A comparison of different chemometrics approaches for the robust classification of electronic nose data

Abstract

Accurate detection of certain chemical vapours is important, as these may be diagnostic for the presence of weapons, drugs of misuse or disease. In order to achieve this, chemical sensors could be deployed remotely. However, the readout from such sensors is a multivariate pattern, and this needs to be interpreted robustly using powerful supervised learning methods. Therefore, in this study, we compared the classification accuracy of four pattern recognition algorithms which include linear discriminant analysis (LDA), partial least squares-discriminant analysis (PLS-DA), random forests (RF) and support vector machines (SVM) which employed four different kernels. For this purpose, we have used electronic nose (e-nose) sensor data (Wedge et al., Sensors Actuators B Chem 143:365-372, 2009). In order to allow direct comparison between our four different algorithms, we employed two model validation procedures based on either 10-fold cross-validation or bootstrapping. The results show that LDA (91.56% accuracy) and SVM with a polynomial kernel (91.66% accuracy) were very effective at analysing these e-nose data. These two models gave superior prediction accuracy, sensitivity and specificity in comparison to the other techniques employed. With respect to the e-nose sensor data studied here, our findings recommend that SVM with a polynomial kernel should be favoured as a classification method over the other statistical models that we assessed. SVM with non-linear kernels have the advantage that they can be used for classifying non-linear as well as linear mapping from analytical data space to multi-group classifications and would thus be a suitable algorithm for the analysis of most e-nose sensor data.

Probing the action of a novel anti-leukaemic drug therapy at the single cell level using modern vibrational spectroscopy techniques

“…Statistics were then generated for the 1000 models in terms of correct classification rates and confusion matrices. This technique gives well represented estimates of the average model and is therefore less likely to be biased 55 .…”

Exhaled breath analysis, a simple tool to study the pathophysiology of obstructive sleep apnoea

“…The analytical potential of these devices depends heavily on the material of the sensors [87]. Because electronic noses cannot identify individual substances, but able to compare and distinguish complex molecular patterns, appropriate statistical approaches in electronic nose analysis are essential [88]. In addition, as numerous physiological and sampling-related methodological factors might modify exhaled volatile compound pattern [87], an external validation should be performed in an independent cohort to ensure results are reliable and valid [89].…”

Biomimetic cross-reactive sensor arrays: prospects in biodiagnostics

“…104–107 Each of the sensor types have various advantages and disadvantages; thus, it is important to consider the ultimate aim of analysis and type of sample to be characterized. Both mass and electrical/electrochemical sensors both provide sensitivity and selectivity in vapour and liquid analysis; however, these devices have a delicate setup, are unfortunately prone to drift due to changes in environment such as temperature and humidity, and require complex data analysis to extract the de-convolute the sensor response signal from the background noise.…”