Development of artificial neural network models to predict the concentration range of formation of m

Microemulsions have gained prominence in the research for biomolecules nanocarriers due to their thermodynamic stability and auto-organization. However, the formation of these systems requires a high experimental effort. To minimise it, an Artificial Neural Network to predict the concentration range of microemulsion formation was developed using weight fraction, hydrophilic-lipophilic balance as model input. We also evaluated the effect of including surfactant viscosity as input. Experimental data were generated using formulations with babassu oil, water, Tween® 80, and Labrasol®. After training, the proposed model presented accuracies up to 93%, with the addition of viscosity reducing the cross-validation variance.