A SMART methodology for assessment of hexanal in potato crisps using electronic nose technology: sensor screening by scalar machine learning classifier method
- p150–160
There is a pertinent need to develop a rapid and accurate methodology for the detection of the onset and the progression of rancidity in the most popular savory product worldwide, viz. fried potato crisps for food safety and health concerns. Rancidity in the fried crisps—one set prepared using C18:2-lean deodorized virgin coconut oil under modified deep frying conditions (140 °C, 5 min),—and another set deep fried (170 °C, 3 min) in C18:2-rich oil (simulating commercial frying conditions) was determined by ‘rancidity indices’ generated (using Mahalanobis distance) from the data obtained by MO-based electronic nose analysis of hexanal (in Likens-Nickerson extract of volatiles from potato crisps), the most prominent rancidity marker, using screened sensors calibrated with standard hexanal, and classified using support vector machine. This also allowed unambiguous discrimination of the two sets of potato fries. The correlation of hexanal contents with the said indices yielded robust regression models which could accurately predict rancidity status of the crisps, forgoing GC-FID analysis of rancidity marker in the same. The ‘SMART’ models developed would allow rapid-cum-accurate detection of the onset and progression of rancidity in fried potato crisps on an industrial scale, forgoing the need to conduct biochemical analyses.
Potato crisps Hexanal Support vector machines Electronic nose Correlation equation