Multiway Processing of Data Generated with a Potentiometric Electronic Tongue in a SIA System

Abstract

In this work a potentiometric electronic tongue based on an array of three Ion Selective Electrodes (ISEs) as sensor elements and multiway partial least squares (N-PLS2) regression method as processing tool has been developed for simultaneous quantification of Cd 2 + , Cu 2 + and Pb 2 + in aqueous solution mixtures. The sensor array was formed by two chalcogenide glass sensors selective to Cd 2 + and Cu 2 + plus one PVC membrane sensor selective to Pb 2 + . Preparation, measurement and injection of samples in a step profile were performed by an automated SIA system. Concentration range covered was 1.05 mM to 29.8 mM for Cd 2 + , 0.05 mM to 1.5 mM for Cu 2 + and 0.55 mM to 14.7 mM for Pb 2 + . Potentiometric recordings were arranged into a three-way array and treated by N-PLS2 regression method to obtain the calibration model. Prediction ability yielded RMSE values as low as 4.20 AE 0.37 mM, 0.094 AE 0.007 mM and 0.915 AE 0.006 mM, and correlation coefficients from linear regression analysis of 0.82 AE 0.038, 0.97 AE 0.005 and 0.97 AE 0.004 for Cd 2 + , Cu 2 + and Pb 2 + , respectively.

Determination of total polyphenol index in wines employing a voltammetric electronic tongue

“…Despite great advantages provided by the use of ANNs, when voltammetric sensors are used, the high complexity of the generated data matrix hinders their treatment. The straightforward solution is the use of multiway processing methods (samples x sensors x polarization potential), but the complexity of this technique is also critical [31,32]. One solution when dealing with a set of voltammograms is to employ a preprocessing stage for data reduction; this option permits to gain advantages in training time, to avoid redundancy in input data and to obtain a model with better generalization ability.…”

Bioelectronic Tongues Employing Electrochemical Biosensors

“…[24]. Obviously , chemometrics still form a significant part of ET research [25], particularly when the issues involved are artificial intelligence, among others the work with ANNs, or the multiway data treatment in the cases of highest dimensionality [26,27]. As already commented, 3-D data can be easily generated (e.g.…”

Bioelectronic tongues: New trends and applications in water and food analysis

“…Despite the progress made in the fields of chemometrics and computing, the high dimensionality of the generated data by an array of sensors still represents a limitation from both the modelling and computational perspective; and hence, a data pre-processing stage is required to achieve data compression and/or reduction of the high dimensionality of the generated data (generally by its unfolding into a 2D matrix (Cartas et al 2011)), while preserving the relevant information in a format compatible with the modelling tool. This is especially critical when data come from e.g.…”