Premium
Modeling the Shelf Life of Fruit‐Filled Snack Bars Using Survival Analysis and Sensory Profiling Techniques
Author(s) -
Corrigan Virginia,
Hedderley Duncan,
Harvey Winna
Publication year - 2012
Publication title -
journal of sensory studies
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.61
H-Index - 53
eISSN - 1745-459X
pISSN - 0887-8250
DOI - 10.1111/joss.12006
Subject(s) - shelf life , sensory system , sensory analysis , odor , food science , sample (material) , computer science , statistics , mathematics , psychology , chemistry , cognitive psychology , chromatography , neuroscience
Survival analysis and accelerated storage techniques were used to evaluate the shelf life of fruit‐filled snack bars. Survival analysis of the consumer data gave estimated shelf lives of 37, 16 and 8 weeks for bars stored at 20, 25 and 30 C , respectively. Reanalyzing the data with the first sample assessed excluded, greatly improved the shelf life confidence intervals indicating the value of including a warm‐up sample in the consumer sample set. Storage at 25 and 30 C reduced the shelf life by a factor of approximately 2 for every 5‐ C increase in storage temperature, as judged by consumer acceptability, and accelerated the rate of change of key sensory attributes evaluated by a trained sensory panel. The sensory attributes of fruit odor intensity, color development, sourness, freshness, uncharacteristic odors and uncharacteristic flavors were highly correlated with consumer rejection data and could be used as predictors of end of shelf life. Practical Applications This article supports survival analysis as a relatively simple methodology for use by the industry for the estimation of end of shelf life of shelf‐stable foods such as fruit‐filled snack bars, which can be slow and difficult to determine, driven by relatively small quality changes, rather than microbiological safety. Increased storage temperatures can be used to facilitate the process, accelerating the rate of change of key sensory attributes associated with consumer rejection. The data set can be modeled, allowing the estimation of such product rejection at any point in the product's shelf life and the level of risk selected accordingly. Shelf life confidence intervals are often wide when using this technique, but our findings indicate that including a warm‐up sample in the product sample set increases the precision of shelf life estimates. This has not been done previously, but we recommend that it be considered for future testing.