Infrared spectroscopy coupled with machine learning algorithms for predicting the detailed milk mineral profile in dairy cattle

Aug 4, 2026

Upload by Aurelia

Milk minerals support human health while offering informative signals for milk quality and cow well-being. This study evaluates Fourier Transformed mid-Infrared (FTIR) spectroscopy to predict a detailed panel of 17 macro, trace, and environmental elements in bovine milk using partial least squares regression (PLS) and machine learning. AutoML models significantly exceed PLS in mineral prediction performance. For macrominerals, R2 spans 0.59–0.78, with strong results for Cu and B (R2=0.66 and 0.74) and moderate predictability for Fe, Mn, Zn, and Al (R2=0.48–0.58).

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages11

Reading Time28 min

AccessFree to download

Introduction

Milk minerals and their relevance

Conventional mineral quantification methods

Motivation for FTIR and machine learning

Materials and methods

FTIR spectroscopy and preprocessing

Modeling approaches (PLS and autoML)

Model evaluation and performance metrics

Results and discussion

Overall prediction performance

Performance by mineral category

Informative wavelengths and interpretability

Conclusions

Rapid and cost-effective mineral quantification

Food Chemistry 461 (2024) 140800 Contents lists available at ScienceDirect Food Chemistry journal [homepage: www.elsevier.com/locate/foodchem](homepage: www.elsevier.com/locate/foodchem) | Infrared spectroscopy coupled with machine learning algorithms for predicting the detailed milk mineral profile in dairy cattle | | | | | --- | --- | --- | --- | | *
Vittoria Bisutti , Lucio Flavio Macedo Mota , Diana Giannuzzi , Alessandro Toscano , Nicolo` Amalfitano , Stefano Schiavon , Sara Pegolo , Alessio Cecchinato
Department of Agronomy, Food, Natural Resources, Animals and Environment (DAFNAE), University of Padova, Viale dell’ Universit`a 16, 35020, Legnaro (PD), Italy | | | | | A R T I C L E I N F O | | A B S T R A C T | | | Keywords:
FTIR prediction
ICP-OES
autoML
Informative wavelengths | | Milk minerals are not only essential components for human health, but they can be informative for milk quality and cow’s health. Herein, we investigated the feasibility of Fourier Transformed mid Infrared (FTIR) spectroscopy for the prediction of a detailed panel of 17 macro, trace, and environmental elements in bovine milk, using partial least squares regression (PLS) and machine learning approaches. The automatic machine learning significantly outperformed the PLS regression in terms of prediction performances of the mineral elements. For macrominerals, the R2 ranged from 0.59 to 0.78. Promising predictability was achieved for Cu and B (R2 = 0.66 and 0.74, respectively) and more moderate ones for Fe, Mn, Zn, and Al (R2 from 0.48 to 0.58). These results provide a reliable basis for a rapid and cost-effective quantification of these traits, serving as a resource for dairy farmers seeking to enhance the quality of milk production and optimize cheese properties. | | 1. Introduction Bovine milk is a complex food that serves as a valuable source of essential components, each with different biological functions. Nutritionally, it is a rich reserve of essential amino acids, fatty acids, minerals, and vitamins. Moreover, it harbors immunoglobulins, hormones, and various other bioactive compounds (Haug et al., 2007). Even though minerals constitute only a small proportion (0.7%) of milk components (Kaufmann & Hagemeister, 1987), they are essential to human health as they contribute to different physiological functions such as the skeletal tissue formation, bone maintenance, biosynthesis of specific molecules, acid-base regulation and nerve impulse transmission (Stergiadis et al., 2021). In addition, some minerals play an important role in determining milk technological properties. Specifically, Ca and P serves as fundamental components of casein micelles, directly impacting both milk coagulation and final consistency of coagulum (Gustavsson et al., 2014; Stocco et al., 2021). Furthermore, elements such as Na and K have proven to be potential indicators of udder health as both their concentration is influenced by the modified permeability of blood-milk membrane induced by mastitis (Khatun et al., 2019). The concentration of milk minerals is dependent also on different factors, like the animals’ feeding systems and environmental features. Among milk minerals, iodine has been poorly described given the difficulty in determining its actual content in milk (Niero et al., 2020), which is influenced by different aspects such as the use of teat dipping sanitizers, differences in farming system and dietary supplements. Milk mineral elements are conventionally quantified through analytical techniques such as mass spectrometry or inductively coupled plasma - optical emission spectrometry (ICP - OES) (Di et al., 2009; Soyeurt et al., 2009). However, these methods are time consuming, expensive and require expert personnel, and therefore unsuitable for large scale analysis. Therefore, finding fast and affordable methods to measure the milk mineral profile might be beneficial for the dairy sector, considering the impli

FAQs

What is the main goal of this research on bovine milk?

To assess whether FTIR spectroscopy, combined with machine learning, can predict a detailed set of 17 milk mineral elements efficiently and accurately.

Which modeling approaches are compared for predicting milk minerals?

Partial least squares regression (PLS) is compared against automatic machine learning (autoML) approaches.

How well does autoML predict the mineral elements in milk?

AutoML outperforms PLS, achieving macromineral R2 values from 0.59 to 0.78, with stronger predictability for Cu and B and more moderate performance for Fe, Mn, Zn, and Al.

Aug 4, 2026

Upload by Aurelia