Gemma Matute

Linking Chemical Parameters to Sensory Panel Results through Neural Networks To Distinguish Olive Oil Quality

A wide variety of olive oil samples from different origins and olive types has been chemically analyzed as well as evaluated by trained sensory panelists. Six chemical parameters have been obtained for each sample (free fatty acids, peroxide value, two UV absorption parameters (K232 and K268), 1,2-diacylglycerol content, and pyropheophytins) and linked to their quality using an artificial neural network-based model. Herein, the nonlinear algorithms were used to distinguish olive oil quality. Two different methods were defined to assess the statistical performance of the model (a K-fold cross-validation (K = 6) and three different blind tests), and both of them showed around a 95-96% correct classification rate. These results support that a relationship between the chemical and the sensory analyses exists and that the mathematical tool can potentially be implemented into a device that could be employed for various useful applications.

Identifying and Quantifying Adulterants in Extra Virgin Olive Oil of the Picual Varietal by Absorption Spectroscopy and Nonlinear Modeling

In this research, the detection and quantification of adulterants in one of the most common varieties of extra virgin olive oil (EVOO) have been successfully carried out. Visible absorption information was collected from binary mixtures of Picual EVOO with one of four adulterants: refined olive oil, orujo olive oil, sunflower oil, and corn oil. The data gathered from the absorption spectra were used as input to create an artificial neural network (ANN) model. The designed mathematical tool was able to detect the type of adulterant with an identification rate of 96% and to quantify the volume percentage of EVOO in the samples with a low mean prediction error of 1.2%. These significant results make ANNs coupled with visible spectroscopy a reliable, inexpensive, user-friendly, and real-time method for difficult tasks, given that the matrices of the different adulterated oils are practically alike.

Inputting molecular weights into a multilayer perceptron to estimate refractive indices of dialkylimidazolium-based ionic liquids-A purity evaluation

Neural networks to estimate the refractive index of imidazolium-based ionic liquids.Purity determination by comparison of estimated and experimental refractive indices.Only the molecular weights are required as previously known information.Chemical approach by relating molecular weights and refractive indices.Successful results obtained with both bibliographical and experimental data. Dialkylimidazolium-based ionic liquids (ILs) are one of the most employed and accessible ILs. These novel chemicals possess unique physicochemical properties which, unfortunately, are greatly altered by impurities. A simple method to evaluate the purity level of ILs is proposed, as a direct relationship exists between refractive index (RI) and purity. Two multilayer perceptrons (MLPs) have been designed to estimate the RI values using the molecular weights (MWs) of the imidazolium-based ILs. The RI is defined as the single output of the created neural network models. These MLPs offered low verification prediction errors (less than 0.48% in both cases), thus leading to useful mathematical tools that are able to more than adequately estimate the RI of imidazolium-based ILs by solely relying on the MWs. Therefore, an extremely manageable mathematical tool that can accurately estimate the RIs of imidazolium-based ILs, and, in the end, their purity, has been created. Additional tests were developed with experimental data regarding two imidazolium-based ILs to evaluate the applicability of the networks, and the results were successful in terms of RI and purity estimation.

Algorithmic modeling of spectroscopic data to quantify binary mixtures of vinegars of different botanical origins

Multiple binary mixtures of different kinds of vinegars have been analyzed through UV-Vis absorption. Two types of mathematical models (multiple linear regression (MLR) and artificial neural networks (ANNs)) have been employed to identify and quantify the components of such blends. Six different vinegars were used to prepare these mixtures, each one with a particular botanical origin: white wine, red wine, apple cider, apple, molasses, and rice. The best results have been obtained with ANN based models, offering mean estimation error value averages of 1% (v/v) and mean correlation coefficients (R2) over 0.99. This model is adequate to perform the estimation and achieve an accurate and reliable tool. Nevertheless, although the MLR models provide worse results (0.88 in terms of R2 and 5% v/v error), they can be used depending on the application and required accuracy.

Volatile Organic Compound Detection with FET Sensors and Neural Network Data Processing as a Preliminary Step to Early Lung Cancer Diagnosis

Cancer is currently one of deadliest and most feared diseases in the developed world, and, particularly, lung cancer (LC) is one of the most common types and has one of the highest death/incidence ratios. An early diagnosis for LC is probably the most accessible possibility to try and save patients and lower this ratio. Recently, research concerning LC-related breath biomarkers has provided optimistic results and has become a real option to try and obtain a fast, reliable, and early LC diagnosis. In this paper, a combination of field-effect transistor (FET) sensors and artificial neural networks (ANNs) has been employed to classify and estimate the partial pressures of a series of polar and nonpolar volatile organic compounds (VOCs) present in prepared gaseous mixtures. The objective of these preliminary tests is to give an idea of how well this technology can be used to analyze artificial or real breath samples by quantifying the LC-related VOCs or biomarkers. The results of this step are very promising and indicate that this methodology deserves further research using more complex samples to find the existing limitations of the FET-ANN combination.

Modelling of Hydrocarbon Solubility in Isomeric Ionic Liquids Using Mathematical Regressions

Linear Regression and Radial Basis Network models were explored for the prediction of heptane and toluene solubility in eleven ionic liquids at 313.15 K. The ILs were grouped in three groups of isomers. The models fitting performance was analyzed calculating statistical parameters (correlation coefficient and mean prediction error). For every model tested, the mean prediction error values in all three systems studied is less than 9.26 and 0.8% for linear regression and radial basis network models, respectively.

P1.9.16 Nonlinear Algorithms to Reduce the Dimension of Databases without Loss of Information

In this work, self-organizing maps have been used in the reduction of the dimensionality of the database, extracting the essential information of the database, facilitating its handling and reducing the time needed by the sensor to give the measurement. This tool has been applied to a database composed of 220 1H NMR and 31P NMR spectra of 13 using edible vegetal oils (hazelnut, sunflower, corn, soybean, sesame, walnut, rapeseed, almond, palm, groundnut, safflower, coconut, and extra virgin olive oils). With this tool, the dimension of the databases decreases from 11 x 192 to 2 x 192. The loss of information was checked by comparing the statistical results shown here with others that can be found in literature. Here, using a low dimension database and without any other physicochemical feature, the statistical results have been slightly improved (the misclassification percentage decreases from 3 to less than 2.8%).