Improving preclinical models in psychiatry

In biomedical research, rodents are often used to study neuropsychiatric conditions. Currently, scientists measure animal behaviour through short tests that provide only a snapshot of activity and can be stressful for the animals. These standard tests miss subtle but important changes in behaviour that occur naturally over time.

This project aims to create a more natural and continuous way to study rodent behaviour using artificial intelligence. By combining standard infrared cameras with open-source analysis tools, researchers can record activity continuously without specialised or expensive equipment. Computer programs will then analyse the video to recognise movement, sleep, and other daily activities. By tracking these patterns, the system can detect early signs that an animal might be stressed, unwell, or showing changes in mood.

This approach will give researchers a richer and more accurate picture of behaviour whilst improving animal welfare. It also makes these technologies more accessible to smaller laboratories by removing the need for costly proprietary cages currently available on the market.

The research team will develop and test a low-cost, open-source proof-of-concept system that can later be scaled into a full platform. Specific aims include extending existing home-cage monitoring approaches, identifying digital signatures of welfare and neuropsychiatric-like states, integrating digital behavioural measures with conventional tests, and establishing a sustainable, shareable computational platform.

The project will deliver a validated pilot workflow, FAIR-structured data templates, and an interactive dashboard for visualising results. These outputs will improve research quality and animal welfare under the 3Rs principles (replacement, reduction, refinement). The open-source toolkit will be reusable and accessible to any laboratory, making continuous monitoring feasible across institutions.

MDAP will provide essential technical expertise by containerising and deploying pose-estimation software on university high-performance computing infrastructure. MDAP will automate feature extraction, develop analytical modules to detect behavioural changes, build an interactive dashboard, establish FAIR-compliant data structures, and provide training and documentation. These capabilities are not available through existing Faculty resources, making MDAP's partnership essential to deliver tangible, scalable outputs within six months.