|Harnly, James - Jim
|HARRINGTON, PETER - Ohio University
|JABLONSKI, JOSEPH - Food And Drug Administration(FDA)
|CHANG, CLAIRE - Food And Drug Administration(FDA)
|BOTROS, LUCY - Us Pharmacipeia (USP)
|POTTS, ALAN - Us Pharmacipeia (USP)
|MOORE, JEFF - Us Pharmacipeia (USP)
|BERGANA, MARTI - Abbott Nutrition
|WEHLING, PAUL - General Mills, Inc
Submitted to: Journal of Agricultural and Food Chemistry
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 7/10/2014
Publication Date: 7/10/2014
Citation: Harnly, J.M., Harrington, P., Jablonski, J., Chang, C., Botros, L., Potts, A., Moore, J., Bergana, M., Wehling, P. 2014. Characterization of near infrared spectral variance in the authentication of skim and nonfat dry milk powder collection using ANOVA-PCA, Pooled-ANOVA, and partial least squares regression. Journal of Agricultural and Food Chemistry. 62:8060-8067.
Interpretive Summary: Samples of skim milk powder (SMP) and non-fat dry milk (NFDM) from different suppliers, production sites, and with different processing temperatures were analyzed by NIR diffuse reflectance spectrometry over a period of three days. The purpose of the study was to determine the variance introduced by each of these factors and to determine if a single SMP or NDF standard could be used as a reference for identifying adulterants. It was found that each of these factors introduced statistically significant differences. In addition, it was found that the percentage of moisture, fat, and protein also produced systematic differences. As a result, it was concluded that a single SMP or NFDM sample could not be used as a reference for screening for adulterants. Instead, local models that are specific for milk class, process type, and production site will be necessary for screening.
Technical Abstract: Forty-one samples of skim milk powder (SMP) and non-fat dry milk (NFDM) from 8 suppliers, 13 production sites, and 3 processing temperatures were analyzed by NIR diffuse reflectance spectrometry over a period of three days. NIR reflectance spectra (1700-2500 nm) were converted to pseudo-absorbance and examined using (a) analysis of variance-principal component analysis (ANOVA-PCA) (b) pooled-ANOVA based on data sub-matrices, and (c) partial least squares regression (PLSR) coupled with pooled ANOVA. ANOVA-PCA score plots showed clear separation of the samples with respect to milk class (SMP or NFDM), day of analysis, production site, processing temperature, and individual samples. Pooled ANOVA provided statistical levels of significance for the separation of the averages, some of which were many orders of magnitude below 10-3. With the exception of the milk class and processing type factor interactions were insignificant. PLSR showed that the correlation with Certificate of Analysis (COA) concentrations varied from a weak coefficient of determination (R2) of 0.32 for moisture to moderate R2 values of 0.61 for fat and 0.78 for protein for this multinational study. However, with better correction of ambient moisture, improved storage conditions, and precise values for concentration instead of COA concentrations that are lot specific, the correlation with concentration would be expected to improve. In this study, pooled-ANOVA was applied for the first time to PLS modeling and demonstrated that even though the calibration models may not be precise, the contribution of the correlated peaks in the NIR spectra accounted for large proportions of the variation. The main effects were all significant in this study, which implies that local models that are specific for milk class, process type, and production site will be necessary for screening milk powders for adulterants. The significance of the variation with respect to the day of analysis indicates that a better approach for collecting reference spectra may be beneficial especially with respect to moisture.