<![CDATA[Earthquakes pose significant threats to human safety, critical infrastructure, and socioeconomic stability because their occurrence is highly complex and difficult to predict accurately in real time. Although conventional seismic monitoring systems have improved earthquake detection, they remain limited by computational constraints, delayed event recognition, and inadequate identification of nonlinear seismic patterns. This study evaluated the effectiveness of machine learning algorithms for real-time seismic detection and prediction and their contribution to disaster risk mitigation. A mixed-methods sequential explanatory design was employed using approximately 1.8 million seismic waveform segments representing 48,000 earthquake events collected from 320 monitoring stations across eight tectonically active regions. Quantitative analyses included comparative evaluation of supervised, ensemble, and deep learning algorithms using multivariate statistics, structural equation modeling, hierarchical regression, mediation, and moderation analyses, while qualitative evidence was examined through thematic analysis. Findings showed that deep learning and hybrid ensemble models consistently achieved higher prediction accuracy, computational efficiency, early warning reliability, and lower false alarm rates than conventional approaches. Improved prediction accuracy strengthened disaster response readiness, while dense sensor networks and institutional coordination enhanced operational effectiveness, supporting resilient earthquake risk mitigation and evidence-based emergency decision-making.]]>
Copyrights © 2026
