Harnessing multimodal emotion features in depression detection across gender: Integrating large language model, acoustic fusion and facial expression recognition

Key Points

• Emotion features significantly enhance depression detection across genders, with a focus on audio, language, and visual data.
• The integration of large language models and acoustic fusion approaches demonstrates a 30% increase in accuracy.
• Analysis incorporates facial expression recognition models, highlighting their role in identifying emotional states in real-time.
• Findings stress the need for multi-faceted approaches in depression detection, emphasizing gender differences for better diagnosis.