Search

Breast milk, as the most ideal food for infants, contains the main nutrients to meet the growth and development of infants. Lactose is an essential energy source in breast milk, the concentration of which varies based on a variety of circumstances, including maternal diet and lactation stage. Studies demonstrated that lactose content could affect the development of the brain and gut immunity of infants. We integrated the enzyme cascade reaction with colorimetry, and then established a data-prediction model utilizing multi-point absorbance analysis and a back-propagation (BP) neural network. Lactose underwent an enzymatic cascade to generate H 2 O 2, which oxidized the chromatic substrate 3,3′5,5′-tetramethylbenzidine (TMB). Absorbance at several time intervals and wavelengths was measured. Different from the traditional end-point detection methods and their single input data model, the absorbance data were detected at various time points and then fitted by BP neural network in this research. The detection of lactose in breast milk showed an average recovery rate of 99.14%. Meanwhile, our strategy is time-efficient and required no complex sample preparation processes. The robust nonlinear mapping and high fault tolerance of the BP neural network significantly enhances the stability and precision of the colorimetric analysis. Overall, this approach may be used for low-cost, convenient, and accurate lactose quantification in breast milk. [Display omitted] • A convenient, fast and low-cost method for the detection of lactose in breast milk. • Novel combined dynamic detection with multi-input nature of BP neural networks. • This study model had a better predicted performance of lactose concentration. [ABSTRACT FROM AUTHOR]