Tokamak disruptions are a major challenge for the safe operation of next-step fusion devices such as ITER. A key concern is the formation of relativistic runaway-electron (RE) beams, which can deposit highly localized heat loads and threaten plasma-facing components. In ITER, the baseline disruption mitigation strategy relies on the rapid delivery of large impurity quantities via shattered pellet injection (SPI). Predictive modelling of RE generation in SPI-mitigated disruptions is therefore essential, but the computational cost of high-fidelity approaches motivates reduced models that enable extensive parameter scans. A recent study [1] used the DREAM fluid-RE model [2] to investigate RE generation in realistic ITER disruptions, incorporating (i) RE scrape-off during plasma vertical displacement [3] and (ii) plasmoid-drift effects on impurity deposition during pellet ablation [4]. A major remaining uncertainty is the hot-tail RE generation,which is typically misestimated by fluid models. In this work, we improve hot-tail seed predictions in ITER-relevant SPI simulations with the DREAM code, by formulating and numerically solving an isotropic kinetic equation, which resolves the momentum-space dynamics neglected in fluid models. We introduce a local, dynamic transition between fluid and kinetic equation sets, triggered once the isotropic model assumptions are satisfied. In particular, since the isotropic formulation employs a collision operator linearized about a cold background, it is activated in regions where impurity deposition rapidly builds up the cold electron population. Building on [5], which found that the DREAM fluid hot-tail model yields larger seeds than the isotropic kinetic approach, we re-assess this result for ITER by extending the analysis to the SPI mitigated disruptions considered in [1]. [1] Votta L. et al. (2026). Submitted to Nuclear Fusion. https://doi.org/10.48550/arXiv.2602.22177 [2] Hoppe M. et al. (2021). Computer Physics Communications, 268. https://doi.org/10.1016/j.cpc.2021.108098 [3] Vallhagen O. et al. (2025a). Journal of Plasma Physics, 91(3). https://doi.org/10.1017/S0022377825000327 [4] Vallhagen O. et al. (2025b). Plasma Physics and Controlled Fusion, 67(10).https://doi.org/10.1088/1361-6587/ae140f [5] Ekmark I. et al. (2024). Journal of Plasma Physics, 90. https://doi.org/10.1017/S0022377824000606