A mixture-process design methodology, i.e., the crossed design, is proposed for experimental analysis and optimization. Five mixture materials for membrane formulation and two process factors for casting condition were fixed in the design methodology. The study was to generate a regression model for each of the responses, based on the experimental data and analysis variance of the study. Based on the response models, the optimal blend and casting condition were predicted. The highest desirability function, D, which prevailed from the optimization was 0.66. This optimal blend composition of the cast solution and the process condition in crossed design is proven to increase the membrane performance through the high membrane porosity, the high protein binding ability, and the fast lateral wicking rate.