News

A recent study presents a novel approach for the automated detection of fighting events in group-housed male mice by combining deep learning with Tecniplast's DVC® (Digital Ventilated Cage) technology. The work addresses a long-standing challenge in laboratory animal science, as aggressive interactions are often difficult to identify through routine observations and may occur without producing visible injuries.

The researchers developed and validated a convolutional neural network using approximately 3,000 manually annotated video segments linked to DVC® sensor data. The resulting model was able to detect fighting behavior with high specificity, achieving a false-positive rate of only 1% while correctly identifying 70% of fighting events.

Validation studies conducted across multiple research sites, mouse strains, group sizes, and housing configurations demonstrated the robustness of the approach. Importantly, the automated system frequently detected aggressive interactions before wounds were observed and identified conflicts that would have been missed using conventional welfare assessment methods. The findings also showed that the absence of visible injuries does not necessarily indicate the absence of aggression, highlighting the value of continuous behavioral monitoring.

Overall, the study demonstrates how the integration of artificial intelligence with home-cage monitoring can provide a scalable and objective method for behavioral assessment, supporting earlier welfare interventions while improving the quality and reproducibility of preclinical research. The article further reinforces the role of DVC® as a foundation for advanced digital biomarkers and AI-driven applications in modern laboratory animal facilities.

Find the study here (JAALAS, 2026)