The increasing number of inverter interfaced distributed generation units in electrical distribution grids can influence the functionality of protective devices in case of grid faults. The result can be an inadmissible delay in tripping time or failure to trip altogether. This study describes a novel approach to identify and amend potentially unreliable protection configurations. Reliance on conservative worst-case approximations of the grid state is avoided by using distribution system state estimation instead. Thereby, protection configurations can be more accurately matched to the specific grids' requirements. The overall computational effort is reduced using methods of regression analysis while maintaining original accuracy.