Emotion-Aware Mental Health Intervention Strategies on Mobile Computing Platforms

With the rapid advancement of mobile computing technologies and the continuous growth of global mobile device users, smartphones and other mobile terminals have created new possibilities for the widespread delivery of mental health services. At the same time, the increasing psychological pressure in modern society—especially in the post-pandemic era—has led to a surging demand for mental health interventions. Leveraging mobile computing platforms for accurate emotion recognition and effective intervention has thus become a critical area of research. However, current studies show that unimodal emotion recognition models achieve less than 60% accuracy in complex scenarios, while multimodal fusion methods often overlook the impact of spatiotemporal context, resulting in significant accuracy degradation in cross-regional emotion recognition. Furthermore, existing approaches lack effective strategies to address data sparsity and noise in mobile environments. This study focuses on emotion analysis in mental health service dialogues conducted via mobile networks. We propose a multimodal fusion module and a long-distance emotion fusion module. The former integrates multi-source data—including text, voice, and facial expressions—for comprehensive emotion capture, while the latter constructs a cross-regional emotional feature mapping mechanism to incorporate spatiotemporal contextual information. The contribution of this study lies in overcoming the limitations of existing models in terms of recognition accuracy and adaptability to diverse scenarios, thereby offering a feasible technical framework for mental health interventions on mobile computing platforms and advancing the intelligent development of digital mental health services.