AI Model Transforms Rodent Behavior Analysis with Multi-Scale Transformer

Researchers developed MSGL-Transformer, an AI system that automatically recognizes rodent social behaviors from pose data, eliminating error-prone manual scoring. The model achieved 87.1% accuracy on standard datasets, outperforming existing methods and enabling faster analysis of neural and behavioral mechanisms in rodent studies.