New robot to detect Parkinson’s disease could be used on humans in the next 5 years
Researchers at the University of Leicester have created the world’s first artificial intelligence system capable of detecting Parkinson’s disease. The major breakthrough could lead to improved diagnosis in humans within the next five years.
The robot consists of a multi-camera system that uses motion and spatio-temporal data to monitor social behaviours in mice. It can then identify animals with forms of Parkinson’s disease by recording observational traits, such as posture, alongside genetic behavioural anomalies. It’s the first time in the world that researchers have developed an AI system to detect the condition in mice and the overall detection success rate (71.4%) is nearly 10% better than any previous assessment models (62.6%).
Professor Huiyu Zhou, an expert in machine learning at the University of Leicester, led the project, which has been published in the peer-reviewed journal, IEEE Transactions on Image Processing.
Parkinson’s disease is caused by a loss of nerve cells in part of the brain called the substantia nigra. As a result, dopamine – a chemical that helps to regulate movement of the body – drops. Experts don’t yet know what causes the disease and there’s no known cure, but early treatment can reduce symptoms and prolong life. Currently, 145,000 people are living with the disease in the UK and it’s estimated that the number will grow to 172,000 by 2030 due to the growing, ageing population.
The use of artificial intelligence is growing in the medical field, particularly in diagnostics. The study of mouse social behaviours is also increasingly being used in neuroscience research but automated quantification of mouse behaviours poses a challenging problem. Tracking methods can be intrusive and consequently interfere with the movements of mice in a dynamic environment. Previous monitoring systems have mainly relied on sensors, including radio-frequency identification transponders, infrared sensors, and photo beams. However, these approaches can only be applied to simple behaviours such as running and resting and they cannot be used to handle more complex mouse behaviours such as eating, attacking, or sniffing.
The vision-based system, designed by researchers at the University of Leicester, is able to recognise more subtle, complex mouse behaviours and can, therefore, perform automated quantification of social behaviours for freely interacting mice. The technology will now be further developed to address an even wider range of conditions and scenarios. Researchers then hope it will be ready to trial in humans in the near future.