Design of formulated fragrant products using rough set machine learning and molecular design tools

Fragrance is a desirable and often significant quality in various personal care products. While the components of fragrance -based products may initially be unknown, their intended attributes and functions are typically well understood. However, predicting odour is challenging due to the complex nature of fragrance. In addition, it is essential to consider the physicochemical properties, product quality, and all other associated aspects in a holistic manner. In this work, a systematic framework has been developed to design fragrant products. The proposed framework integrates Rough Set Machine Learning (RSML) into a computer -aided molecular design (CAMD) model for fragrance design. RSML is employed to generate deterministic rules based on the structural attributes of molecules, enabling the prediction of odour attributes. The results from the model are utilised as decision rules, establishing quantitative connections between the structural attributes of a molecule and its odour characteristics. The suitable surfactants required in the final product are subsequently designed through another CAMD model. The remaining components in the formulation are determined based on the suitability of ingredients for their defined roles. A case study was conducted to demonstrate the applicability of this suggested framework in developing fragrances for use in liquid soap.