Rapid Detection and Quantification of Adulterants in Fruit Juices Using Machine Learning Tools and Spectroscopy Data

Abstract

Fruit juice production is one of the most important sectors in the beverage industry, and its adulteration by adding cheaper juices is very common. This study presents a methodology based on the combination of machine learning models and near-infrared spectroscopy for the detection and quantification of juice-to-juice adulteration. We evaluated 100% squeezed apple, pineapple, and orange juices, which were adulterated with grape juice at different percentages (5%, 10%, 15%, 20%, 30%, 40%, and 50%). The spectroscopic data have been combined with different machine learning tools to develop predictive models for the control of the juice quality. The use of non-supervised techniques, specifically model-based clustering, revealed a grouping trend of the samples depending on the type of juice. The use of supervised techniques such as random forest and linear discriminant analysis models has allowed for the detection of the adulterated samples with an accuracy of 98% in the test set. In addition, a Boruta algorithm was applied which selected 89 variables as significant for adulterant quantification, and support vector regression achieved a regression coefficient of 0.989 and a root mean squared error of 1.683 in the test set. These results show the suitability of the machine learning tools combined with spectroscopic data as a screening method for the quality control of fruit juices. In addition, a prototype application has been developed to share the models with other users and facilitate the detection and quantification of adulteration in juices.

Novel Method Based on Ion Mobility Spectrometry Combined with Machine Learning for the Discrimination of Fruit Juices

“…It should be noted that other authors, such as Mansour Rasekh et al, reported similar results, with superior performance of the SVM model compared to LDA, for the detection of adulterations in fruit juices using data from an electronic nose [ 38 ]. In contrast, other studies have reported better performance with SVM models compared to RF and LDA when using NIR spectroscopic data [ 28 ], while, using FT-IR data, SVM and LDA models outperform RF [ 13 ].…”

“…Support vector machines (SVM) and random forest (RF) models are notable examples of these techniques. Both SVM and RF can be applied for classification and regression purposes, and they have shown excellent results in the characterization of different fruit juices using different spectroscopic data [ 13 , 28 , 29 , 30 ].…”

Current Techniques for Fruit Juice and Wine Adulterant Detection and Authentication

“…The applicability of NIRS for fruit juice and wine authenticity has been confirmed in previous studies. For example, in the case of juices, Calle et al [57] achieved a detection limit of 5% when using a combination of NIRS and machine learning models to detect juice-to-juice adulteration of pineapple, apple, and orange juices with grape juice. Following that, an accuracy of up to 97.67% was obtained for their discrimination model (LDA and RF models), and their support vector regression (SVR) model achieved a prediction error of less than 1.7% for adulteration level [57].…”

“…For example, in the case of juices, Calle et al [57] achieved a detection limit of 5% when using a combination of NIRS and machine learning models to detect juice-to-juice adulteration of pineapple, apple, and orange juices with grape juice. Following that, an accuracy of up to 97.67% was obtained for their discrimination model (LDA and RF models), and their support vector regression (SVR) model achieved a prediction error of less than 1.7% for adulteration level [57]. Shafiee et al [58] optimized data mining method on the NIR spectrum of lime juice to obtain the best classification result of 97% using support vector machine (SVM) method.…”

“…Compared with destructive techniques, nondestructive techniques have the advantages of little or no sample preparation, ease of use, reduced cost, high portability, and no need for reagents. Especially, some spectroscopic techniques have very rapid analysis that is performed within 1 min [57]. Although the feasibility of these techniques in adulteration detection and authentication was proven by many authors, there were challenges for applying these methods in real life.…”