Integrating mid-infrared spectroscopy, machine learning, and graphical bias correction for fatty acid prediction in water buffalo milk

BACKGROUND

Buffalo milk, constituting 15% of global production, has higher fatty acids content than Holstein milk. Fourier-transform mid-infrared (FT-MIR) spectroscopy is widely used for dairy analysis, but its application to buffalo milk, with larger fat globules, remains understudied. The ultimate goal of this study is to develop machine learning models based on FT-MIR spectroscopy for predicting fatty acids in buffalo milk and to assess the accuracy of commercial milk analyzers. This research provides a convenient, fast, and environmentally friendly method for detecting the fatty acid composition in buffalo milk.

RESULTS

We employed six machine learning algorithms to establish a detection model for 34 fatty acids in buffalo milk. The predictive models demonstrated robust capabilities for high-content fatty acids [C14:0, C15:0, C16:0, C17:0, C18:0, C18:1, saturated fatty acid (SFA), monounsaturated fatty acid (MUFA)], with errors within a 15% range. Traditional FT6000 detection methods exhibited limitations in measuring SFAs and polyunsaturated fatty acids (PUFA). Implementing a mean difference correction of 0.21 for MUFAs and applying regression equations (SFA × 1.0639 + 0.0705; PUFA × 0.5472 + 0.0047) significantly improved measurement accuracy.

CONCLUSION

This study successfully developed a predictive model for fatty acids in Mediterranean buffalo milk based on FT-MIR spectroscopy. Additionally, a correction was applied to the existing measurement device, FT6000, enabling more accurate measurements of fatty acids in buffalo milk. The findings have practical implications for the food industry, offering a faster and more reliable approach to assess and monitor fatty acid composition in buffalo milk, potentially influencing product development and quality control processes. © 2024 Society of Chemical Industry.

SVR Chemometrics to Quantify β-Lactoglobulin and α-Lactalbumin in Milk Using MIR

Protein content variation in milk can impact the quality and consistency of dairy products, necessitating access to in-line real time monitoring. Here, we present a chemometric approach for the qualitative and quantitative monitoring of β-lactoglobulin and α-lactalbumin, using mid-infrared spectroscopy (MIR). In this study, we employed Hotelling T2 and Q-residual for outlier detection, automated preprocessing using nippy, conducted wavenumber selection with genetic algorithms, and evaluated four chemometric models, including partial least squares, support vector regression (SVR), ridge, and logistic regression to accurately predict the concentrations of β-lactoglobulin and α-lactalbumin in milk. For the quantitative analysis of these two whey proteins, SVR performed the best to interpret protein concentration from 197 MIR spectra originating from 42 Cornell University samples of preserved pasteurized modified milk. The R2 values obtained for β-lactoglobulin and α-lactalbumin using leave one out cross-validation (LOOCV) are 92.8% and 92.7%, respectively, which is the highest correlation reported to date. Our approach introduced a combination of preprocessing automation, genetic algorithm-based wavenumber selection, and used Optuna to optimize the framework for tuning hyperparameters of the chemometric models, resulting in the best chemometric analysis of MIR data to quantitate β-lactoglobulin and α-lactalbumin to date.

Feeding spent hemp biomass to lactating dairy cows: Effects on performance, milk components and quality, blood parameters, and nitrogen metabolism

The legalization of industrial hemp by the 2018 Farm Bill in the United States has driven a sharp increase in its cultivation, including for cannabinoid extraction. Spent hemp biomass (SHB), produced from the extraction of cannabinoids, can potentially be used as feed for dairy cows; however, it is still illegal to do so in the United States, according to the US Food and Drug Administration Center for Veterinary Medicine, due to the presence of cannabinoids and the lack of data on the effect on animals. To assess the safety of this byproduct as feed for dairy cows, late-lactation Jersey cows (245 ± 37 d in milk; 483 ± 38 kg body weight; 10 multiparous and 8 primiparous) received a basal total mixed ration (TMR) diet plus 13% alfalfa pellet (CON) or 13% pelleted SHB for 4 wk (intervention period [IP]) followed by 4 wk of withdrawal period (WP), where all cows received only the basal TMR during WP. The dry matter intake (DMI), body weight, body condition score, milk yield, milk components, and fatty acid profile, blood parameters, N metabolism, methane emission, and activity were measured. Results indicated that feeding SHB decreased DMI mainly due to the low palatability of the SHB pellet, as the cows consumed only 7.4% of the total TMR with 13.0% SHB pellet offered in the ration. However, milk yield was not affected during the IP and was higher than CON during the WP, leading to higher milk yield/DMI. Milk components were not affected, except for a tendency in decreased fat percentage. Milk fat produced by cows fed SHB had a higher proportion of oleate and bacteria-derived fatty acids than CON. The activity of the cows was not affected, except for a shorter overall lying time in SHB versus CON cows during the IP. Blood parameters related to immune function were not affected. Compared with CON, cows fed SHB had a lower cholesterol concentration during the whole experiment and higher β-hydroxybutyric acid during the WP, while a likely low-grade inflammation during the IP was indicated by higher ceruloplasmin and reactive oxidative metabolites. Other parameters related to liver health and inflammatory response were unaffected, except for a tendency for higher activity of alkaline phosphatase during IP and a lower activity of gamma-glutamyl transferase during WP in the SHB group versus CON. The bilirubin concentration was increased in cows fed SHB, suggesting a possible decrease in the clearance ability of the liver. Digestibility of the dry matter and protein and methane emission were not affected by feeding SHB. The urea, purine derivatives, and creatinine concentration in urine was unaffected, but cows fed SHB had higher N use efficiency and lower urine volume. Altogether, our data revealed a relatively low palatability of SHB affecting DMI with minimal biological effects, except for a likely low-grade inflammation, a higher N use efficiency, and a possible decrease in liver clearance. Overall, the data support the use of SHB as a safe feed ingredient for lactating dairy cows.

Establishment and Validation of Fourier Transform Infrared Spectroscopy (FT–MIR) Methodology for the Detection of Linoleic Acid in Buffalo Milk

Buffalo milk is a dairy product that is considered to have a higher nutritional value compared to cow’s milk. Linoleic acid (LA) is an essential fatty acid that is important for human health. This study aimed to investigate and validate the use of Fourier transform mid-infrared spectroscopy (FT-MIR) for the quantification of the linoleic acid in buffalo milk. Three machine learning models were used to predict linoleic acid content, and random forest was employed to select the most important subset of spectra for improved model performance. The validity of the FT-MIR methods was evaluated in accordance with ICH Q2 (R1) guidelines using the accuracy profile method, and the precision, the accuracy, and the limit of quantification were determined. The results showed that Fourier transform infrared spectroscopy is a suitable technique for the analysis of linoleic acid, with a lower limit of quantification of 0.15 mg/mL milk. Our results showed that FT-MIR spectroscopy is a viable method for LA concentration analysis.

Diurnal variation of milk fatty acids in early-lactation Holstein cows with and without hyperketonemia

Estimates of milk constituents by Fourier-transform mid-infrared (FTIR) analysis have been shown to be a useful tool in monitoring energy deficit in early-lactation dairy cows. Our objectives were to describe the diurnal variation in milk fatty acids (FAs) and estimate the association of hyperketonemia with concentrations and diurnal patterns of FTIR estimates of milk FA. Blood samples were collected via jugular catheters bihourly for 5 d from multiparous Holstein cows (n = 28) enrolled between 3 and 9 days in milk. Milk samples were collected thrice daily at 0600, 1400, and 2200 h for d 2, 3, and 4 of the study period. Cows were retrospectively classified as hyperketonemic (HYK; n = 13) or non-HYK (n = 15) based on blood beta-hydroxybutyrate (bBHB) concentrations analyzed during the study period. Cows were classified as HYK if bBHB was ≥ 1.2 mmol/l for ≥ 50% (22/44) of bihourly timepoints; cows were classified as non-HYK if bBHB was ≥ 1.2 mmol/l for < 50% of bihourly timepoints. The HYK cows had bBHB ≥ 1.2 mmol/l for 31.4 ± 6.8 timepoints while the non-HYK cows had bBHB ≥ 1.2 mmol/l for 8.0 ± 3.9 timepoints. We used generalized linear mixed models to analyze concentrations of milk FA over time and differences between HYK groups. The relative percentage of de novo, mixed, and preformed FAs all followed diurnal patterns, however only the yield of preformed FA diurnally cycled, reaching a nadir at 0600 h and peaking at 1400 h. The yield per milking of preformed FA was also greater in the HYK cows than in the non-HYK cows. Oleic acid in milk followed a similar diurnal pattern to the yield of preformed FA, likely driving the cyclical nature of preformed FA. Finally, stearic acid was greater in HYK cows. Our results suggest that FTIR estimates of milk FA offer the potential to provide insight on the energy status of early-lactation cows, and when interested in understanding the absolute concentrations and yields of milk FA, diurnal variation should be considered.

Nitrogen isotopic discrimination as a biomarker of between-cow variation in the efficiency of nitrogen utilization for milk production: A meta-analysis

Estimating the efficiency of N utilization for milk production (MNE) of individual cows at a large scale is difficult, particularly because of the cost of measuring feed intake. Nitrogen isotopic discrimination (Δ¹⁵N) between the animal (milk, plasma, or tissues) and its diet has been proposed as a biomarker of the efficiency of N utilization in a range of production systems and ruminant species. The aim of this study was to assess the ability of Δ¹⁵N to predict the between-animal variability in MNE in dairy cows using an extensive database. For this, 20 independent experiments conducted as either changeover (n = 14) or continuous (n = 6) trials were available and comprised an initial data set of 1,300 observations. Between-animal variability was defined as the variation observed among cows sharing the same contemporary group (CG; individuals from the same experimental site, sampling period, and dietary treatment). Milk N efficiency was calculated as the ratio between mean milk N (grams of N in milk per day) and mean N intake (grams of N intake per day) obtained from each sampling period, which lasted 9.0 ± 9.9 d (mean ± SD). Samples of milk (n = 604) or plasma (n = 696) and feeds (74 dietary treatments) were analyzed for natural ¹⁵N abundance (δ¹⁵N), and then the N isotopic discrimination between the animal and the dietary treatment was calculated (Δ¹⁵n = δ¹⁵N_animal - δ¹⁵N_diet). Data were analyzed through mixed-effect regression models considering the experiment, sampling period, and dietary treatment as random effects. In addition, repeatability estimates were calculated for each experiment to test the hypothesis of improved predictions when MNE and Δ¹⁵N measurements errors were lower. The considerable protein mobilization in early lactation artificially increased both MNE and Δ¹⁵N, leading to a positive rather than negative relationship, and this limited the implementation of this biomarker in early lactating cows. When the experimental errors of Δ¹⁵N and MNE decreased in a particular experiment (i.e., higher repeatability values), we observed a greater ability of Δ¹⁵N to predict MNE at the individual level. The predominant negative and significant correlation between Δ¹⁵N and MNE in mid- and late lactation demonstrated that on average Δ¹⁵N reflects MNE variations both across dietary treatments and between animals. The root mean squared prediction error as a percentage of average observed value was 6.8%, indicating that the model only allowed differentiation between 2 cows in terms of MNE within a CG if they differed by at least 0.112 g/g of MNE (95% confidence level), and this could represent a limitation in predicting MNE at the individual level. However, the one-way ANOVA performed to test the ability of Δ¹⁵N to differentiate within-CG the top 25% from the lowest 25% individuals in terms of MNE was significant, indicating that it is possible to distinguish extreme animals in terms of MNE from their N isotopic signature, which could be useful to group animals for precision feeding.

Routine Monitoring of Instrument Stability in a Milk Testing Laboratory With ASCA: A Pilot Study

Mid-infrared spectroscopy has been developed as a reliable and rapid tool for routine analysis of fat, protein, lactose and other components in liquid milk. However, variations within and between FTIR instruments, even within the same milk testing laboratory, present a challenge to the accuracy of measurement of particularly minor components in the milk, such as individual fatty acids or proteins. In this study we have used Analysis of variance–Simultaneous Component Analysis (ASCA), to monitor the spectral variation between and within each of four different FOSS FTIR spectrometers over each week in an independent milk testing laboratory over 4 years, between August 2017 and March 2021 (223 weeks). On everyday of each week, spectra of the same pilot milk sample were recorded approximately every hour on each of the four instruments. Overall, variations between instruments had the largest effect on spectral variation over each week, making a significant contribution every week. Within each instrument, day-to-day variations over the week were also significant for all but two of the weeks measured, however it contributed less to the variance overall. At certain times other factors not explained by weekday variation or inter-instrument variation dominated the variance in the spectra. Examination of the scores and loadings of the weekly ASCA analysis allowed identification of changes in the spectral regions affected by drifts in each instrument over time. This was found to particularly affect some of the fatty acid predictions.

Performance evaluation of an enzymatic spectrophotometric method for milk urea nitrogen

Our objective was to determine the within and between laboratory performance of an enzymatic spectrophotometric method for milk urea nitrogen (MUN) determination. This method first uses urease to hydrolyze urea into ammonia and carbon dioxide. Next, ammonia (as ammonium ions) reacts with 2-oxoglutarate, in the presence of reduced nicotinamide-adenine dinucleotide phosphate (NADPH) and the enzyme glutamate dehydrogenase (GlDH), to form l-glutamic acid, water, and NADP⁺. The change in light absorption at 340 nm due to the conversion of NADPH to NADP+ is stoichiometrically a function of the MUN content of a milk sample. The relative within (RSD_r) and between (RSD_R) laboratory method performance values for the MUN enzymatic spectrophotometric method were 0.57% and 0.85%, respectively, when testing individual farm milks. The spectrophotometric MUN method demonstrated better within and between laboratory performance than the International Dairy Federation differential pH MUN method with a much lower RSD_r (0.57 vs. 1.40%) and RSD_R (0.85 vs. 4.64%). The spectrophotometric MUN method also had similar method performance statistics as other AOAC International official validated chemical methods for primary milk component determinations, with the average of all RSD_r and RSD_R values being <1%. An official collaborative study of the enzymatic spectrophotometric MUN method is needed to achieve International Dairy Federation, AOAC International, and International Organization for Standardization official method status.