“Technology is a great enabler. It can make a real difference to our lives. That is why we are using eye-tracking technology and machine learning to aid the early diagnosis of crippling neurodegenerative diseases. ”
Dr Wenhao Zhang, Associate Professor of Computer Vision and Machine Learning
I currently have two main streams of research. One is around agriculture technology, using computer vision and machine learning. The other one is to adapt and apply a similar set of techniques to a very different interdisciplinary challenge: healthcare tech. I am looking at how we can use eye tracking and machine learning for disease diagnosis.
The eyes have it
Eye tracking is a really interesting area – for me personally and for useful, real-world applications – because eye movements are quite an important type of human behaviour. They tell people about your attention, for example, Are you interested? Are you tired? and potentially other clinical signs useful to disease diagnosis. We can absolutely read a lot from eye movements.
Our ambition is to use eye movement information, combined with machine learning analysis, to tell you if you might have a neurodegenerative disease such as dementia or its early pre-disease stages. We are in the early stages of our research work, and we are taking a very novel approach to diagnosis. But we believe in what we are doing based on evidence shown in the relevant literature and the successes we have had with applying machine vision and machine learning to other research areas. The ultimate aim is to enable everyone to be able to undertake eye tracking tests in their own home.
Pick up and play
Essentially, we are working to create a kind of an eye tracking test for dementia. The current gold-standard tests commonly need to be administered by a clinician. It takes time. It is labour intensive. It’s inconvenient for everyone really. Despite the availability of newer forms of eye tracking tests, you have to sit there and complete a visual task, which can be tracking a dot moving in different directions on a screen. The task can be somewhat counterintuitive. And it’s not fully representative of your cognitive functions and higher-level thinking skills.
What we want to do is build everything into a game. A game you can play on your laptop, phone and tablet. A game that uses the built-in camera to track and record your eye movements.
Through machine learning analysis of subtle but high-speed eye movement, we aim to spot errors in movement trajectories, latencies in responses to stimuli, and other more complex eye movement patterns. We would then compare those patterns to a healthy group of people who are not cognitively impaired and also to people who have dementia.
Increasing the data set
Also, by making the test more accessible we can increase our data set which often improves machine learning performance. While there are already commercial eye trackers available, they are commonly very, very expensive. Also, people can’t get to or use them easily, especially people who have a neurodegenerative condition. Through our research work, we want to enable just a normal webcam, or the camera on your phone or tablet, to achieve accurate and ubiquitous eye tracking. And ultimately, the more accessible the test, the more people will use it, and the bigger difference we can make.
Looking to tomorrow
Once we are up and running, we want to conduct some kind of longitudinal study. To see, for example, what happens if you play the game once a day or a few times a day. What happens over the course of a year. Are there any neuropsychological changes detectable by our approach? Could playing the game delay the onset of dementia? How early could dementia be detected before clinical symptoms manifest? Certain aspects of our future work are entirely hypothetical, but these are the kind of questions we want to ask of ourselves.
Contribution to the UN 2030 sustainable development goals
UWE Bristol is proud to align our research to the UN sustainable development goals. The above research aligns with the following goals:
