Researchers from the University of St Andrews have developed an AI tool that reads animal movement from video and turns it into clear, human-readable descriptions, making behavioural analysis faster, cheaper, and scalable across species.
Published on Wednesday 21 January by The Royal Society, the PoseR plug has been developed to remove a major bottleneck in neuroscience, psychology and biology to enable larger faster, and more reproducible studies.
Animal behaviour is a window into how their brains work, or not, in the case of disease. Traditionally, scientists have spent hours manually scoring videos to analyse behaviour; often taking weeks or months per project. Speeding this up can therefore have a big impact.
This new breakthrough makes behaviour analysis much faster and more consistent, helping researchers study animal actions and brain function more efficiently.
The AI tool uses a method called "Graph Neural Networks". These networks can be fed data in the shape of graphs which correspond to the varying shapes of animals. It allows the user to determine what the subjects in the recording are doing, and gives categories of behaviour.
The tool was developed by a team of researchers led by Dr Maarten Zwart from St Andrews School of Psychology and Neuroscience, whose research looks at how the brain produces behaviour, They are particularly interested in how movements are produced, and using this tool will speed up their analyses.
He said "Our team aims to find out how the brain and spinal cord interact to produce all the different movements that we and other animals make. This tool, which started as a COVID lockdown project, is an exciting new development driven by Dr Pierce Mullen, along with Dr Holly Armstrong, our research technician, and two amazing undergraduate researchers, Beatrice Bowlby and Angus Gray."
It's hoped that by removing a major bottleneck in neuroscience, psychology and biology, it will enable larger, faster, and more reproducible studies as well as rapid screening of animal models for disease. Eventually making it easier to make fundamental discoveries and find cures for neurological diseases.
