Decoding depression with computer vision-assisted analysis of synchronized facial expressions.

Area of Science:

• Neuroscience
• Psychology
• Computer Science

Background:

• Facial expressions are crucial nonverbal indicators of emotional states.
• They show potential as markers for affective disorders like depression.

Purpose of the Study:

• To identify diagnostic cues for depression using facial expression synchronization.
• To develop depression detection models based on naturalistic emotional responses.

Main Methods:

• Utilized a naturalistic paradigm and inter-subject correlation (ISC) framework.
• Analyzed time-series activity patterns of action units (AUs) from facial expressions.
• Developed depression detection models using AU-ISC vectors as features.

Main Results:

• Models achieved 72-90% accuracy in detecting depression.
• AU-ISC models demonstrated significant diagnostic utility.
• Models using mean AU activity showed poor performance.

Conclusions:

• Facial expression data can advance objective diagnostic tools for depression.
• Facial expression dynamics offer a new direction for emotion research.
• This approach can complement established self-report measures.