Sirius AI System Speeds Up Search for Treatments for Depression and Dementia

Scientists at the Sirius University have developed an AI-based system that analyzes laboratory rat behavior in minutes instead of requiring hours of manual video processing.

The neural network was trained on more than 7,000 manually annotated frames, each marking key points of the animal’s body, including the nose tip, paws, and torso. The algorithm can automatically classify behaviors such as running, freezing, and standing on hind legs. Classification accuracy reached 84% for hyperactive animals and 98% for rats with serotonin deficiency. The system enables researchers to evaluate potential treatments for depression, dementia, and other affective disorders at the preclinical stage.

The findings were published in the journal Neuroscience. The research involved specialists from St. Petersburg State University, the Institute of Experimental Medicine, and the Pavlov Institute of Physiology of the Russian Academy of Sciences.