Xiaodong Song1
Haohan Ding
94% of researchers rate our articles as excellent or good
Learn more about the work of our research integrity team to safeguard the quality of each article we publish.
Find out more
- 1Key Laboratory of Dairy Quality Digital Intelligence Monitoring Technology, State Administration for Market Regulation, Hohhot, China
- 2School of Artificial Intelligence and Computer Science, Jiangnan University, Wuxi, China
- 3Science Center for Future Foods, Jiangnan University, Wuxi, China
This paper summarizes the existing methods of milk powder detection, and classifies them according to the direct and indirect characteristics of the detection methods, mainly introducing the detection methods of milk powder nutrition, recombination characteristics, transportation convenience and sensory characteristics. The direct detection methods of milk powder include traditional chemical analysis and modern instrument technology, most of which are based on the International Dairy Federation (IDF) standard method and powder detection instrument method. These methods can give accurate quantitative results, but often require complex sample preparation processes and long experimental operations. The indirect detection methods of milk powder mainly use microscopic imaging, spectral analysis, electronic nose system, environmental parameter monitoring and other technologies to establish complex mathematical models and provide a fast and non-destructive alternative. In addition, this paper summarizes the development of milk powder quality detection in three main directions: first, the traditional chemical detection method to environmental protection indirect analysis technology; Secondly, the development direction of multidisciplinary comprehensive evaluation; Finally, there is the wider use of artificial intelligence (AI) and automation. Future developments in the field are expected to focus on innovation across disciplines, combining technologies such as spectroscopy, high-definition microscopic imaging, digital twin with modern technologies such as AI and the Internet of Things. These advances are expected to improve the efficiency, sustainability and intelligence of milk powder quality assessment systems, while ensuring their accuracy and reliability.
1 Introduction
Milk powder is a favorite source of nutrition for many consumers, and it lasts a long time. Therefore, its quality, safety and resilience directly affect the consumer experience. According to the World Health Organization, milk powder is an important source of protein and calcium worldwide (Pancholi et al., 2023), and is especially important for children and the elderly (Li et al., 2023; Nebbia et al., 2023). As more and more people drink milk powder, ensuring its nutritional value and ease of brewing has become a problem for the food industry to solve.
The main ingredients of milk powder include protein, fat and lactose. The quantity and proportion of these ingredients not only determine the nutritional value of the product, but also affect consumer acceptance and market competitiveness (Schaefer et al., 2023). In order to ensure the quality of milk powder, the industry usually uses a variety of methods to evaluate the physical and chemical properties, structural properties and sensory properties of milk powder (Wang et al., 2024). In this paper, the detection methods are divided into direct detection and indirect detection according to whether auxiliary indicators are used. The direct detection method is mainly formulated by ISO and other organizations, its advantage is high accuracy, the experimental process has standards to follow. However, this method also has the problems of complex operation, error prone and low efficiency. In contrast, indirect detection methods predict hard-to-measure parameters by measuring easily available indicators. The advantage of this method is that the detection speed is fast and the damage to the sample is small, but it also has the problem of low accuracy and relying on the quality of the direct detection data. In recent years, many efforts have been made to improve the accuracy of indirect detection methods. Khan et al. (2020a) summarized the application of hyperspectral imaging in real-time quality control of milk powder, and Sharma et al. (2021) modeled the hygroscopic isotherm in milk powder by using computational neural genetic algorithm. In actual production, sometimes the precise measurement of certain milk powder parameters is not as important as quickly determining whether the product meets regulatory requirements. In view of this situation, Liu et al. (2023b) developed a method combining near-infrared spectroscopy and algorithm for rapid detection of trace elements in infant formula milk powder, and Ding et al. (2023) realized the LSTM algorithm for rapid detection of energy content in infant formula milk powder.
In the following sections, this review will discuss in detail the direct and indirect methods for detection the physical and chemical parameters, remodeling properties, and sensory properties of milk powder. Particular attention will be paid to the role of AI technology in indirect detection methods. Finally, the future development trend of milk powder detection technology will be discussed.
2 Main focus of Milk powder quality detection
The quality detection of milk powder is mainly carried out based on the aspects of nutritional value, recombination difficulty, transportation efficiency, sensory experience and so on. This chapter discusses, in detail, the key indicators to be measured in milk powder quality detection.
2.1 Nutrient content
When checking the quality of milk powder, the content of protein, fat and lactose are the most critical chemical indicators, which directly affect the nutritional level and market performance of the product. According to ISO 23318:2022 (Poo et al., 2024), the protein content of whole milk powder accounts for 34% of skim milk solids, while the protein content of reconstituted milk powder should meet or exceed 16.7 g /100 g. Whole milk powder should have a fat content of 26 g/100 g or more, and lactic acid concentration of reconstituted milk powder should not exceed 18°T. According to ISO 20633:2015 (Vassos et al., 2024), infant formula has more stringent standards for protein, lactose, linoleic acid, vitamins, minerals, taurine and other ingredients than ordinary milk powder.
2.2 Ease of reconstitution
In addition to controlling the nutritional composition of milk powder, it is also important to precisely adjust its recombination characteristics, because this will affect the production efficiency and the stability of product quality. In addition, these characteristics will directly affect the feelings of consumers when using and the market competitiveness of the product. Wettability, dispersion, sinkability and solubility are the important criteria to evaluate the recombination characteristics of milk powder, and also the key factors to judge the quality of milk powder products (Ji et al., 2016b).
Wettability is defined as the diffusion and permeability of water molecules on the surface of milk powder particles when they come into contact with water molecules (Hailu et al., 2023), which directly affects the dissolution kinetics and dissolution uniformity of milk powder when dissolved, and is an important indicator in the preparation of milk powder (McSweeney et al., 2020). The wettability of milk powder is mainly affected by factors such as particle density, particle size distribution, porosity, specific surface area, surface charge characteristics, and interface properties (Fournaise et al., 2020). The study by Yasukuni et al. showed that particle size distribution and surface characteristics were the most important indicators affecting the wettability of milk powder.
Dispersibility is defined as the ability of milk powder aggregates to disperse into individual particles in water with gentle agitation (McCarthy et al., 2014). There is a strong correlation between milk powder dispersibility and wettability, and milk powders, which usually exhibit excellent dispersibility, also have excellent wettability. The dispersibility of milk powder is mainly affected by factors such as particle morphology, particle size distribution and surface characteristics. Crowley et al. (2015) showed that milk powders with particle sizes in the range of 150–250 μm had the best dispersion and performed better than 30% of those outside of this range. In another recent study, the effect of manufacturing parameters on the dispersion of milk powder was equally important (Schuck et al., 2016). Dispersibility is particularly important in infant formula, as it is used as an indicator for the preparation of milk powders, because infant formulas with excellent dispersibility will have higher nutrient utilization and digestibility.
Sinkability is an important functional property of milk powder, which means that milk powder particles can travel through the surface of the water when they hit it instead of floating on it (Anup Sharma et al., 2012). It is found that this characteristic has a great influence on the transformation effect, and is related to the wettability and dispersion. The main factor of sinkability is the relationship between particle density and surface tension. Dense particles sink better than lighter ones. In addition, the porosity of the particles also affects the subsidence. In the most suitable porosity range, the particles can both remain structurally intact and quickly immerse in water.
Solubility is an important characteristic of milk powder, which refers to the ability of milk powder to form a uniform solution after being remixed in water (Wu et al., 2022). This property involves several aspects, such as wetting, dispersion and sedimentation. Although it is difficult to study these individual processes, solubility is still the best criterion for evaluating the reaction between powder and water (Schuck et al., 2012). Studies have shown that solubility has a great impact on production efficiency and economic benefits. Sikand et al. (2011) found that protein content and storage conditions would affect the solubility of powder. The more protein, the lower the solubility. This conclusion is similar to those observed by Gaiani et al. (2007), who also noted that the composition of the powder’s surface has an important effect on its dispersibility in water. In addition, Crowley et al. pointed out that environmental factors, especially humidity and temperature during storage, would also significantly affect solubility.
2.3 Transport convenience
Flowability is an important feature of milk powder, which refers to the ease with which particles can move during packaging, storage and transportation. Physicochemical factors affecting fluidity include particle size distribution, cohesiveness between particles, amount of oil on the surface, average particle size and roughness of the surface. In particular, it should be noted that the particle size distribution is the key to determine the quality of fluidity; Powders with good fluidity usually contain more large particles and a small number of small particles. If the proportion of fine particles (≤100 μm) is too large, it will affect the flow and make the flow difficult. A higher proportion of larger particles (≥200 μm) helps to improve fluidity, as it reduces friction between large particles (Liu et al., 2023a). In addition, the content of water and fat can also affect mobility; Too much water makes the particles more likely to stick together and reduce fluidity, while a high fat content can form a lubricating film on the surface of the particles, helping to reduce friction and thus improve fluidity (Fitzpatrick et al., 2004).
Bulk density is an important criterion to evaluate the transportation economy and market display effect of milk powder, which directly affects the performance of milk powder in the process of production, storage, transportation and consumption. This is also an important factor in judging the quality of milk powder. Bulk density refers to the quality of milk powder per unit volume and is a key indicator used to measure the efficiency of packaging, storage and transportation. This property is affected by many factors, such as the size of the particles, the Spaces within the particles, and the arrangement of the particles in the container. In addition, determining bulk density is also very important for controlling transportation costs, because it will affect the economic benefits of long-distance transportation of milk powder (Ding et al., 2020b).
2.4 Sensory characteristics
Sensory characteristics such as color, fragrance and taste are very important for the overall evaluation of milk powder. Generally speaking, milk powder should be white or light yellow, mainly because of the fat content and heating process at the time of production (Czarnowska-Kujawska et al., 2024). For example, whole milk powder is usually yellowish in color, while skim milk powder is closer to white. In terms of smell, milk powder should have the taste of fresh milk, not sour or other strange taste, which usually indicates the preservation of the product and whether it is fresh (Guo et al., 2023). Taste is the most direct feeling of consumers; Good milk powder should taste delicate, sweet, should not have bitter, sour or metallic taste and other bad taste (Coolbear et al., 2022).
3 Direct methods for quality detection of Milk powder
In order to accurately measure the important characteristics of milk powder, international standards organizations such as IDF have developed quantitative measurement methods. These methods generally require the use of laboratory equipment to directly detection the milk powder. In this paper, direct detection refers to the use of laboratory instruments, by measuring the number of some specific indicators, to directly or indirectly determine the indicators to be detected. This chapter will explain these direct detection methods in detail.
3.1 Nutritional value assessment methods
There are several standard techniques for the direct detection of nutrients in milk powder. Fat content was determined mainly by Soxhlet extraction and modified Mojonnier method (Gallier et al., 2017). For protein determination, Kjeldahl nitrogen determination is still the most reliable method, but spectrophotometry can also be a faster option (Dupont et al., 2012). Lactose is usually determined by enzyme method and high performance liquid chromatography (HPLC) technology. The latest research shows that advanced HPLC-MS technology improves the accuracy of determination (Hou et al., 2019). Acidity is usually measured by electropotentiometry and titration, while the total solids content is determined by traditional oven drying and refraction techniques (Murphy et al., 2015). Table 1 lists the direct methods for detecting the nutritional value of milk powder.
Table 1
Table 1. Direct detection methods for determining the nutritional composition.
3.2 Reconstitution performance assessment methods
3.2.1 Wettability
According to ISO/TS 17758:2014|IDF/RM 87:2014, methods for improving the dispersibility and wettability of instant milk powder are described. Common wettability detection methods include contact Angle measurement, settling time and capillary rise method. Figure 1 shows these methods in action.
Figure 1
Figure 1. Illustration of direct detection methods for milk powder wettability.
Contact Angle measurement is to detect the contact Angle between the water drop and the surface of the milk powder by using an goniometer. The contact Angle can be used to roughly judge the wettability of milk powder. The smaller the contact Angle, the better the wettability of the milk powder (Ji et al., 2016a).
The IDF method is as follows: Put 10 grams of skim milk powder or 13 grams of whole milk powder into a 400 mL beaker, and then place the beaker in 25°C water. Next, the already weighed powder is poured into a glass cylinder, which is placed on the glass plate above the beaker. After that, remove the glass plate and start the timer. Wettability, also known as wetting time, refers to the time it takes for the last particle in the powder to pass through the water surface (Warncke and Kulozik, 2020).
The Niro method uses a funnel made of antistatic plastic foil, placed on the side of a beaker, with a glass rod as a stopper, and a water temperature of more than 20°C. Sprinkle 10 grams of powder around the glass rod. When the glass rod is lifted, start timing with a stopwatch (van Boven et al., 2023).
3.2.2 Dispersibility
The main methods for directly detecting the dispersion of milk powder are dissolution method and New Zealand Dairy Board method (NZDB method).
Dissolution detection requires manual operation, and the steps are complicated. The method is to take the right amount of milk powder and water, mix at 25 degrees, stir continuously for 20 s, and then use a 150 micron sieve to filter out the solid particles that are not dispersed. The proportion of solids in the reconstituted milk is then calculated to determine the dispersion of the milk powder. The more solids in the liquid, the better the powder is dispersed.
The NZDB method (Písecký, 2012) simplifies laboratory operations and reduces overall time, but does not solve the real-time measurement problems of the international standard method. According to the NZDB method, 26 grams of milk powder is dissolved in 200 mL of distilled water at 20 degrees Celsius. Stir quickly with a fork to swirl the solution, then pour the mixture into a settling funnel and vacuum it through a 300-micron stainless steel sieve within 5 s. The undissolved milk powder is left on the sieve, dried and weighed, and the dispersion is expressed as a percentage of the undissolved substance.
3.2.3 Sinkability
The main direct detection methods for evaluating the subsidence characteristics of milk powder include subsidence detection, turbidity measurement and density measurement.
Sinkability detection is to dissolve a certain amount of milk powder in water, put it in the water for a period of time to see the situation of the sediment, and then measure the volume or weight of the sediment to judge the characteristics of the sediment.
The solubility of milk powder was measured by turbidimetry. A low turbidity may indicate a high concentration of suspended particles or sediment.
Specific gravity measurement evaluates the settlement by measuring the change in the specific gravity of milk powder in water. Higher specific gravity generally means better dispersion and less likelihood of settlement.
3.2.4 Permission to reuse and copyright
The main basis for evaluating the solubility of milk powder is ISO 8156:2005|IDF 129:2005. In addition to this standard, there are two methods: transparency assessment and turbidity measurement.
The insolubility index method is a basic method used to measure the solubility of milk powder (Lin et al., 2022). Because this method can be repeated and is closely related to actual production applications, it is widely used.
Transparency evaluation is to put the milk powder into water and observe the clarity of the water to determine whether the milk powder can dissolve well. A spectrophotometer measures how much light passes through; The more it penetrates, the better it dissolves.
Turbidity method is the use of turbidity meter to determine the turbidity of dissolved milk powder solution. The lower the turbidity, the better the dissolution effect. Turbidimeters provide stable and repeatable measurement results, making them suitable for quality control in mass production.
The equations should be inserted in editable format from the equation editor.
3.3 Factors affecting transport convenience assessments
3.3.1 Flowability
The measuring methods of fluidity include Angle of rest method, discharge time method and drum method. The Angle of rest detection is to drop the milk powder from a certain height to the horizontal surface, and then measure the diameter of the cone formed, which is the Angle of rest. The smaller the Angle of repose, the better the fluidity. This method is simple to understand, but is easily affected by ambient humidity and temperature. The discharge time detection uses a Hall flow meter to measure the time required for a given amount of milk products to pass through a standard hole. The shorter the time, the better the liquidity.
3.3.2 Bulk density
In accordance with ASTM D7481-18 (Astm, 2018) method for measuring powder loose and vibrational bulk density in a cylinder, we use a powder characteristic tester to measure the loose bulk density of milk powder. First, a small spoon is used to add the milk powder sample through the feed port and pass the sample through a sieve into a special loose density measuring container. When the container is full and slightly overflows, stop filling. Next, use a triangular metal scraper to smooth the surface of the container. Finally, the total weight of the powder and container is recorded, and the loose packaging density is calculated using the corresponding formula.
According to ISO 8967:2005|IDF 134:2005 Standard method for measuring the vibration density of powder products, the vibration densitometer consists of a fixed 100 mL empty cup and an extension cylinder. Put the milk powder into this long container and close the lid. The device is then fixed in a specific position and subjected to 3,000 vibrations. After the vibration is over, use a triangular metal scraper to smooth the surface of the container. The total weight of the powder and container is then recorded, and the corresponding formula is used to calculate the vibration density.
3.4 Sensory characteristic assessment methods
Visual, olfactory, gustatory, tactile and auditory evaluation are the most basic and direct evaluation methods. The sensory evaluation of milk powder is mainly subjective. In order to quantify sensory properties, these methods can be listed in Table 2.
Table 2
Table 2. Sensory evaluation methods.
4 Indirect methods for quality detection of milk powder
In this paper, indirect detection is defined as the measurement of the quality characteristics of milk powder in the laboratory, and then the establishment of links between these quality characteristics and other representative characteristics of the sample, so that only these optional sample characteristics can be used to quickly assess the quality characteristics. Figure 2 illustrates this process.
Figure 2
Figure 2. Logic diagram of direct and indirect detection.
Due to the difficulty of establishing indirect detection correlations and the fact that most methods rely on AI, these methods may not fully cover all aspects of milk powder quality detection. Table 3 lists the main differences and advantages of existing direct and indirect detection methods. This section describes the methods in terms of each specific method, rather than categorizing them by the quality attribute being measured.
Table 3
Table 3. Differences and advantages between direct and indirect detection methods.
4.1 Image-based indirect detection
Microscopic imaging can provide detailed information about milk powder particles. The optical microscope works at a certain magnification and can simultaneously display the details and distribution of different particles (Murrieta-Pazos et al., 2012; Gaiani et al., 2011). As the magnification of the microscope increases, more surface details of milk powder particles can be seen, but some sense of proportion may be lost (Kim et al., 2002). The characteristics of these microscopic images are helpful for indirect quality detection of milk powder (Burgain et al., 2014; Fu et al., 2012). The morphological characteristics observed by microscope are closely related to the quality parameters of milk powder. Recent studies have shown that particle size distribution and surface micro-structure are related to some important quality properties, such as wettability, impermeability, and fluidity. Advanced image analysis techniques can now measure these appearance features, making it an effective tool for rapid and non-damaging sample quality inspection.
Ding et al. (2020a) introduced a method to predict the dispersion degree of milk powder through the image of milk powder particles, which was completed under a tenfold optical microscope. They took pictures of milk powder particles with an optical microscope and then processed these pictures with computer technology. After removing noise and converting them into black and white images, they extracted characteristics such as roundness, convence, solidity, maximum diameter, minimum diameter, area, equivalent diameter, circumference and elongation of milk powder particles. Then, they used principal component analysis (PCA) to study the effects of these characteristics on the dispersity of milk powder, and found that roundness, convexity and area were the main factors affecting the dispersity of milk powder. They then trained the model with partial least square method (PLS) and artificial neural network (ANN), and the results showed that the ANN-based model had better prediction results. In addition, Ding et al. also conducted another study using a similar method to predict the loose and packing density of milk powder by analyzing the image of milk powder particles under a tenfold optical microscope.
The latest study by Ding et al. (2024) proposed a simpler method, which is to put together pictures of milk powder particles taken under a 10x microscope, so that pictures containing the proportion and shape information of milk powder particles can be obtained. They trained a ResNet model using these images and the corresponding dispersion and volume density data as datasets. The trained model can directly predict the dispersion quality and packaging density of milk powder samples from the input pictures.
Harindran et al. (2022) used laser confocal scanning microscopy to observe the dry state of the recombinant emulsion under three conditions of dilution, solidification and storage, so as to predict the situation of milk powder. These pictures show what the reconstituted emulsion looks like at different stages of treatment, and the different states of the milk powder can be analyzed through these pictures. They chose a feedforward neural network (FFN) as the basic model for classification and trained it with representative picture data for the three states. The trained model was able to predict the type of composite emulsion with 85% accuracy and effectively determine whether the sample was diluted, solidified, or deteriorated.
Yan et al. (2020) developed a method for predicting the type of milk powder suspension using a lenless camera. They recorded the transmission speckle image of suspension under coherent He-Ne laser irradiation with a lenceless camera. These images show dispersion and transport patterns related to the concentration of the suspension. The convolutional neural network (CNN) was trained using these images as a dataset, and the trained model was more than 99% accurate on the test set.
4.2 Spectrum-based indirect detection
The spectroscopy can be used to detect milk powder quickly, and can obtain the composition and quality information of milk powder comprehensively. There are different kinds of this technology depending on the wavelength. Near infrared spectroscopy (NIRS) has a working range of 780–2,500 nm and can detect protein, moisture, and fat in milk powder with very few samples (Wu et al., 2008; Hebling et al., 2022). As the frequency increases, more detailed chemical information on milk powder can be obtained by spectroscopy, although the accuracy of this information may be affected by the signal-to-noise ratio (Gastélum-Barrios et al., 2020). The spectral technology can effectively complete the indirect quality detection of milk powder (Velásquez-Ferrín et al., 2021; Du, 2024). By analyzing the spectrogram, we can understand the quality parameters of milk powder, such as the change of milk powder protein, the crystallization of lactose and the stability of reconstructed milk powder through the specific spectral band and its intensity change. Combined with AI technology, this method can achieve rapid and accurate quantitative indirect quality detection of milk powder.
Huang W. et al. (2022) performed a rapid elemental analysis of milk powder using laser-induced breakdown spectroscopy (LIBS). They compared five AI techniques: convolutional neural networks (CNN), linear discriminant Analysis (LDA), K-nearest neighbor (KNN), random forest (RF), and support vector machine (SVM). The average accuracy rate of CNN in detecting the adulteration of different foreign protein milk powder reached 97.8%. This research has developed a simple, reliable and efficient method for the detection of milk powder adulteration.
Near infrared spectroscopy (NIR) is very helpful for quality assessment and adulteration detection of milk powder. Ejeahalaka and On (2020) studied a near infrared spectroscopy method for the detection of melamine and urea in fertilizer milk powder (FMP) containing vegetable oils such as coconut oil, palm oil, soybean oil, and sunflower oil. They used a partial least squares regression model that predicted a sum of squares (PRESS R2) of more than 95% when the amount of adulteration reached 1.00% or more. This method provides an affordable option for the testing of FMP products and helps safeguard public health. Huang G. et al. (2022) developed a fast and simple method for detecting food adulteration through spectral analysis. Their approach is to extract simple features from complex spectral data with an autoencoder and reconstruct the original spectrum. By comparing the errors before and after coding to determine whether the food is adulterated. The effect of this model is consistent with the infrared spectral data of milk powder and its adulterated version. The experimental results show that this classification method is better than the existing method based on support vector machine, and can be flexibly used for spectral detection of food adulteration. Khan et al. (2021) and Khan et al. (2020b) used near infrared spectroscopy, data preprocessing and multivariate analysis to predict the fine particle size fraction, dispersion and packing density of different milk powder samples. They achieved a high prediction accuracy of 88–90% using a partial least squares (PLS) regression model. This provides a new way to detect the characteristics of milk powder, which can be effectively applied in large-scale quality control in factories.
MALDI-TOF mass spectrometry can detect protein in milk very sensitively and is very effective in detecting adulteration of milk powder. Rysova et al. (2022) used this method to detect adulteration in small ruminant milk and developed calibration models for goat and sheep milk. They used PLS regression and Lasso regularized generalized linear models to analyze the concentration of adulterated milk in small ruminant milk. Zhang et al. (2022) developed a method combining MALDI-TOF mass spectrometry and AI for the analysis of the composition and quality of milk and dairy products. They selected the characteristic peptide mass spectrometry signals by PLS-DA, LASSO, and RFECV. The prediction accuracy of SVM-L, RF, LDA and PDA are 0.97, 0.96, 0.96 and 0.96 respectively, which can accurately identify heat treated milk samples.
Nuclear magnetic resonance (NMR) spectroscopy and chromatography-mass spectrometry have become effective tools for the identification and characterization of milk. Rysova et al. (2021) used NMR spectroscopy and multivariate analysis (PCA and OPLS-DA) in the spectral range of 0.04 ppm to find specific biomarkers that could distinguish cow’s and goat’s milk. They found that nacetylated carbohydrates are unique to cow’s milk, and that nacetylglucosamine and nacetylgalactosamine can be used to check whether cow’s milk has been mixed with cow’s milk. Studies have shown a strong positive correlation between the amount of nacetyl carbohydrates and the proportion of milk. The ROC curve was used to verify the model. The results showed that the detection accuracy of 5% adulterated milk could reach 84.78%, and the detection accuracy of 10% adulterated goat milk could reach 95.65%. Sanchez et al. (2021) conducted a comprehensive study to develop a method combined with nuclear magnetic resonance (NMR) and gas chromatography–mass spectrometry (GC–MS) for characterizing the liquid and volatile components of New Zealand GMP (sugar macropeptide). Their non-targeted 1H-NMR analysis identified 44 metabolites from the fat, sugar, and aromatic regions of the liquid portion. In addition, their nontargeted HS-SPME-GC–MS method isolated 50 volatile compounds from different chemical families. These volatiles include 26 aliphatic hydrocarbons (mainly alkanes), 8 ketones, 4 alcohols, 3 aromatics, 3 olefins, 2 aldehydes, 2 esters, 1 acid and 1 sulfur compound. This study demonstrates the power of NMR based metabolomics in characterizing the water-soluble components of GMP.
4.3 Environmental parameter-based indirect detection
Production environment monitoring provides important information for the quality of milk powder. In the production of milk powder, the process control system operated by multiple sensors can simultaneously collect real-time environmental parameters and production line conditions (Vieira et al., 2019; Munir et al., 2015). Although the extended monitoring scope may affect some operational flexibility, we can obtain more comprehensive information about the production batch environment (Wang et al., 2018). These environmental monitoring data can be used for preliminary quality checks of milk powder (O’Callaghan and Cunningham, 2005; Uthayaseelan, 2024). The environmental parameters obtained by continuous monitoring are closely related to the quality characteristics of milk powder. The latest study found that temperature changes, humidity and air flow patterns in particular reflect important quality characteristics of milk powder, such as moisture content, particle aggregation and microbial stability. Advanced data analysis methods can predict quality parameters based on environmental conditions, making it a useful tool for proactive quality control and batch consistency assessment.
Boiarkina et al. (2017) devised a test to assess how poorly powdered milk is absorbed during production. This method is easy to operate and solves some problems of the previous detection methods. It is only necessary to collect some data on the production line and information about the size of milk powder particles, and you can roughly judge the dispersion of milk powder. However, this test can only roughly classify dispersion into high, medium and low levels, and cannot perform specific numerical analysis. This method provides a fast and effective inspection method for factories.
Wagh and Agashe (2023) used machine learning and data analysis techniques to design an analytical system to monitor changes in product density in a four-effect falling film evaporator for milk. This system handles different process parameters collected by the evaporator, such as temperature, pressure, and flow. After feature extraction and data preprocessing, a density prediction model is established, which can realize real-time monitoring and optimization of production process. This research focuses on the development of models to predict product density by using data collected from multiple sensors and machine learning methods to improve the quality of production and products.
4.4 E-nose-based indirect detection
Electronic nose technology provides characteristic information about the volatile characteristics of milk powder. Gas sensor arrays work with multiple sensing elements to simultaneously capture overall aroma patterns and specific volatile compound signals during milk powder analysis (Yu et al., 2008; Yakubu et al., 2022). With the increase in optional specialized sensor categories, the detection methods for specific volatile compounds in milk powder have also correspondingly increased (Darvishi et al., 2024). The signals generated by these electronic noses can be effectively applied to the indirect detection tasks of milk powder, as the volatile profiles produced by the electronic noses have a strong correlation with various quality parameters in milk powder (Mu et al., 2020). The odor characteristics and specific volatile markers in milk powder can reflect the oxidative state, storage stability, and quality attributes related to thermal damage. The use of electronic nose technology combined with AI pattern recognition enables non-destructive testing of milk powder.
Tian et al. (2023) combined electronic nose technology with gas chromatography to propose a rapid detection method for milk adulteration from multiple perspectives. The process of this method includes sample preparation, data preprocessing, feature extraction and selection, model construction and model validation. In the data preprocessing phase, we use techniques such as normalization and noise reduction to ensure data stability and quality. In the stage of feature selection and extraction, we use statistical and machine learning methods to select important features from the preprocessed d`ata and build a predictive model. In the model building stage, we chose Lasso regression, decision tree and random forest to train the training set, so as to build the prediction model. Finally, in the model validation stage, we verify the trained model on the test set through cross-validation, mean square error, determination coefficient and other evaluation indicators. Experimental results show that this analytical framework can effectively predict milk adulteration.
Chi et al. (2021) focus on the research of process control in food processing using electronic noses. They proposed a method for monitoring product quality in food processing that utilizes electronic noses as sensors, based on data analysis and machine learning. This method trains machine learning predictive models using processed electronic nose data, effectively monitors the food production process, optimizes production workflows, and ensures product consistency to achieve cost reduction and efficiency improvement.
Shuba et al. (2024) proposed an innovative sensor system and data processing framework to improve the monitoring efficiency of multiple variables in complex environments. By integrating e-nose data with machine learning algorithms, the system can accurately identify changes in key variables, providing real-time monitoring and prediction capabilities. The experimental results show that the system is significantly superior to the traditional methods in terms of the detection accuracy and response speed and effectively improves the operational efficiency and safety in industrial applications.
4.5 Water activity-based indirect detection
Water activity monitoring is the key index of milk powder quality analysis and shelflife prediction. Advanced sensor systems are integrated with the production line to continuously measure water activity and other processing parameters during milk powder production (Olaimat et al., 2020). As the measurement accuracy improves, a more detailed profile of the water activity in production batches can be obtained, although some operational complexity may increase (Juarez-Enriquez et al., 2022). These continuous monitoring data provide valuable insights for the rapid quality assessment of milk powder. Water activity measurements obtained through real-time monitoring showed that there was a strong correlation between various quality attributes of milk powder. Recent studies have shown that water activity patterns, especially during storage and processing, are highly indicative of key quality characteristics, such as powder stability, microbial safety and reconfiguration characteristics.
Wei et al. (2020) conducted a comprehensive study on the influence of water activity on the mechanics of Salmonella heatkilling activity in milk powder. Their study found a relationship between water activity and bacterial heat resistance, suggesting that the heat resistance of salmonella increased as water activity decreased. The study also pointed out that the different fat content of whole milk powder and skim milk powder had no significant effect on the heat resistance of Salmonella. The researchers created a model to connect the water activity in milk powder to the inactivation of salmonella, which could help the food industry predict the reference D value of salmonella under certain temperature conditions, that is, the time it takes to achieve bactericidal effect.
Yan et al. (2022) studied the relationship among water activity, lactose crystallization (LC) and the Maillard reaction (MR) in the simulated milk powder system. Their research showed that increasing the water activity (from 0.11 to 0.69) accelerated LC and MR processes across all model systems. Whey protein-lactose (WP-Lac) model showed the greatest change, the glass transition temperature decreased rapidly, and the structural change was also obvious. This is because whey protein has a simple structure and small molecular weight, which can also explain the phenomenon of high lysine consumption. As water activity increases, lactose forms a tighter crystal structure, and in the WP-Lac model, anhydrous beta-lactose crystals exhibit greater proteinbinding capacity and more water release, resulting in accelerated glycosylation and Browning. In contrast, casein-based models show fewer molecular changes and reduced lactose crystallization, while in casein-whey protein-lactogen models, interactions between proteins inhibit the production of glycosylation products.
5 Application of AI in milk powder quality detection
The technology of milk powder quality detection has changed a lot, from the traditional detection method to the modern analysis method (Ma et al., 2024c). In particular, with the rapid development of AI, its new application in milk powder quality detection is bringing new models for quality assurance (Ma et al., 2024a). Figure 3 shows the application of AI in milk powder detection. This chapter mainly introduces the latest research and future possibilities of AI technology in this field.
Figure 3
Figure 3. Methods of applying artificial intelligence technology in milk powder detection.
5.1 Application of regression models
Regression analysis is a statistical method used to determine the quantitative relationship between variables. The main idea is to build a predictive model through data fitting. Although the traditional ordinary least squares (OLS) regression method is widely used, it is not effective in dealing with multicollinearity problems. Partial least squares (PLS) (Stocchero et al., 2021) regression addresses this challenge by extracting the underlying factors from the original variables and performing dimensionality reduction in the predictor variables and response space. This kind of application has been widely used in some fields, such as chemometrics. Support vector machine (SVM) regression takes a completely different perspective, rooted in the structural risk minimization principles of statistical learning theory. By introducing kernel functions to map low-dimensional features to high-dimensional spaces, SVM effectively deals with nonlinear regression problems. Kernel-based regression models, such as kernel ridge regression (KRR) (H. Zhang et al., 2023), solve nonlinear problems by identifying linear relationships in high-dimensional feature spaces. Each approach has its advantages: PLS is good at managing multi-collinear data, SVM is robust to outliers and exhibits strong generalization capabilities and KRR offers computational efficiency advantages.
At present, the quantitative methods used in milk powder quality detection mainly involve regression models, such as PLS and SVM. For example, Ding et al. used PCA and PLS for the rapid, indirect quantification of the dispersibility and bulk density of milk powder. Ejeahalaka et al. used PLS to detect whether the fat in milk pulp was adulterated. Similarly, PLS was also used by Khan et al. to analyze the fine particle size fraction, dispersion and packing density of milk powder. In addition, Tian et al. used linear regression, decision tree and random forest methods to quickly detect whether milk powder was adulterated. These regression models can predict difficult quality parameters from easily measured data, which greatly improves the efficiency of indirect detection of milk powder quality.
5.2 Application of computer vision
Computer vision technology is a deep learning method specifically used to process image data. Its basic principle is to analyze image features by imitating the layers of the visual system. The design of convolutional neural networks (CNNS) was influenced by the work of biologists Hubel and Wiesel (Hubel and Wiesel, 1962), who investigated the layers of information processing in the cat visual cortex. This layered design allows the deeper layers of the network to capture features that are relevant to a larger range of input images. In a neural network, this means that deeper nodes can see a larger range. This attention-based network model mimics the way the human brain processes visual information. The human brain scans roughly the entire field of vision and then quickly focuses on a specific area (Luo et al., 2016). This helps people quickly filter out unimportant information and focus on the key parts. Compared to other methods, the attention mechanism can extend the scope of observation better, cover the entire image, and pay more attention to the important parts (Itti et al., 1998).
Computer vision technology is widely used in milk powder quality detection (Ma et al., 2024b). For example, Ding et al. used ResNet model (A deep convolutional neural network architecture utilizing residual blocks with skip connections to mitigate vanishing gradients, enabling training of extremely deep networks and achieving breakthroughs in image recognition tasks) to predict the dispersion and packaging density of milk powder. Yan et al. trained the CNN model by transmitting the speckle model to distinguish different types of suspended matter. Huang et al. used the CNN model trained by laser induced breakdown spectra to analyze the element composition in milk powder. Compared with regression model, computer vision technology makes data analysis easier and can further improve the efficiency of milk powder quality detection. However, the application of computer vision technology in this field is still relatively small, and many advanced algorithms have not been developed. Because the image data related to milk powder is very diverse, the computer vision technology has great application prospects in the future.
5.3 Prospects for the application of recommendation systems and large language models
The rapid development of AI technology provides new opportunities for the quality inspection of milk powder. A large language model (LLM) built on the Transformer architecture (Vaswani, 2017) (a neural network model using self-attention mechanisms to process sequential data in parallel, with stacked encoder-decoder layers and positional encodings for context awareness) can analyze the historical test data, expertise and scientific literature to develop high quality early warning systems. These systems leverage multi-modal data integration, combining sensory assessments, physicochemical parameters and expert insights to provide more comprehensive quality assessment recommendations. The recommendation system technology rooted in collaborative filtering algorithms can operate on the configuration of test parameters (Koren et al., 2009). By analyzing the detection history of different product batches, these systems can automatically recommend the most effective and reliable detection regimen for a particular type of milk powder, thereby increasing the efficiency of detection and the reliability of results. Graph neural networks (GNNs) (Veličković, 2023) implement supply chain traceability systems by building diagrams that connect raw materials, production processes and test phases. This helps monitor quality from start to finish and ensures a transparent and reliable quality control system. The combination of advanced technologies such as LLM for intelligent analysis, recommendation system for process optimization and GNN for supply chain monitoring will promote the progress of milk powder quality detection to a more intelligent and accurate direction, and provide stronger technical support for the safety of dairy products and consumers’ trust.
6 Outlook on milk powder quality detection methods
In recent years, the quality detection technology of milk powder has improved greatly, both direct analysis and indirect detection, which use the technology of the time. With the continuous progress of science and technology, some new technologies have not been applied to the detection of milk powder, but the existing methods are also continuing to improve, which makes people full of expectations for future quality detection. Traditional detection methods are still the basis, while the application of new technologies makes the quality assessment of milk powder more accurate and efficient (Sokoliuk et al., 2022). At present, the combination of traditional methods and new technologies provides a new way of quality inspection, which requires us to seriously think about the future development path. This chapter will discuss the potential and possible application of new technologies in milk powder quality detection, focusing on new methods and technological innovation (Bowler et al., 2020).
6.1 Traceability and automation
Intelligent automation is the key development direction of milk powder quality detection in the future. The integration of big data analysis and automated systems will improve the efficiency, safety and reliability of milk powder quality inspection. Future automated systems will go beyond a single test phase to cover the entire quality inspection workflow. From automated sampling on the production line to automated analysis processes to automated data processing and feedback mechanisms, these systems will significantly improve test efficiency and accuracy (Jiang et al., 2022). For example, the combination of a robotic sampling arm and high-throughput detection equipment can significantly improve the detection flux and accuracy of milk powder analysis (Weston et al., 2021).
Internet of Things (IoT) technology will make milk powder quality monitoring smarter and more transparent throughout the supply chain. Smart sensors embedded in milk powder packaging can monitor key parameters, such as temperature and humidity, in real time during production, storage and transportation, ensuring quality compliance at every stage (Younan et al., 2020). The implementation of blockchain technology can further ensure the security and immutability of data, achieve full traceability from production to consumption and enhance consumer confidence in milk powder products (Rejeb et al., 2020).
6.2 Interdisciplinary integration
The future development of milk powder quality detection will not only depend on the progress of a single discipline but also depend on the integration and integration of multiple technical fields. The integration of knowledge of chemistry, physics, biology and information technology will make a more comprehensive, accurate and efficient milk powder quality detection method possible (Farag et al., 2022). Interdisciplinary cooperation is becoming a key trend in milk powder quality control. For example, near infrared spectroscopy combined with immune sensors can detect trace contaminants, such as antibiotic residues and aflatoxins, in milk powder with high accuracy (Liang et al., 2020). Microfluidic technology from physics, when integrated with biosensors, enables a fast and accurate analysis with minimal sample consumption, making it particularly suitable for the point-of-demand detection of milk powder quality and safety events (Farag et al., 2021). In addition, new nanomaterials play a crucial role in milk powder quality detection (Ma et al., 2024b). For example, graphene-based biosensors can detect ultra-low concentrations of harmful substances in milk powder, including heavy metal ions and organic pollutants (Seth and Rathinasabapathi, 2022). This combination of technologies enables the optimization of the entire process, enabling efficient management of every link from production to quality control to logistics on a single platform. An important example of interdisciplinary collaboration is the use of cold plasma technology in milk processing. This method is different from traditional chemical and heat treatment methods, and shows good bactericidal effect on whole, semi-skimmed and skimmed milk, while not significantly changing important indicators such as color and pH value. Although plasma sterilization technology is becoming increasingly popular, the application in milk purification, especially using low-temperature plasmas, demonstrates new areas of plasma physics and dairy science. This suggests that an interdisciplinary approach has the potential to transform the traditional way dairy products are processed (Ucar et al., 2021). Cooperation between different disciplines is increasing. Future research will not only advance in a single field, but also foster collaboration among experts in chemistry, physics, and data science (Amiri et al., 2021). Through indepth multidisciplinary integration, more intelligent and comprehensive detection methods can be developed to improve the quality detection level of milk powder.
6.3 Environmental protection and sustainability
In the traditional milk powder quality detection, the traditional chemical methods often rely heavily on toxic chemical reagents and consumables. Not only do these methods pose potential health risks and negative environmental impacts for operators, but there are also resource inefficiencies and waste generation issues (Alizadeh et al., 2021). For example, Kjeldahl nitrogen determination for protein content uses large amounts of sulfuric acid and selenium compounds as catalysts, which release toxic fumes during heating (Hueso et al., 2022). Soxhlet extraction relies on organic solvents (such as ether or chloromethane) to determine fat content (Masotti et al., 2020) and involves volatile and flammable substances that pose environmental hazards and need to be properly disposed of after analysis (Câmara et al., 2022). In addition, these solvent-based extraction methods typically require longer extraction times and high energy consumption (Mao et al., 2024).
However, people are increasingly concerned about environmental protection and sustainable development, and many modern detection techniques are slowly replacing traditional methods. For example, non-destructive techniques such as spectroscopic analysis (NIR, FTIR) can quickly determine the composition without the use of chemical reagents (Ziyaina et al., 2020). Biosensors can be used for specific detection with little waste generation (Girigoswami et al., 2021). New technologies such as ultrasound-assisted extraction and supercritical fluid extraction offer a more environmentally friendly option for composition analysis (Nde and Foncha, 2020). Microfluidic devices enable accurate analysis while significantly reducing reagent usage (Neethirajan et al., 2011). Advanced membrane separation technology provides an efficient and environmentally friendly sample handling method (Du et al., 2022). These emerging technologies, along with assistive tools such as AI algorithms, IoT monitoring and blockchain traceability, are pushing milk powder quality inspection in a more sustainable direction. These environmental technologies not only improve the safety of testing, reduce environ-mental pollution, but also improve the efficiency of resource use, in line with the global sustainable development goals.
6.4 Prospects of more future technologies
Looking to the future, the integration of new technologies provides new opportunities for milk powder quality detection. The development of quantum sensing technology can significantly improve the accuracy of inspection, and advanced nanomaterials and smart sensors can allow real-time online monitoring systems to better integrate into intelligent production processes. In addition, advances in edge computing and 5G/6G technology will make it possible to establish a distributed test network, enabling instant data sharing and analysis across the global supply chain, and ensuring consistent quality standards worldwide. In addition, the use of synthetic biology and biosensing platforms can provide highly specific and sensitive detection methods for trace substances and contaminants, while advances in microfluidic technology may lead to lab-on-a-chip solutions that require a small number of samples to complete a full quality assessment. These technologies will further promote the popularization of the application of non-destructive full inspection, leaving no potential problems on each milk powder production line unnoticed.
7 Conclusion
This paper summarizes the key indicators of milk powder quality detection, the existing direct detection methods and the emerging indirect methods. It discusses in detail the application of AI in indirect detection. Milk powder quality is evaluated on four dimensions: nutritional content, reconstitution, ease of transport and organoleptic properties. Direct detection methods, including traditional chemical analysis and modern instrumentation techniques, provide precise measurements but require complex sample preparation and lengthy processing. In contrast, indirect methods such as microscopy, spectroscopic analysis, and electronic nose techniques can enable rapid, non-destructive detection, but often require complex calibration models. The integration of AI enables indirect detection with greater accuracy and efficiency while eliminating human-induced bias, effectively bridging the performance gap between direct and indirect methods.
According to current research, the quality detection of milk powder has been developed in three main aspects. First, in terms of intelligence and automation, with the progress of AI technology, it will be more applied to milk powder detection, and with the development of intelligent detection technology, more open data sets will be provided. More advanced algorithms will make quality assessment faster and more accurate. New technologies such as the Internet of Things and blockchain will further improve the quality inspection system, ensure the implementation of inspection responsibilities, and promote the evaluation of test results. Secondly, in the field of interdisciplinary fusion, the future indirect detection methods will utilize more advanced equipment for extracting the physical and chemical characteristics of milk powder and obtain richer spectral information through improved imaging equipment and advanced spectral technology, so as to obtain enhanced image features. Future feature extraction will go beyond current imaging and spectroscopy methods, as developments in materials science, acoustics and other technologies will bring a whole new perspective to milk powder detection. Finally, in terms of green and sustainable development, although some traditional methods have become the standard process for testing relevant parameters, because these methods rely on toxic chemicals and consume a lot of resources, people are looking for more green and sustainable alternatives.
Author contributions
XS: Funding acquisition, Supervision, Writing – original draft. SS: Writing – original draft, Writing – review & editing. GD: Formal analysis, Writing – review & editing. HD: Funding acquisition, Writing – original draft, Writing – review & editing. ZX: Writing – review & editing. LW: Investigation, Writing – original draft. WC: Formal analysis, Writing – review & editing.
Funding
The author(s) declare that financial support was received for the research and/or publication of this article. This research was funded by the National Key Research and Development Program of China (2024YFE0199500).
Conflict of interest
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
Generative AI statement
The author(s) declare that no Gen AI was used in the creation of this manuscript.
Publisher’s note
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.
References
Alizadeh, M., Adel, M. M., Shakooie, M., Khajavi, M. Z., and Farhoodi, M. (2021). Trends and applications of intelligent packaging in dairy products: a review. Crit. Rev. Food Sci. Nutr. 62, 383–397. doi: 10.1080/10408398.2020.1817847
Amiri, S., Aghamirzaei, M., Mostashari, P., Sarbazi, M., Tizchang, S., and Madahi, H. (2021). “The impact of biotechnology on dairy industry” in Microbial biotechnology in food and health. ed. R. C. Ray (Amsterdam, Netherlands: Elsevier), 53–79.
Andrade, E. H., Paz, L. M., da Fonseca, M., de Souza, R., de Andrade Morais Penna, C. F., Cerqueira, M. M. O. P., et al. (2022). Fat content in fermented milk beverages: determination by the Gerber method. Semina Ciências Agrárias 43, 441–448. doi: 10.5433/1679-0359.2022v43n1p441
Astm, D. (2018). 7481–18. Standard test methods for determining loose and tapped bulk densities of powders using a graduated cylinder. PA, USA: ASTM International West Conshohocken.
Boiarkina, I., Depree, N., Yu, W., and Di Wilson, B. R. Y. (2017). Rapid particle size measurements used as a proxy to control instant whole milk powder dispersibility. Dairy Sci. Technol. 96, 777–786. doi: 10.1007/s13594-016-0302-5
Bowler, A. L., Bakalis, S., and Watson, N. J. (2020). A review of in-line and on-line measurement techniques to monitor industrial mixing processes. Chem. Eng. Res. Des. 153, 463–495. doi: 10.1016/j.cherd.2019.10.045
Burgain, J., Scher, J., Lebeer, S., Vanderleyden, J., Cailliez-Grimal, C., Corgneau, M., et al. (2014). Significance of bacterial surface molecules interactions with milk proteins to enhance microencapsulation of Lactobacillus rhamnosus GG. Food Hydrocoll. 41, 60–70. doi: 10.1016/j.foodhyd.2014.03.029
Câmara, J. S., Perestrelo, R., Berenguer, C. V., Andrade, C. F. P., Gomes, T. M., Olayanju, B., et al. (2022). Green extraction techniques as advanced sample preparation approaches in biological, food, and environmental matrices: a review. Molecules 27:2953. doi: 10.3390/molecules27092953
Chen, B., and Ellefson, W. C. (2024). “Fat Analysis” in Nielsen's food analysis. eds. B. Pam Ismail and S. Suzanne Nielsen (Cham, Switzerland: Springer), 273–286.
Chi, X., Shao, Y., Pan, M., Yang, Q., Yang, Y., Zhang, X., et al. (2021). Distinction of volatile flavor profiles in various skim milk products via HS-SPME–GC–MS and E-nose. Eur. Food Res. Technol. 247, 1539–1551. doi: 10.1007/s00217-021-03730-0
Coolbear, T., Janin, N., Traill, R., and Shingleton, R. (2022). Heat-induced changes in the sensory properties of milk. Int. Dairy J. 126:105199. doi: 10.1016/j.idairyj.2021.105199
Crowley, S. V., Desautel, B., Gazi, I., Kelly, A. L., Huppertz, T., and O’Mahony, J. A. (2015). Rehydration characteristics of milk protein concentrate powders. J. Food Eng. 149, 105–113. doi: 10.1016/j.jfoodeng.2014.09.033
Czarnowska-Kujawska, M., Starowicz, M., Paszczyk, B., Klepacka, J., Popielarczyk, M., and Tońska, E. (2024). The chemical, antioxidant and sensorial properties of milk and plant based kombucha analogues. LWT 206:116610. doi: 10.1016/j.lwt.2024.116610
Darvishi, P., Mirzaee-Ghaleh, E., Ramedani, Z., Karami, H., and Wilson, A. D. (2024). Detecting whey adulteration of powdered Milk by analysis of volatile emissions using a MOS electronic nose. Int. Dairy J. 157:106012. doi: 10.1016/j.idairyj.2024.106012
de Oliveira, N., Leandra, N., and Leal de Oliveira, M. A. (2021). Assessment of enzymatic hydrolysis of lactose in lactose-free milk production-a comparative study using capillary zone electrophoresis and cryoscopy. LWT 138:110585. doi: 10.1016/j.lwt.2020.110585
Ding, H., Li, B., Boiarkina, I., Wilson, D. I., Wei, Y., and Young, B. R. (2020a). Effects of morphology on the bulk density of instant whole milk powder. Food Secur. 9:1024. doi: 10.3390/foods9081024
Ding, H., Shen, S., Xie, Z., Cui, X., and Wang, Z. (2024). "detection of Dispersibility and bulk density of instant whole Milk powder based on residual network." food. Science 45, 9–18. doi: 10.7506/spkx1002-6630-20240129-262
Ding, H., Wei, Y., Boiarkina, I., Depree, N., and Young, B. R. (2020b). Effects of morphology on the dispersibility of instant whole milk powder. J. Food Eng. 276:109841. doi: 10.1016/j.jfoodeng.2019.109841
Ding, Y., Zhao, M., Shu, Y., Ao, H., Chen, J., Chen, W., et al. (2023). Energy value measurement of milk powder using laser-induced breakdown spectroscopy (LIBS) combined with long short-term memory (LSTM). Anal. Methods 15, 4684–4691. doi: 10.1039/D3AY01144E
Du, L. (2024). New insights into raw milk adulterated with milk powder identification: ATR-FTIR spectroscopic fingerprints combined with machine learning and feature selection approaches. J. Food Compos. Anal. 133:106443. doi: 10.1016/j.jfca.2024.106443
Du, Y., Pramanik, B. K., Zhang, Y., Dumée, L., and Jegatheesan, V. (2022). Recent advances in the theory and application of nanofiltration: a review. Curr. Pollut. Rep. 8, 51–80. doi: 10.1007/s40726-021-00208-1
Dupont, D., Croguennec, T., Brodkorb, A., and Kouaouci, R. (2012). “Quantitation of proteins in milk and milk products” in Advanced dairy chemistry: Volume 1A: Proteins: Basic aspects. 4th ed. eds. P. L. H. McSweeney and P. F. Fox (New York, USA: Springer), 87–134.
Ejeahalaka, K. K., and On, S. L. W. (2020). Effective detection and quantification of chemical adulterants in model fat-filled milk powders using NIRS and hierarchical modelling strategies. Food Chem. 309:125785. doi: 10.1016/j.foodchem.2019.125785
Farag, M. A., Khalifa, I., Gamal, M., and Bakry, I. A. (2022). The chemical composition, production technology, authentication, and QC analysis of dried milk. Int. Dairy J. 133:105407. doi: 10.1016/j.idairyj.2022.105407
Farag, M. A., Tanios, M., AlKarimy, S., Ibrahim, H., and Guirguis, H. A. (2021). Biosensing approaches to detect potential milk contaminants: a comprehensive review. Food Addit. Contamin. Part A 38, 1169–1192. doi: 10.1080/19440049.2021.1914864
Fitzpatrick, J. J., Iqbal, T., Delaney, C., Twomey, T., and Keogh, M. K. (2004). Effect of powder properties and storage conditions on the flowability of milk powders with different fat contents. J. Food Eng. 64, 435–444. doi: 10.1016/j.jfoodeng.2003.11.011
Fournaise, T., Burgain, J., Perroud, C., Scher, J., Gaiani, C., and Petit, J. (2020). Impact of formulation on reconstitution and flowability of spray-dried milk powders. Powder Technol. 372, 107–116. doi: 10.1016/j.powtec.2020.05.085
Fu, X., Huck, D., Makein, L., Armstrong, B., Willen, U., and Freeman, T. (2012). Effect of particle shape and size on flow properties of lactose powders. Particuology 10, 203–208. doi: 10.1016/j.partic.2011.11.003
Gaiani, C., Boyanova, P., Hussain, R., Murrieta Pazos, I., Karam, M. C., Burgain, J., et al. (2011). Morphological descriptors and colour as a tool to better understand rehydration properties of dairy powders. Int. Dairy J. 21, 462–469. doi: 10.1016/j.idairyj.2011.02.009
Gaiani, C., Schuck, P., Scher, J., Desobry, S., and Banon, S. (2007). Dairy powder rehydration: influence of protein state, incorporation mode, and agglomeration. J. Dairy Sci. 90, 570–581. doi: 10.3168/jds.S0022-0302(07)71540-0
Gallier, S., Acton, D., Garg, M., and Singh, H. (2017). Natural and processed milk and oil body emulsions: bioavailability, bioaccessibility and functionality. Food Struct. 13, 13–23. doi: 10.1016/j.foostr.2016.07.005
Gastélum-Barrios, A., Soto-Zarazúa, G. M., Escamilla-García, A., Toledano-Ayala, M., Macías-Bobadilla, G., and Jauregui-Vazquez, D. (2020). Optical methods based on ultraviolet, visible, and near-infrared spectra to estimate fat and protein in raw milk: a review. Sensors 20:3356. doi: 10.3390/s20123356
Girigoswami, A., Ghosh, M. M., Pallavi, P., Ramesh, S., and Girigoswami, K. (2021). Nanotechnology in detection of food toxins–focus on the dairy products. Biointer. Res. Appl. Chem. 11, 14155–14172. doi: 10.33263/BRIAC116.1415514172
Guo, K., Zhang, Y., Zhang, H., Wang, L., Song, H., and Li, Z. (2023). Odor changes in breast milk during different storage temperatures and times using GC× GC-O-MS. Food Res. Int. 168:112792. doi: 10.1016/j.foodres.2023.112792
Hailu, Y., Maidannyk, V. A., Murphy, E. G., and McCarthy, N. A. (2023). Improving the physical and wettability properties of skim milk powders through agglomeration and lecithination. J. Food Eng. 357:111597. doi: 10.1016/j.jfoodeng.2023.111597
Harindran, A., Hashmi, S., and Madhurima, V. (2022). Pattern formation of dried droplets of milk during different processes and classifying them using artificial neural networks. J. Dispers. Sci. Technol. 43, 1838–1847. doi: 10.1080/01932691.2021.1880927
Hebling, T., Pedro, J., da Silva Medeiros, M. L., and Barbin, D. F. (2022). Near-infrared techniques for fraud detection in dairy products: a review. J. Food Sci. 87, 1943–1960. doi: 10.1111/1750-3841.16143
Hou, X., Ma, J., Chen, L., He, X., and Wang, S. (2019). “Analysis of additives in Milk powders with SPE-HPLC or 2D-HPLC method” in Milk production, processing and marketing. ed. K. Javed (London, UK: IntechOpen).
Huang, W., Guo, L., Kou, W., Zhang, D., Zhenlin, H., Chen, F., et al. (2022). Identification of adulterated milk powder based on convolutional neural network and laser-induced breakdown spectroscopy. Microchem. J. 176:107190. doi: 10.1016/j.microc.2022.107190
Huang, G., Yuan, L.-m., Shi, W., Chen, X., and Chen, X. (2022). Using one-class autoencoder for adulteration detection of milk powder by infrared spectrum. Food Chem. 372:131219. doi: 10.1016/j.foodchem.2021.131219
Hubel, D. H., and Wiesel, T. N. (1962). Receptive fields, binocular interaction and functional architecture in the cat's visual cortex. J. Physiol. 160, 106–154. doi: 10.1113/jphysiol.1962.sp006837
Hueso, D., Fontecha, J., and Gómez-Cortés, P. (2022). Comparative study of the most commonly used methods for total protein determination in milk of different species and their ultrafiltration products. Front. Nutr. 9:925565. doi: 10.3389/fnut.2022.925565
Hui, C., Xing-xin, G., Jia-li, P., Li-ping, T., and Hai-zi, W. (2022). Determination of nine mineral elements in infant formula milk powder by rapid wet digestion-inductively coupled plasma optical emission spectrosmetry. Food Mach. 38, 76–79. doi: 10.13652/j.spjx.1003.5788.2022.90162
Itti, L., Koch, C., and Niebur, E. (1998). A model of saliency-based visual attention for rapid scene analysis. IEEE Trans. Pattern Anal. Mach. Intell. 20, 1254–1259. doi: 10.1109/34.730558
Jaudzems, G., Fengxia Zhang, W., Bolong, L. B., and Xiao, J. (2019). Chloride in milk, milk powder, whey powder, infant formula, and adult nutritionals potentiometric titration: collaborative study, final action 2016.03. J. AOAC Int. 102, 564–569. doi: 10.5740/jaoacint.18-0244
Ji, J., Fitzpatrick, J., Cronin, K., Crean, A., and Miao, S. (2016a). Assessment of measurement characteristics for rehydration of milk protein based powders. Food Hydrocoll. 54, 151–161. doi: 10.1016/j.foodhyd.2015.09.027
Ji, J., Fitzpatrick, J., Cronin, K., Maguire, P., Zhang, H., and Miao, S. (2016b). Rehydration behaviours of high protein dairy powders: the influence of agglomeration on wettability, dispersibility and solubility. Food Hydrocoll. 58, 194–203. doi: 10.1016/j.foodhyd.2016.02.030
Jiang, J., Ma, X., Ouyang, D., and Williams III, R. O. (2022). Emerging artificial intelligence (AI) technologies used in the development of solid dosage forms. Pharmaceutics 14:2257. doi: 10.3390/pharmaceutics14112257
Johnson, R. G., and King, B. S. L. (1987). Energy-dispersive X-ray fluorescence spectrometry. U.S. Geol. Surv. Bull. 1770, F1–F5.
Juarez-Enriquez, E., Olivas, G. I., Zamudio-Flores, P. B., Perez-Vega, S., Salmeron, I., Ortega-Rivas, E., et al. (2022). A review on the influence of water on food powder flowability. J. Food Process Eng. 45:e14031. doi: 10.1111/jfpe.14031
Junior, A., José, E., de Jesus, P. C., and Borges, E. M. (2023). Whey protein analysis using the Lowry assay and 96-well-plate digital images acquired using smartphones. J. Chem. Educ. 100, 2329–2338. doi: 10.1021/acs.jchemed.2c00830
Khan, A., Munir, M. T., Wei, Y., and Young, B. (2020b). Wavelength selection for rapid identification of different particle size fractions of milk powder using hyperspectral imaging. Sensors 20:4645. doi: 10.3390/s20164645
Khan, A., Munir, M. T., Wei, Y., and Young, B. R. (2021). Near-infrared spectroscopy and data analysis for predicting milk powder quality attributes. Int. J. Dairy Technol. 74, 235–245. doi: 10.1111/1471-0307.12734
Khan, A., Munir, M. T., Yu, W., and Young, B. R. (2020a). A review towards hyperspectral imaging for real-time quality control of food products with an illustrative case study of Milk powder production. Food Bioprocess Technol. 13, 739–752. doi: 10.1007/s11947-020-02433-w
Kim, E. H.-J., Dong Chen, X., and Pearce, D. (2002). Surface characterization of four industrial spray-dried dairy powders in relation to chemical composition, structure and wetting property. Colloids Surf. B: Biointerfaces 26, 197–212. doi: 10.1016/S0927-7765(01)00334-4
Koren, Y., Bell, R., and Volinsky, C. (2009). Matrix factorization techniques for recommender systems. Computer 42, 30–37. doi: 10.1109/MC.2009.263
Kučerová, P., Komenská, P., Tomková, H., Skopalová, J., and Barták, P. (2017). Determination of lactose in milk products: a comparison of three-enzyme amperometric biosensor and gas chromatography/tandem mass spectrometry. Monatshefte für Chemie-Chemical Monthly 148, 517–524. doi: 10.1007/s00706-016-1903-7
Li, B.-Y., Mahe, J.-L., Hao, J.-Y., Ye, W.-H., Bai, X.-F., Feng, H.-T., et al. (2023). Formula Milk supplementation and bone acquisition in 4-6 years Chinese children: a 12-month cluster-randomized controlled trial. Nutrients 15:2012. doi: 10.3390/nu15082012
Liang, W., Wei, Y., Gao, M., Yan, X., Zhu, X., and Guo, W. (2020). Detection of melamine adulteration in milk powder by using optical spectroscopy technologies in the last decade—a review. Food Anal. Methods 13, 2059–2069. doi: 10.1007/s12161-020-01822-3
Liang, X., Yang, H., Sun, J., Cheng, J., Luo, X., Wang, Z., et al. (2021). Effects of enzymatic treatments on the hydrolysis and antigenicity reduction of natural cow milk. Food Sci. Nutr. 9, 985–993. doi: 10.1002/fsn3.2066
Lin, R., Wang, Y., and Selomulya, C. (2022). Physical properties of dairy powders. Encyclop. Dairy Sci. eds. P. L. H. McSweeney and J. P. McNamara, Amsterdam, Netherlands. 504–520. doi: 10.1016/B978-0-12-818766-1.00028-3
Liu, S., Jiang, X., Feifan, F., Aziz, T., Li, G., Zhao, J., et al. (2023a). Assessing the effect of powder characteristics of infant milk on the compressibility of milk powder compression molding. Food Sci. Nutr. 11, 4625–4633. doi: 10.1002/fsn3.3425
Liu, S., Lei, T., Li, G., Liu, S., Chu, X., Hao, D., et al. (2023b). Rapid detection of micronutrient components in infant formula milk powder using near-infrared spectroscopy. Front. Nutr. 10:1273374. doi: 10.3389/fnut.2023.1273374
Luo, W., Li, Y., Urtasun, R., and Zemel, R. (2016). Understanding the effective receptive field in deep convolutional neural networks. Adv. Neural Inf. Proces. Syst. 29, 4905–4913. doi: 10.5555/3157382.3157645
Ma, P., Jia, X., He, Y., Tao, J., Wang, Q., and Wei, C.-I. (2024a). Recent progress of quantum dots for food safety assessment: a review. Trends Food Sci. Technol. 143:104310. doi: 10.1016/j.tifs.2023.104310
Ma, P., Jia, X., Zhang, X., Li, Y., He, Y., Li, T., et al. (2024b). Ion-chelated porous chitosan nanocrystal for highly efficient postharvest preservation. Matter 7, 2567–2580. doi: 10.1016/j.matt.2024.06.004
Ma, P., Tsai, S., He, Y., Jia, X., Zhen, D., Ning, Y., et al. (2024c). Large language models in food science: innovations, applications, and future. Trends Food Sci. Technol. 148:104488. doi: 10.1016/j.tifs.2024.104488
Mandal, R., Karmakar, P., and Roy, K. (2024). “Physicochemical analysis of milk powder” in Analytical methods for milk and milk products. eds. M. R. Goyal, N. Veena, and S. K. Mishra (New York, USA: Apple Academic Press), 93–156.
Mangan, D., McCleary, B. V., Culleton, H., Cornaggia, C., Ivory, R., McKie, V. A., et al. (2019). A novel enzymatic method for the measurement of lactose in lactose-free products. J. Sci. Food Agric. 99, 947–956. doi: 10.1002/jsfa.9317
Mao, Y., Ma, P., Li, T., Liu, H., Zhao, X., Liu, S., et al. (2024). Flash heating process for efficient meat preservation. Nat. Commun. 15:3893. doi: 10.1038/s41467-024-47967-1
Masotti, F., Cattaneo, S., Stuknytė, M., Pica, V., and De Noni, I. (2020). Analytical advances in the determination of calcium in bovine milk, dairy products and milk-based infant formulas. Trends Food Sci. Technol. 103, 348–360. doi: 10.1016/j.tifs.2020.07.013
McCarthy, N. A., Kelly, P. M., Maher, P. G., and Fenelon, M. A. (2014). Dissolution of milk protein concentrate (MPC) powders by ultrasonication. J. Food Eng. 126, 142–148. doi: 10.1016/j.jfoodeng.2013.11.002
McSweeney, D. J., Maidannyk, V., Montgomery, S., O’Mahony, J. A., and McCarthy, N. A. (2020). The influence of composition and manufacturing approach on the physical and rehydration properties of milk protein concentrate powders. Food Secur. 9:236. doi: 10.3390/foods9020236
Mu, F., Yu, G., Zhang, J., and Zhang, L. (2020). Milk source identification and milk quality estimation using an electronic nose and machine learning techniques. Sensors 20:4238. doi: 10.3390/s20154238
Munir, M. T., Wei Yu, B. R., Young, D. I., Wilson, I. I., and Centre, C. (2015). The current status of process analytical technologies in the dairy industry. Trends Food Sci. Technol. 43, 205–218. doi: 10.1016/j.tifs.2015.02.010
Murphy, E. G., Roos, Y. H., Hogan, S. A., Maher, P. G., Flynn, C. G., and Fenelon, M. A. (2015). Physical stability of infant milk formula made with selectively hydrolysed whey proteins. Int. Dairy J. 40, 39–46. doi: 10.1016/j.idairyj.2014.08.012
Murrieta-Pazos, I., Gaiani, C., Galet, L., Calvet, R., Cuq, B., and Scher, J. (2012). Food powders: surface and form characterization revisited. J. Food Eng. 112, 1–21. doi: 10.1016/j.jfoodeng.2012.03.002
Nde, D. B., and Foncha, A. C. (2020). Optimization methods for the extraction of vegetable oils: a review. PRO 8:209. doi: 10.3390/pr8020209
Nebbia, S., Ménard, O., Cochet, M.-F., Henry, G., Daniel, N., Moran, L., et al. (2023). Impact of process and composition of formulas for elderly on in vitro digestion using the dynamic DIDGI® model. Food Res. Int. 167:112716. doi: 10.1016/j.foodres.2023.112716
Neethirajan, S., Kobayashi, I., Nakajima, M., Dan, W., Nandagopal, S., and Lin, F. (2011). Microfluidics for food, agriculture and biosystems industries. Lab Chip 11, 1574–1586. doi: 10.1039/c0lc00230e
Niero, G., Thomas, S. A., Mouratidou, K., Visentin, G., De Marchi, M., Penasa, M., et al. (2023). Lactoferrin concentration in bovine milk: validation of radial immunodiffusion technique, sources of variation, and association to udder health status. Ital. J. Anim. Sci. 22, 230–238. doi: 10.1080/1828051X.2023.2180440
O’Callaghan, D., and Cunningham, P. (2005). Modern process control techniques in the production of dried milk products–a review. Lait 85, 335–342. doi: 10.1051/lait:2005021
Olaimat, A. N., Osaili, T. M., Al-Holy, M. A., Al-Nabulsi, A. A., Obaid, R. S., Alaboudi, A. R., et al. (2020). Microbial safety of oily, low water activity food products: a review. Food Microbiol. 92:103571. doi: 10.1016/j.fm.2020.103571
Pancholi, N., Gajera, H., and Shah, D. (2023). Improving quality of maintenance task for milk powder manufacturing unit through TOPSIS. J. Qual. Maint. Eng. 29, 27–49. doi: 10.1108/JQME-04-2021-0028
Písecký, J. (2012). Handbook of milk powder manufacture. Copenhagen, Denmark: GEA Process Engineering A/S.
Poo, S., Palma, M., and Muñoz, O. (2024). A new Design for Reference Values Assignment in proficiency testing for fat and crude protein in raw Milk for a limited number of participants. Food Secur. 13:2693. doi: 10.3390/foods13172693
Portnoy, M., and Barbano, D. M. (2021). Lactose: use, measurement, and expression of results. J. Dairy Sci. 104, 8314–8325. doi: 10.3168/jds.2020-18706
Pugliese, A., Cabassi, G., Chiavaro, E., Paciulli, M., Carini, E., and Mucchetti, G. (2017). Physical characterization of whole and skim dried milk powders. J. Food Sci. Technol. 54, 3433–3442. doi: 10.1007/s13197-017-2795-1
Rejeb, A., Keogh, J. G., Zailani, S., Treiblmaier, H., and Rejeb, K. (2020). Blockchain technology in the food industry: a review of potentials, challenges and future research directions. Logistics 4:27. doi: 10.3390/logistics4040027
Rysova, L., Cejnar, P., Hanus, O., Legarova, V., Havlik, J., Nejeschlebova, H., et al. (2022). Use of MALDI-TOF MS technology to evaluate adulteration of small ruminant milk with raw bovine milk. J. Dairy Sci. 105, 4882–4894. doi: 10.3168/jds.2021-21396
Rysova, L., Legarova, V., Pacakova, Z., Hanus, O., Nemeckova, I., Klimesova, M., et al. (2021). Detection of bovine milk adulteration in caprine milk with N-acetyl carbohydrate biomarkers by using 1H nuclear magnetic resonance spectroscopy. J. Dairy Sci. 104, 9583–9595. doi: 10.3168/jds.2020-20077
Sánchez, L., Pérez, M. D., Puyol, P., Calvo, M., and Brett, G.Collaborators: Cattaneo TMP Haasnoot W López-Fandiño R Manso MA Mata L Razquín P Sánchez A Svenning C Vegarud GE (2002). Determination of vegetal proteins in milk powder by enzyme-linked immunosorbent assay: Interlaboratory study. J. AOAC Int. 85, 1390–1397. doi: 10.1093/jaoac/85.6.1390
Sanchez, L. J., Zhu, D., Frew, R., and Kebede, B. (2021). Optimization of nuclear magnetic resonance and gas chromatography-mass spectrometry-based fingerprinting methods to characterize goat milk powder. J. Dairy Sci. 104, 102–111. doi: 10.3168/jds.2020-18467
Schaefer, J., Etges, B. I., and Schaefer, J. L. (2023). A literature database review of the competitive factors that influence the production and use of whey in the Brazilian dairy industry. Food Secur. 12:3348. doi: 10.3390/foods12183348
Schuck, P., Jeantet, R., Bhandari, B., Chen, X. D., Perrone, Í. T., Fernandes, A., et al. (2016). Recent advances in spray drying relevant to the dairy industry: a comprehensive critical review. Dry. Technol. 34, 1773–1790. doi: 10.1080/07373937.2016.1233114
Schuck, P., Jeantet, R., and Dolivet, A. (2012). Analytical methods for food and dairy powders. New Jersey, USA: John Wiley & Sons.
Seth, S., and Rathinasabapathi, P. (2022). A short review on detection of antibiotics in milk using nanomaterial-based biosensor. Food Anal. Methods 15, 2181–2192. doi: 10.1007/s12161-022-02291-6
Sharma, A. K., Bhatia, A. K., Kulshrestha, A., and Sawhney, I. K. (2021). Intelligent modeling of moisture sorption isotherms in Indian Milk products using computational neuro-genetic algorithm. SN Comp. Sci. 2:289. doi: 10.1007/s42979-021-00693-7
Sharma, A., Jana, A. H., and Chavan, R. S. (2012). Functionality of milk powders and milk-based powders for end use applications—a review. Compr. Rev. Food Sci. Food Saf. 11, 518–528. doi: 10.1111/j.1541-4337.2012.00199.x
Shuba, A., Umarkhanov, R., Bogdanova, E., Anokhina, E., and Burakova, I. (2024). Possibilities of an electronic nose on piezoelectric sensors with Polycomposite coatings to investigate the microbiological indicators of Milk. Sensors 24:3634. doi: 10.3390/s24113634
Sikand, V., Tong, P. S., Roy, S., Rodriguez-Saona, L. E., and Murray, B. A. (2011). Solubility of commercial milk protein concentrates and milk protein isolates. J. Dairy Sci. 94, 6194–6202. doi: 10.3168/jds.2011-4477
Sokoliuk, V. M., Dukhnytsky, V. B., Krupelnytsky, T. V., Ligomina, I. P., Revunets, A. S., and Prus, V. M. (2022). "influence of technological factors on milk quality indicators." scientific messenger of LNU of veterinary medicine and biotechnologies. Series Vet. Sci. 24, 37–43. doi: 10.32718/nvlvet10506
Stocchero, M., De Nardi, M., and Scarpa, B. (2021). PLS for classification. Chemom. Intell. Lab. Syst. 216:104374. doi: 10.1016/j.chemolab.2021.104374
Tian, H., Di, W., Chen, B., Yuan, H., Haiyan, Y., Lou, X., et al. (2023). Rapid identification and quantification of vegetable oil adulteration in raw milk using a flash gas chromatography electronic nose combined with machine learning. Food Control 150:109758. doi: 10.1016/j.foodcont.2023.109758
Ucar, Y., Ceylan, Z., Durmus, M., Tomar, O., and Cetinkaya, T. (2021). Application of cold plasma technology in the food industry and its combination with other emerging technologies. Trends Food Sci. Technol. 114, 355–371. doi: 10.1016/j.tifs.2021.06.004
Uthayaseelan, M. (2024). Data-driven optimization of an industrial cheese production process. Utgiver, Norwegian: Norwegian University of Life Sciences.
van Boven, A. P., Calderon, S. M., Novoa, R. K., and Schutyser, M. A. I. (2023). Investigation on nozzle zone agglomeration during spray drying using response surface methodology. Powder Technol. 429:118910. doi: 10.1016/j.powtec.2023.118910
Vasiljevic, T., Toebes, A., and Huppertz, T. (2021). Moisture sorption by dairy powders studied by low-field NMR. Int. Dairy J. 119:105062. doi: 10.1016/j.idairyj.2021.105062
Vassos, V., Voltezou, A., Stavropoulos, A., Stavropoulou, E., Stefanis, C., Tsigalou, C., et al. (2024). The role of Total quality Management in the Pharmaceutical, food, and nutritional supplement sectors. Food Secur. 13:2606. doi: 10.3390/foods13162606
Velásquez-Ferrín, A., Babos, D. V., Marina-Montes, C., and Anzano, J. (2021). Rapidly growing trends in laser-induced breakdown spectroscopy for food analysis. Appl. Spectrosc. Rev. 56, 492–512. doi: 10.1080/05704928.2020.1810060
Veličković, P. (2023). Everything is connected: graph neural networks. Curr. Opin. Struct. Biol. 79:102538. doi: 10.1016/j.sbi.2023.102538
Vieira, G. N. A., Olazar, M., Freire, J. T., and Freire, F. B. (2019). Real-time monitoring of milk powder moisture content during drying in a spouted bed dryer using a hybrid neural soft sensor. Dry. Technol. 37, 1184–1190. doi: 10.1080/07373937.2018.1492614
Wagh, N., and Agashe, S. D. (2023). A data-analytic framework to monitor product density of four-effect falling-film evaporator for skimmed milk production. J. Food Process Eng. 46:e14364. doi: 10.1111/jfpe.14364
Wang, D., Cao, Z., Gao, Y., Yang, L., and Zhao, L. (2024). Impact of the pre-dehydration and drying methods on the mass transfer and quality attributes of yak milk casein. Food Secur. 13:1062. doi: 10.3390/foods13071062
Wang, J., Zheng, Y., Xi, X., Hou, Q., Haiyan, X., Zhao, J., et al. (2018). Application of PacBio single molecule real-time (SMRT) sequencing in bacterial source tracking analysis during milk powder production. Food Control 93, 226–234. doi: 10.1016/j.foodcont.2018.05.044
Warncke, M., and Kulozik, U. (2020). Impact of temperature and high pressure homogenization on the solubility and rheological behavior of reconstituted dairy powders of different composition. Powder Technol. 376, 285–295. doi: 10.1016/j.powtec.2020.08.039
Wei, X., Lau, S. K., Chaves, B. D., Danao, M.-G. C., Agarwal, S., and Subbiah, J. (2020). Effect of water activity on the thermal inactivation kinetics of Salmonella in milk powders. J. Dairy Sci. 103, 6904–6917. doi: 10.3168/jds.2020-18298
Wei, D., Wang, X., Wang, N., and Zhu, Y. (2017). A rapid ion chromatography column-switching method for online sample pretreatment and determination of L-carnitine, choline and mineral ions in milk and powdered infant formula. RSC Adv. 7, 5920–5927. doi: 10.1039/C6RA25711A
Weston, M., Geng, S., and Chandrawati, R. (2021). Food sensors: challenges and opportunities. Adv. Mat. Technol. 6:2001242. doi: 10.1002/admt.202001242
Wu, S., Cronin, K., Fitzpatrick, J., and Miao, S. (2022). Updating insights into the rehydration of dairy-based powder and the achievement of functionality. Crit. Rev. Food Sci. Nutr. 62, 6664–6681. doi: 10.1080/10408398.2021.1904203
Wu, S., Fitzpatrick, J., Cronin, K., and Miao, S. (2019). The effect of pH on the wetting and dissolution of milk protein isolate powder. J. Food Eng. 240, 114–119. doi: 10.1016/j.jfoodeng.2018.07.022
Wu, D., He, Y., Feng, S., and Sun, D.-W. (2008). Study on infrared spectroscopy technique for fast measurement of protein content in milk powder based on LS-SVM. J. Food Eng. 84, 124–131. doi: 10.1016/j.jfoodeng.2007.04.031
Xi, X., Kwok, L.-Y., Wang, Y., Ma, C., Mi, Z., and Zhang, H. (2017). Ultra-performance liquid chromatography-quadrupole-time of flight mass spectrometry MSE-based untargeted milk metabolomics in dairy cows with subclinical or clinical mastitis. J. Dairy Sci. 100, 4884–4896. doi: 10.3168/jds.2016-11939
Yakubu, H. G., Kovacs, Z., Toth, T., and Bazar, G. (2022). Trends in artificial aroma sensing by means of electronic nose technologies to advance dairy production–a review. Crit. Rev. Food Sci. Nutr. 63, 234–248. doi: 10.1080/10408398.2021.1945533
Yan, J., Jin, M., Zhousu, X., Chen, L., Zhu, Z., and Zhang, H. (2020). Recognition of suspension liquid based on speckle patterns using deep learning. IEEE Photon. J. 13, 1–7. doi: 10.1109/JPHOT.2020.3044912
Yan, H., Ziyin, Y., and Liu, L. (2022). Lactose crystallization and Maillard reaction in simulated milk powder based on the change in water activity. J. Food Sci. 87, 4956–4966. doi: 10.1111/1750-3841.16335
Ye, M. P., Zhou, R., Shi, Y. R., Chen, H. C., and Du, Y. (2017). Effects of heating on the secondary structure of proteins in milk powders using mid-infrared spectroscopy. J. Dairy Sci. 100, 89–95. doi: 10.3168/jds.2016-11443
Younan, M., Houssein, E. H., Elhoseny, M., and Ali, A. A. (2020). Challenges and recommended technologies for the industrial internet of things: a comprehensive review. Measurement 151:107198. doi: 10.1016/j.measurement.2019.107198
Yu, H., Wang, J., Yao, C., Zhang, H., and Yong, Y. (2008). Quality grade identification of green tea using E-nose by CA and ANN. LWT Food Sci. Technol. 41, 1268–1273. doi: 10.1016/j.lwt.2007.08.018
Zhang, S., Li, H., Qiongzheng, H., Wang, Z., and Chen, X. (2022). Discrimination of thermal treated bovine milk using MALDI-TOF MS coupled with machine learning. Food Control 142:109224. doi: 10.1016/j.foodcont.2022.109224
Zhang, Haobo, Li, Yicheng, Weihao, Lu, and Lin, Qian. (2023). "On the optimality of misspecified kernel ridge regression." International Conference on Machine Learning.
Keywords: milk powder, quality detection, artificial intelligence, rapid detection, spectrum, indirect detection
Citation: Song X, Shen S, Dong G, Ding H, Xie Z, Wang L and Cheng W (2025) A comprehensive review of direct, indirect, and AI-based detection methods for milk powder. Front. Sustain. Food Syst. 9:1571317. doi: 10.3389/fsufs.2025.1571317
Edited by:
Qingli Dong, University of Shanghai for Science and Technology, ChinaReviewed by:
Seydi Yıkmış, Namik Kemal University, TürkiyePeihua Ma, Chinese Academy of Agricultural Sciences, China
Copyright © 2025 Song, Shen, Dong, Ding, Xie, Wang and Cheng. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
*Correspondence: Haohan Ding, ZGluZ2hhb2hhbkBqaWFuZ25hbi5lZHUuY24=