New research published in Psychological Medicine suggests that face and eye movement tracking may offer a reliable and inexpensive method of screening for depression. The tracking data detected heightened symptoms of depression with an accuracy approaching clinical significance.
Major depressive disorder (MDD) is often tricky to diagnose because symptom severity varies widely across patients, and the disorder often co-occurs with anxiety. Some studies have explored brain imaging data as a possible method of detecting depressive symptomology, but these methods remain expensive and require skilled technicians.
Study authors Aleks Stolicyn and his team conducted an experiment to test whether face and eye movement tracking might offer an alternative method of identifying patients with depressive symptomology. The researchers were motivated by previous research suggesting that depression is associated with particular face and eye movements when viewing emotional stimuli.
Stolicyn and his colleagues had a final sample of participants take part in two cognitive tasks while their face and eye movements were tracked. After patients with missing data were excluded, the final sample included 48 participants — 25 participants scoring above the threshold for clinical depression, and 23 who scored below the threshold and were considered asymptomatic.
The two cognitive tasks consisted of a working memory task and a sustained attention task. Importantly, both tasks contained distraction words that were flashed across the screen at various time points. These distraction words were either neutral, positive, or negative words.
At the end of the study, the researchers had obtained 663 facial movement measures and 132 eye-tracking measures for each subject. Eye movement metrics included: the delay between the distraction word appearing and the first eye fixation, the number of fixations, and the total time spent fixating on the word. Facial movements were coded using the Facial Action Coding System (FACS).
The researchers applied the data to a learning model called a support vector machine (SVM) to see if the model could discriminate between those with and without depressive symptomology. It was found that the eye-tracking data on its own resulted in a 65% accuracy in detecting symptoms, and the face movement data on its own allowed a 67% accuracy. Both of these measures combined, however, resulted in an accuracy of 79%.
Participants with depressive symptomology showed a lower number of fixations to positive words during the distraction portion of the working memory task. This is in line with studies showing that people with heightened depressive symptoms tend to fixate less on positive stimuli. Contrary to previous findings, however, those with elevated symptoms did not show more fixations to negative words.
The authors say the level of accuracy seen in their study is similar to what has been reported in studies using neuroimaging data. They suggest that their face and eye tracking methods might be ideal for conducting screenings for depression in smaller settings such as general practices. More advanced hospitals could then make use of MRI technology to inform treatment options for those with more severe symptoms.
Stolicyn and his colleagues note that their study was met with technical difficulties that resulted in the loss of data from 22 participants. They say that future studies should focus on methods of improving the technical setup of face and eye tracking systems.
The study, “Prediction of Depression Symptoms in Individual Subjects with Face and Eye Movement Tracking”, was authored by Aleks Stolicyn, J. Douglas Steele, and Peggy Seriès.