A Polish team of researchers has combined eye movement tracking with artificial neural networks to create a tool that will support early diagnosis of mental and neurological disorders. 

Working on the development of the new diagnosis method were psychologists and experts in artificial intelligence research, including Dr. Krzysztof Krejtz and Dr. Izabela Krejtz of SWPS University, Dr. Karol Chlasta of Kozminski University, and Dr. Katarzyna Wisiecka of the Warsaw University of Economics and Human Sciences. The study was published in the International Journal of Marketing, Communication and New Media. 

Researchers from three Polish universities have developed an AI-based system for the rapid detection of mental disorders. It detects depression, anxiety and other disorders in just 10 seconds by analyzing eye movement. 

A total of 101 people took part in the study – including patients diagnosed with depression, people with social anxiety and healthy participants (the control group). The subjects looked at the photos of faces with different emotional expressions for 10 seconds while special sensors in oculographs recorded their eye movements. The collected data was used to generate so-called “gaze paths,” which were then analyzed by models of deep convolutional neural networks that learn to recognize patterns in visual data and operate on a similar principle to the human brain, that is, processing information layer by layer and searching for meaningful signals. 

When it comes to distinguishing between depression and social anxiety, the accuracy of the method reaches 60-70%, but the researchers see room for further improvement.