Rising heatwaves threaten urban sustainability, necessitating a shift toward heat resilience. This study examines 38 cities across China’s three major coastal urban agglomerations (2016–2024) to quantify dynamic resilience responses. Utilizing a dual-threshold identification method and the Baidu Search Index to construct a Standardized Stress Index (SSI), the research evaluates urban heat vulnerability (UHV) through an exposure–sensitivity–adaptive capacity framework while applying NMF and machine learning models (XGBoost/SHAP) to analyze spatiotemporal heterogeneity. The results show that heatwave pressures peaked in 2022–2023, with Jing–Jin–Ji’s UHV evolving from localized clusters toward regional homogenization. Regional UHV profiles reveal that Jing–Jin–Ji is constrained by population pressures, the Yangtze River Delta (YRD) by resource allocation, and the Pearl River Delta by industrial attributes; notably, the YRD’s systematic coordination effectively offsets structural vulnerability. Furthermore, the optimized XGBoost model achieves strong predictive performance (R2 = 0.673), revealing that core factors like summer heat exposure intensity (SHE, 25.65% importance) trigger sharp non-linear surges in social stress upon crossing critical inflection thresholds (e.g., SHE at −0.10). The conclusion will lead to the formulation of differentiated, forward-looking climate adaptation strategies to enhance urban resilience across major regions.
Chen et al. (Sun,) studied this question.