non-invasive monitoring of pre-weaning group-housed mice within their home-cage environment, facilitating more robust and reproducible longitudinal data.

Methods and System Design

We focus on mice carrying mutations in the GRIN2A gene, which alters NMDA receptor subunit composition. While NMDA dysfunction is a known contributor to schizophrenia, its specific influence on early-life behavioral trajectories remains under-explored. To study this, I developed a low-cost, high-throughput video monitoring system designed for easy deployment in standard animal facilities. This system is designed to be easy-to-implement and open-source. The setup enables:

• Real-time motion tracking via automated algorithms.

• Post-hoc behavioral analysis using AI-driven trajectory modeling.

• Scalable data pipelines to manage the high computational demands of continuous recording.

Results and Applications

I will present the technical hurdles overcome during the implementation of this pipeline, particularly regarding the processing of large-scale datasets. Furthermore, I will demonstrate how these analytic methods extract meaningful insights from pup behavior that traditional assays might miss. Beyond neurodevelopmental research, these tools offer significant promise for enhancing animal welfare monitoring and refining longitudinal phenotyping across the lifespan.