A community public health emergency resilience assessment framework based on contrastive learning and hyperbolic embedding

Introduction Recent global health crises have exposed critical gaps in community preparedness for public health emergencies, revealing that existing assessment frameworks often rely on generic indicators that fail to capture specific vulnerabilities. Methods We developed an evidence-based assessment framework through systematic analysis of 230 peer-reviewed studies using contrastive learning algorithms and hyperbolic embedding techniques. The framework was validated by 15 international experts across public health, urban planning, and disaster management disciplines. Results The framework comprises 39 indicators systematically organized into four actionable dimensions: (i) medical and safety measures, (ii) spatial design and infrastructure, (iii) community services and support, and (iv) landscape and ecology. Significantly, 23 indicators (59%) represent novel additions to public health emergency preparedness literature, including telemedicine infrastructure, community health surveillance