We have investigated disruptions mitigated with combined deuterium and noble gas injection in SPARC using multi-objective Bayesian optimization. More specifically, we have optimized the densities of the injected material, taking into account limits on the maximum runaway current, the transported fraction of the heat loss, and the current quench time. The simulations were conducted using the numerical framework DREAM (Disruption Runaway Electron Analysis Model). During deuterium operation, we found that runaway generation can be avoided with material injection, even when accounting for runaway electron generation from DD-induced Compton scattering. However, when including the latter, the region in the injected material density space corresponding to successful mitigation is reduced. During deuterium-tritium operation, acceptable levels of runaway current and transported heat losses are only obtainable at the highest levels of achievable injected deuterium densities. Furthermore, disruption mitigation was found to be more favourable when combining deuterium with neon, compared to deuterium combined with helium or argon.