Deep learning characterizes depression and suicidal ideation in young adults from eye movements

Area of Science:

• Neuroscience
• Psychiatry
• Computer Science

Background:

• Objective biobehavioral markers for mental health conditions are needed, as diagnosis often relies on subjective self-reports and clinical interviews.
• Attentional and mood biases are associated with depression and suicidal ideation.
• Eye tracking presents a potential method for quantifying these biases.

Purpose of the Study:

• To investigate eye tracking as an objective marker for depression and suicidal ideation.
• To analyze eye movement patterns during exposure to emotionally charged stimuli.
• To develop a deep learning framework for identifying mental health conditions using eye movement data.

Main Methods:

• Eye movements of 126 young adults were recorded while they read and responded to emotionally loaded sentences.
• A deep learning framework was employed to analyze intra-trial and inter-trial variations in eye movements.
• The model's performance was evaluated using Area Under the Curve (AUC) for classification tasks.

Main Results:

• The deep learning model achieved an AUC of 0.793 for identifying depression/suicidality against healthy controls.
• The model showed higher accuracy (AUC: 0.826) for identifying suicidality specifically.
• Moderate accuracy (AUC: 0.609) was observed in differentiating depressed from suicidal individuals.
• Discriminative eye movement patterns were more evident during response generation and with negative emotional stimuli.

Conclusions:

• Eye tracking can serve as an objective biobehavioral marker for assessing self-reported symptom severity in depression and suicidal ideation.
• Oculomotor control, influenced by emotional stimuli, provides quantifiable insights into mental health status.
• This approach holds promise for enhancing the objective assessment of mental health conditions.