Physiological Signal-Based Deepfake Detection Using Multi-Modal Fusion and Visual Attention

As deepfake technology advances, it poses severe harm to digital authenticity and information integrity. The traditional detection methods focus on surface level visual artifacts, but they are increasingly vulnerable to in generative adversarial networks. The utilization of physiological cues, such as eye blinks, cardiac rhythms, and the tiny facial expressions, as more precise detection markers is examined in this work. The research gaps and limitations are identified, and five state-of-the-art physiology-based approaches are reviewed. To address these problems, a new framework known as Physiological-Visual Attention Deepfake Detection (PVADD) is proposed. PVADD enhances accuracy, clarity, and resilience against hostile manipulations by fusing local physiology signals with visual embeddings through an attention mechanism.