| Abstract |
In the context of smart city emergency management, public psychological resilience plays a crucial role in maintaining societal stability during crises. Traditional methods for assessing psychological distress are often reactive and lack real-time intervention capabilities. This study proposes a Support Vector Machine (SVM)-based model to predict and mitigate psychological distress through multisource data integration, including social media sentiment analysis, physiological indicators, and environmental stress factors. By leveraging machine learning, the model enables early detection of psychological distress patterns, facilitating targeted intervention strategies. The proposed framework consists of three components: data preprocessing and feature extraction from heterogeneous sources, SVM-based classification and prediction of psychological resilience levels, and an adaptive intervention system that provides real-time recommendations and alerts for emergency response teams. Experimental evaluations demonstrate the model's effectiveness in accurately detecting distress patterns, with an F1-score of over 85% in resilience classification tasks. A case study on a pandemic scenario highlights the practical applicability of the framework in urban emergency planning. The findings suggest that integrating AI-driven psychological resilience assessment into smart city infrastructures can enhance emergency preparedness and mental health support, providing a scalable and efficient solution for crisis management. © 2025 The Authors. |
