Simultaneous quantitative analysis of protein, carbohydrate and fat in nutritionally complete formulas of medical foods by near-infrared spectroscopy

A rapid, efficient, nondestructive and cost-effective method for simultaneous quantitative analysis of protein, carbohydrate and fat in nutritionally complete formulas of medical foods by Fourier transform near-infrared (FT-NIR) spectroscopy was developed. NIR spectra in the range of 3846-7692 cm−1 of 80 emulsion samples with different concentrations of three components were measured. The optimized spectral region was 4167-4878/5405-6452 cm−1. To enhance spectral features, NIR spectra were processed using a first derivative and smoothing method. Partial least squares regression (PLSR) was used to establish NIR models for protein, carbohydrate, fat and energy. The performances of the four models were evaluated by factors such as the values of correlation coefficient (r), root mean square error of prediction (RMSEV), root mean square error of calibration (RMSEC) and residual predictive deviation (RPD). The RMSEC of protein, carbohydrate, fat and energy was 0.29% (w/v), 0.71% (w/v), 0.30% (w/v) and 0.06 kcal/mL, respectively. The RMSEV was 0.29% (w/v), 0.50% (w/v), 0.24% (w/v) and 0.02 kcal/mL, respectively. The r value was 0.995, 0.968, 0.976 and 0.951, respectively. The RPD was 9.12, 4.44, 5.44 and 8.73, respectively. Furthermore, the four models were verified by cross and external validation. These results demonstrated that FT-NIR spectroscopy was a precise and rapid method for quantitative analysis of protein, carbohydrate and fat, and could be successfully applied to the determination of them in commercial medical foods.