Eye movement analysis for depression detection

Depression is a common and disabling mental health disorder, which impacts not only on the sufferer but also on their families, friends and the economy overall. Despite its high prevalence, current diagnosis relies almost exclusively on patient self-report and clinical opinion, leading to a number of subjective biases. Our aim is to develop an objective affective sensing system that supports clinicians in their diagnosis and monitoring of clinical depression. In this paper, we analyse the performance of eye movement features extracted from face videos using Active Appearance Models for a binary classification task (depressed vs. non-depressed). We find that eye movement low-level features gave 70% accuracy using a hybrid classifier of Gaussian Mixture Models and Support Vector Machines, and 75% accuracy when using statistical measures with SVM classifiers over the entire interview. We also investigate differences while expressing positive and negative emotions, as well as the classification performance in gender-dependent versus gender-independent modes. Interestingly, even though the blinking rate was not significantly different between depressed and healthy controls, we find that the average distance between the eyelids (`eye opening') was significantly smaller and the average duration of blinks significantly longer in depressed subjects, which might be an indication of fatigue or eye contact avoidance.

1. Introduction

Unlike mood fluctuation, clinical depression is a common mental disorder that lasts longer and causes disability and reduced function-ality. Moreover, at its most severe level, it might lead to suicide. A recent World Health Organization (WHO [1]) report estimated that 350 million people worldwide are affected by depression. It causes more than two-thirds of suicides each year [2]. The suicide risk is more than 30 times higher among depressed patients than that of the population without these disorders [3]. Although treatment of depression disorders has proven to be effective in most cases [4], misdiagnosing depressed patients is a common barrier [5]. Based on the WHO report, the barriers to effective diagnosis of depression include a lack of resources and trained health care providers. Moreover, the assessment methods of diagnosing depression rely almost exclusively on patient-reported or clinical judgments of symptom severity [6], risking a range of subjective biases. Our goal here is to investigate the general characteristics of depression, which we hope will lead to an objective affective sensing system that assists clinicians in their diagnosis and monitoring of clinical depression. Ultimately, we hope to assist patients with depression to monitor the progress of their illness in a similar way that a patient with diabetes monitors their blood sugar levels with a small portable device.