Abstract Heatwaves intensified by climate change have increasingly threatened public health, highlighting the need for proactive and spatially targeted interventions. This study aimed to provide scientific evidence for managing the risk of heat‐related illness (HRI) by integrating community‐level physical environments and sociodemographic characteristics and applying explainable artificial intelligence techniques. Based on the Hazard–Exposure–Vulnerability–Response framework presented in the IPCC Sixth Assessment Report, we evaluated 20 regression models using Seoul, South Korea, as a case study. Nonlinear models demonstrated superior predictive performance compared to linear models, and Shapley additive explanations analysis revealed that apparent temperature, urbanized area ratio, older‐adult population ratio, outdoor worker count, and accessibility to water areas were the most influential variables. Apparent temperature exhibited a distinct threshold with a sharp increase in HRI risk above 36°C, while less‐urbanized areas were associated with higher incidence rates. Communities with higher proportions of older adults and outdoor workers consistently demonstrated greater vulnerability, and the effect of accessibility to water areas was spatially limited within daily activity spaces. Given the weak linear associations between HRI incidence and all explanatory variables ( r < 0.2), nonlinear dynamics and interactions play a critical role in understanding HRI risk. This study provides actionable insights for designing targeted heat‐health policies that consider diverse community characteristics.

Abstract Heatwaves intensified by climate change have increasingly threatened public health, highlighting the need for proactive and spatially targeted interventions. This study aimed to provide scientific evidence for managing the risk of heat‐related illness (HRI) by integrating community‐level physical environments and sociodemographic characteristics and applying explainable artificial intelligence techniques. Based on the Hazard–Exposure–Vulnerability–Response framework presented in the IPCC Sixth Assessment Report, we evaluated 20 regression models using Seoul, South Korea, as a case study. Nonlinear models demonstrated superior predictive performance compared to linear models, and Shapley additive explanations analysis revealed that apparent temperature, urbanized area ratio, older‐adult population ratio, outdoor worker count, and accessibility to water areas were the most influential variables. Apparent temperature exhibited a distinct threshold with a sharp increase in HRI risk above 36°C, while less‐urbanized areas were associated with higher incidence rates. Communities with higher proportions of older adults and outdoor workers consistently demonstrated greater vulnerability, and the effect of accessibility to water areas was spatially limited within daily activity spaces. Given the weak linear associations between HRI incidence and all explanatory variables ( r < 0.2), nonlinear dynamics and interactions play a critical role in understanding HRI risk. This study provides actionable insights for designing targeted heat‐health policies that consider diverse community characteristics.

This study aims to quantitatively assess the resilience of 229 local governments in South Korea against heatwaves using the Relative Disaster Resilience Framework (Zobel et al., 2021). This method is an extension of the Resilience Triangle Theory that incorporates both loss magnitude and recovery time. The framework allows flexible and objective evaluation by employing resilience thresholds of zero cases of heat illness and average electricity consumption during the milder spring and autumn seasons. Physical resilience was assessed using the number of cases of heat illness, while socioeconomic resilience was evaluated through summer electricity usage. Results showed that socioeconomic resilience was generally higher than physical resilience, indicating that energy systems recover more effectively than public health systems during heatwave events. Latent profile analysis revealed three distinct regional groups: Group 1 (43.7%) with high physical and socioeconomic resilience, Group 2 (50.7%) with low physical but high socioeconomic resilience, and Group 3 (5.7%) with high physical but low socioeconomic resilience. These groupings provide critical insights into differentiated policy responses. Group 2 regions require improved access to emergency medical services and heatwave shelters, while Group 3 regions might benefit from enhanced energy efficiency programs and demand-side energy management. By combining quantitative metrics with a flexible theoretical framework, this study offers a replicable method for evaluating urban resilience to heatwaves. The findings can inform evidence-based policymaking and support the development of tailored local adaptation strategies in response to increasing climate-related risks.

This study developed and measured a disaster resilience index for local communities by utilizing the concept of “disaster resilience” to address the increasing damage caused by climate change-related disasters. Based on prior research, this study conceptualized disaster resilience into three components: resistance, recovery, and transformation. It further categorized resilience into four dimensions: social, economic, infrastructure, and institutional resilience. For each dimension, five measurement variables were selected, resulting in 60 measurement variables used to develop the disaster resilience index. To verify the validity of the index, its relationship with actual disaster damage was investigated using spatial regression analysis. The results of the analysis revealed a negative (-) relationship between the disaster resilience index and disaster damage, confirming that regions with higher disaster resilience experienced lower impacts from disasters.