Premium
Chemosensory aerosol assessment of key attributes for tobacco products
Author(s) -
Soares Frederico L.F.,
Marcelo Marcelo C.A.,
Dias Jailson C.,
Juliano Luciana C.,
Porte Liliane M.F.,
Canova Luciana dos S.,
Ardila Jorge A.,
Pontes Oscar F.S.,
Sabin Guilherme P.,
Kaiser Samuel
Publication year - 2020
Publication title -
journal of chemometrics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.47
H-Index - 92
eISSN - 1099-128X
pISSN - 0886-9383
DOI - 10.1002/cem.3297
Subject(s) - curing of tobacco , sidestream smoke , repeatability , computer science , partial least squares regression , cigarette smoke , biochemical engineering , process engineering , mathematics , machine learning , statistics , engineering , toxicology , biology , botany
Human sensory evaluation plays an important role in assessing quality and supports commercial product development in many industries. However, sensory evaluation of tobacco products often requires highly trained specialists, is thus expensive, time‐consuming with low throughput capacity. To overcome such limitations, analytical platforms based on chemical fingerprints were proposed and evaluated in this work based on two different tobacco matrices. Using cigarette smoke as an example, the method was capable of predicting sensory attributes through chemical fingerprinting key tobacco and cigarette mainstream aerosol compositions. To achieve this, tobacco samples (comprising flue‐cured Virginia and air‐cured Burley types) and cigarette mainstream smoke samples were evaluated using high‐resolution mass spectrometry. The chemical fingerprint of each sample matrix was related to its respective reference sensory attributes, validated by highly trained panellists, to create predictive models based on partial least squares algorithm. These methodologies were further validated through a blind test with suitable prediction of all sensory attributes for single grade tobacco leaf and commercial blended cigarette smoke. The proposed methodology demonstrated satisfactory accuracy, repeatability and robustness, with prediction errors less than 20% for single grade tobaccos, and less than 11% for commercial products. When compared with human sensory evaluation, it significantly improved analytical capacity (over 100 samples per day) at comparable or improved accuracy. This method could be applied to evaluate other novel tobacco products such as heated tobacco products.