Whey Protein Fraud: How Portable NIR Spectroscopy and AI Can Combat This Issue

In today’s modern society, there has never been a greater emphasis on fitness. As a result, health-conscious food products, including dietary supplements, are growing in demand. However, bad actors are capitalizing on this demand by pushing forward fraudulent ingredients found in protein supplements to the market. A recent study published in Food Research International explored this topic. Conducted by researchers from several Chinese institutions, including Tsinghua University and Hainan University, this study presents a nondestructive and accurate method for protein content and detecting adulterants in commercial whey protein concentrate (WPC) products (1).

Whey protein contains significant nutritional value, which makes it a popular choice for fitness enthusiasts and athletes. Some of the health benefits of whey protein include promoting muscle growth, lowering blood pressure, reducing inflammation, and enhancing antioxidant defenses (2). Although excessive whey protein intake can lead to digestive issues, including nausea, bloating, and cramping, moderate consumption of whey protein helps improve a person’s health (2).

What Did The Researchers Do In Their Study?

In their study, the researchers combined several spectroscopic techniques to create their new method. They used portable near-infrared (NIR) spectroscopy, visible near-infrared hyperspectral imaging (HSI), and advanced machine learning algorithms (1). As part of their experimental procedure, the researchers analyzed WPC samples from 15 different brands using portable NIR spectroscopy and Vis-NIR HSI devices (1). These tools were used because they can capture the spectral data and decipher the chemical composition of the samples under study. To identify key differences between adulterated and authentic products, the researchers utilized three chemometric techniques, including principal component analysis (PCA), partial least squares discriminant analysis (PLS-DA), and a machine learning method known as the kernel extreme learning machine (K-ELM), comparing all three techniques to determine which one was most effective (1).

What Were The Findings Of This Study?

The findings of this study revealed the overall effectiveness of both the NIR and HSI data in differentiating between the various WPC brands. The study also showed that chemometrics was vital in enhancing the accuracy of the researchers’ method. PCA, PLS-DA, and K-ELM were integral in achieving high classification accuracy (1). For quality analysis, the team used partial least squares regression (PLSR) and K-ELM to quantify the protein and carbohydrate contents of each sample (1). They found that the PLSR method using the NIR data resulted in a coefficient of prediction (R²_P) as high as 0.991 for both protein and carbohydrate levels, with root mean square errors (RMSE) as low as 0.023 (1).

What Makes This Study Unique?

This study was unique because it integrated several chemometric algorithms and NIR spectroscopy, and HSI. The study also reflected an ongoing trend in food analysis, and that is using portable instrumentation and artificial intelligence (AI) to improve method accuracy and efficiency. Because the team spiked WPC samples with known concentrations of three common adulterants, maltodextrin, wheat flour, and milk powder, ranging from 5% to 50% by weight, the system successfully identified these additives with great precision (1).

Another key aspect of this study was that the researchers showed how K-ELM can help them achieve up to 100% accuracy in detecting the presence of adulterants, and R²_P values ranging from 0.973 to 0.997 for quantifying adulterant levels, with corresponding RMSE values between 0.009 and 0.026 (1).

In addition to demonstrating high performance in adulterant detection, the team also extracted useful spectral fingerprints associated with protein and carbohydrate content, which could further enhance the robustness of future classification models. Based on PCA and PLSR analyses, the researchers were able to pinpoint spectral bands that serve as reliable indicators of nutritional composition, paving the way for automated, handheld authentication tools.

What Are The Next Steps and Future Directions?

Although this study was an encouraging start, the researchers acknowledged that more work needs to be done to refine their model. For example, the researchers want to integrate portable NIR spectroscopy with smartphone-based video imaging. By doing so, they hope to create more accessible quality assessment tools for consumers and regulators (1). Moreover, upcoming studies aim to expand detection capabilities to include nitrogen-based adulterants, which are particularly challenging to identify but commonly used to manipulate protein content readings (1).

Ultimately, this study shows how technological advancements are improving food authentication analysis. As whey protein and other healthy food additives grow in demand, the expectation is that researchers will continue to tinker with spectroscopy methods and technology to build better predictive models for detecting food adulteration.

References

1. Song, W.; Yun, Y.-H.; Lv, Y.; et al. Authentication and Quality Assessment of Whey Protein-based Sports Supplements Using Portable Near-Infrared Spectroscopy and Hyperspectral Imaging. Food Res. Int. 2025, 203, 115807. DOI: 10.1016/j.foodres.2025.115807
2. Anarson, A. 10 Evidence-Based Health Benefits of Whey Protein. Healthline.com. Available at: https://www.healthline.com/nutrition/10-health-benefits-of-whey-protein#:~:text=2.,4%20%2C%205%20%2C%206%20). (accessed 2025-05-15).