Artificial Neural Network in the Development of Halal Cosmetic Formulation Containing Okara

Abstract

The development of halal cosmetic formulations presents a challenge to obtain optimised formulations with desirable qualities as it involves many ingredients. The advancement of cosmetic technologies employs multivariate statistical techniques such as artificial neural networks (ANN) to optimise cosmetic formulation, which aims to overcome the shortcomings of traditional formulation methods, which are laborious and cumbersome. Okara is a by-product of the production of soy-based products. Okara has been found to have numerous benefits for many industries and has been discovered as a promising halal cosmetic ingredient. Okara is a plant-derived ingredient; it can be incorporated as a cosmetic ingredient if essential aspects of production are addressed, such as using permissible substances, manufacturing, storage, packaging, and delivery following Shariah requirements. This study aims to develop an optimised halal cosmetic soap formulation containing okara using ANN to achieve the desired hardness of the soap. The influential input variables were the main compositions of the okara soap formulations, containing different fatty acids and oils, and okara through a saponification process. In contrast, the hardness (N) of the soap was the response used as the output. Five different algorithms trained ANN. Generic Algorithm (GA) 6-09-1 was selected as the final optimum model to optimise the halal cosmetic soap formulation. GA modelling was further validated, and the experimentally obtained actual hardness (N) value was close to the predicted value. In conclusion, they were optimised formulating using ANN to produce a soap with desirable properties better than those of commercial ones.