"The use of high-frequency stimulation for conduction block of the pudendal nerve has potentially high benefits for patients suffering from non-neurogenic urinary retention. Special care has to be put in the design of stimulation parameters to ensure safe operation and prevent electrode and tissue damage. While high-frequency conduction block has been studied and used for many years, only standard waveforms, such as charge balanced sinewaves and square waves, have been utilized. Several studies have anticipated that the use of non-standard, non-symmetrical and slightly charge unbalanced waveforms may provide electrochemically safer stimulation protocols. In this computational simulation study, the MRG model is combined with an electrode-tissue interface (ETI) model based on in vivo experimental data to create a computational model capable of assessing both the efficacy and electrochemical safety of any given stimulation waveform. This model is coupled to a differential evolution algorithm to find the optimal waveform parameters that ensure a successful conduction block and a minimized charge injection through irreversible faradaic reactions. The classical DE algorithm is adapted to include several improvements such as evolutionary adaptive parametrization, elitism, and variable pattern to increase its performance. Additionally, acknowledging the fact that the axonal model is the main bottleneck in computational terms, an improvement baptized as "model down-sampling" is presented. Model down-sampling consists on only executing the axonal model to determine the effectiveness of the block once every N generations. This modification manages to double the execution speed without compromising accuracy. The results show that non-standard waveforms with a slight charge imbalance keep the ETI voltage well within the narrow electrochemical safe window of -0.25V and 0.55V, thus avoiding any irreversible charge injection process. The obtained waveforms show a 39.8% improvement on the safety margin with respect to the best performing standard stimulation waveform. The obtained results prove that well designed non-standard waveforms can lead to electrochemically safer high-frequency stimulation."