Мас-чутливий сенсорний масив та метод регресії на латентні структури для експрес-визначення загальної кількості мікроорганізмів ковбас

Abstract

The paper is reported about the development of novel rapid analytical technique with the use of quartz crystal microbalance (QCM) sensor-array combined with partial least squares regression for rapid determination of total microorganism number in sausages. The mass-sensitive sensors were used to analysed the changes in metabolites composition produced by microorganisms during spoilage of boiled sausages. Electronic nose data were collected from the headspace of sausages in parallel with data from standard microbiological analysis the quantity of mesophilic aerobic and facultative anaerobic microorganisms (microbial counting method). The informativeness of different features extracted from steady-state responses of the multisensor system was investigated. The area values and maximum response values were extracted as features from the electronic nose responses for evaluation and comparison the models fitting and performance of QMAFAnM prediction. The method of partial least squares regression and area values as features allowed to obtain a good performance of QMAFAnM prediction with a relative error less than 12%. For PLS regression model development, correlations above 0.95 and 0.99 were obtained between observed and predicted microbial counts for the training and test data sets, respectively. The new method of microbial counts prediction with the use of the electronic nose in combination with PLS regression will allow to significantly reduce the measurement time and the cost of analysis, and avoid subjective estimation of the results. The obtained recommendations of sensor-array data analysis to solve the analytical problem of effective safety control of food with the use of chemical sensing system. Keywords: mass-sensitive sensors; quantity of microorganisms (QMAFAnM); sausage spoilage; prediction; partial least squares regression