Portable Device Utilizes Machine Learning for On-Site Freshness Evaluation

Researchers from Sichuan University in Chengdu, China, used machine learning algorithms in a portable device to detect and differentiate aromatic molecules in food, according to a recent study published in Analytical Chemistry (1).

Spectroscopic techniques have been used for on-site food freshness assessment. Food quality is often determined by several factors, including the vitamins and minerals that the food product contains (2). Vitamins and minerals can be added to the food products in various ways. Sometimes, the vitamins and minerals naturally enter the food through the soil in which it grows, but in other instances, they are added during food processing (2). Determining the nutrient content in food, which is an important variable for consumers to consider when purchasing food items, requires monitoring the nutrient content in food products (2). Because the number of elemental nutrients is limited and can be present in high concentrations, analysts generally find inductively coupled plasma–optical emission spectroscopy (ICP-OES) to be the best technique for this type of analysis.

The recent study, led by Chengbin Zheng, developed a novel OES method to assess food freshness and adulteration detection. The method involved using a portable device integrated with a point discharge microplasma optical emission spectrometer (μPD-OES) where the output is coupled with machine learning (1).

Read More: Using Hyperspectral Imaging in Food Analysis

The device was equipped with two modular injection units: the headspace solid-phase microextraction (SPME) and headspace purge. These two units helped discern the unique patterns of light emitted by molecular and atomic emission spectra specific to different aromatic molecules present in the sample (1).

In their study, the researchers tested their device by introducing aromas from coffee and meat. Using machine learning algorithms, the portable device achieved higher accuracy rates in assessing meat freshness. It boasted accuracies of 96.0%, 98.7%, and 94.7% for beef, pork, and chicken meat, respectively, by analyzing optical emission patterns of the aroma at different storage times (1).

Moreover, the device could identify beef samples containing varying amounts of duck meat with an astonishing accuracy of 99.5%. It also successfully classified two coffee species without errors, showcasing its efficacy in discriminating food adulteration (1).

Read More: Portable Infrared Attenuated Total Reflection Spectrometer Tested for Food Analysis

Combining μPD-OES and machine learning had several benefits. First, it was a simple, portable technique for food aroma analysis. Second, it was also cost-effective, addressing the pressing need for simple, rapid, and low-cost analytical methods in ensuring food quality and safety (1).

The research presents a significant advancement in on-site food freshness evaluation and adulteration detection. By harnessing the power of machine learning and μPD-OES, the research team offer a versatile solution for real-time food safety monitoring (1).

The researchers also demonstrated that their new OES method could be used in various other sectors in food production, including in distribution and regulatory industries. The portability and ease of use of the device make it suitable for on-site inspections, empowering stakeholders to make informed decisions to safeguard public health (1).

As concerns regarding food safety continue to escalate globally, innovative technologies such as this portable device offer promising avenues for enhancing food quality control measures.

References

(1) Ren, T.; Lin, Y.; Su, Y.; et al. Machine Learning-Assisted Portable Microplasma Optical Emission Spectrometer for Food Safety Monitoring. Anal. Chem. 2024, 96 (13), 5170–5177. DOI: 10.1021/acs.analchem.3c05332

(2) Neubauer, K.; Spivey, N. Micronutrient Analysis from Soil to Food: Determination by ICP-OES. Spectroscopy Suppl. 2016, 31 (11), 8–17.