In this paper, we propose a systematized analytical and AI-assisted framework to compute the end-to-end latency of an unmanned aerial vehicle (UAV)-attached wireless sensor network (WSN) node communicating over a 5G linked communication system and to identify and minimize the latency within this link process. Recent literature has typically analysed networks or mobility, but our model integrates the latency contributions of WSN node, UAV platform and 5G network into a single mathematical model by considering the interactions of the three actors. With parameters of sensing, processing, transmission, and 5G routing, the analytical model enables us to quantify its components latency level that provides clear-base on the analysis of parameter sensitivity. In addition, we employ three AI-based optimization techniques to adaptively set system parameters to minimize latency while adapting to different network conditions, including Supervised Regression, Reinforcement Learning and Hybrid AI–Heuristic control. Using simulation-based evaluation we show that the hybrid approach obtains up to 33% less latency compared with the baseline, and up to 28% and 18% less latency than reinforcement learning and regression methods, respectively. These results confirm the feasibility of AI-driven latency adaptation for UAV-assisted WSNs over 5G, offering a practical and scalable approach toward next-generation low-latency aerial IoT systems.