<p>With the rapid urbanization of the world and the global pandemic of some diseases, the resilience of urban public health systems has regained public attention as a measure of their ability to respond to public health emergencies. Furthermore, predictions and evaluations of future trends in urban public health system resilience can help identify possible gaps or vulnerabilities in the system. These information can enable appropriate plans to be made in advance to prevent any risks. Therefore, this paper proposes a new multi-attribute predictive analysis method to evaluate the resilience of urban public health systems. Firstly, to facilitate experts in expressing their opinions on the resilience of urban public health systems across various historical points, we utilize the Dynamic Probabilistic Linguistic Term Set (DPLTS) to describe expert opinions. Then, we propose a new definition of the time utility of DPLTSs for processing information in the form of DPLTSs for subsequent prediction. Considering the limited data related to the resilience of urban public health systems, we choose to combine grey prediction, suitable for small sample data prediction, with DPLTSs to construct the prediction model in our method. Finally, we integrate regret theory and DPLTSs as an evaluation model to rank the resilience of urban public health systems in cities. Our proposed method is proven through case studies to help identify potential risks in urban public health systems, provide support for future decision-making, and prepare for emergencies.</p>