Clinical Scorecard: Comprehensive Behavioral Profiling for Classifying Depression Spectrum and Assessing Severity through Eye Tracking, Facial Expressions, and Language Analysis from Transcripts

At a Glance

Key Highlights

• Developed a multimodal framework for depression assessment
• Achieved accuracy and balanced accuracy approaching 0.90
• Facial features provided the dominant signal for classification
• Framework supports missing-modality handling and calibrated prediction
• Addresses limitations of prior binary depression-detection models

Guideline-Based Recommendations

Diagnosis

• Utilize multimodal behavioral markers for comprehensive assessment
• Consider subthreshold depression as a distinct category

Management

• Augment clinical assessment with objective multimodal tools

Monitoring & Follow-up

• Implement longitudinal monitoring of depressive-spectrum states

Risks

• Recognize the functional impairment and transition risk associated with subthreshold depression

Patient & Prescribing Data

Individuals experiencing normal control, subthreshold depression, or major depressive disorder

Framework may enhance understanding of individual symptom severity and classification

Clinical Best Practices

• Incorporate objective behavioral assessments in clinical settings
• Utilize quality-aware multimodal frameworks for improved diagnostic accuracy
• Focus on individual-level interpretation of depressive symptoms

Related Resources & Content

• Study on multimodal depression assessment