Abstract
Human milk is the benchmark for formulating infant formula, the latter serving as a substitute when breastfeeding is not possible. Nevertheless, the lipid composition and structure of commercially available infant formulas still differ from those of human milk fat. Accordingly, this paper employs a computational-experimental framework to optimize formulations of prepared lipid (PF). The quality of the optimized product was further validated by analyzing volatile organic compounds (VOCs), color, lipid oxidation indicators, and oxidative stability. The results show that a total of 43 fatty acids (FA) were detected in the base oil, and palmitic acid, oleic acid, and linoleic acid are the main types of FA. Through computer simulation, 6 of PF were obtained, which are superior to commercial products (SP) in the similarity score of the parsimonious model, and PF1 has the highest score (84.15). Multivariate statistical analysis indicates that PF may be more suitable for use in infant formula milk powder due to its lipid composition. Gas chromatography-ion mobility spectrometry was used to study the VOCs content of PF and SP, and a total of 35 VOCs were identified. It was found that alcohols and ketones accounted for the highest proportion in PF, while Nitriles, Aldehydes, and Esters were the most abundant in SP. In the comparison of the basic physical and chemical indices between PF and SP, the peroxide value and p-anisidine value of PF are lower, and the overall oxidation stability is stronger than that of SP. This study provides a reference for the preparation and multi-dimensional evaluation of human milk fat substitutes.