An intelligent portable biosensor for fast and accurate nitrate determination using cyclic voltammetry

During the last two decades, enzyme-based analytical biosensors have been introduced for sensitive and efficient nitrate determination. Generally, in these methods, nitrate is measured based on the biocatalytic reduction of nitrate to nitrite. Since the activity of the immobilised enzyme on the working electrode (WE) reduces over time, the enzyme should be replaced after a small number of nitrate measurements, which limits the commercialisation of these biosensors. In this study, an intelligent portable biosensor is introduced which does not require frequent enzyme replacement. Instead, support vector machine learning method is utilised to predict the nitrate concentration considering the enzyme activity decrement over time. The introduced biosensor consisted of four main units: an enzyme-based three-electrode electrochemical unit, a signal processing and wireless data transfer unit, a decision making unit based on an application for iOS platform, and a unit for sharing the results through an internet of things (IoT)-based cloud server. Results showed that the trained support vector machine with polynomial kernel function resulted in promising nitrate prediction accuracy in which the values of R-2 and mean squared error (MSE) of the nitrate prediction were 0.93 and 0.0016, respectively. The prepared electrode was usable up to at least 10 days after the enzyme immobilisation for analysing nitrate in more than 400 samples without the need for enzyme replacement. Because of the cloud server unit for online sharing of the results, the device operator was able to share the results with environmentalists and plant clinics who work in the field of integrated nutrient management. (C) 2018 IAgrE. Published by Elsevier Ltd. All rights reserved.