Ensuring an adequate reserve, timely supply, and rational distribution of emergency medical supplies in the aftermath of public health emergencies is a critical factor in enhancing emergency response capabilities. To address this issue, this paper first establishes a scenario analysis model for emergency medical supply reserve and supply systems based on convolutional neural networks, using case attribute similarity. Subsequently, by analyzing the information sharing mechanism of reserve and supply entities, a multi-agent emergency medical supply reserve and supply system framework is constructed. Different scenarios such as material demand, entity responsibility, and emergency objectives are considered both vertically and horizontally to achieve resource balance and establish collaborative mechanisms for human, material, financial, and information resources. Taking the emergency medical supply reserve and supply during the COVID-19 pandemic as a case study, simulations are conducted using the Simpy platform. The results demonstrate that 1)scenario analysis models can ensure precise provisioning of emergency medical supplies; 2) the utilization of multi-agent technology promotes intelligent management of emergency medical supplies; 3) collaborative operational mechanisms reduce the risk of disruptions in the emergency medical supply chain. Finally, targeted measures are proposed for the assurance of emergency medical supply reserves and distribution to mitigate the impact of unforeseen events on societal stability.