Study on the Classification of Electrogastrograms Using Artificial Intelligence Analysis in Frequency Space

Olfaction is generally evaluated using sensory evaluation. Objective evaluation has recently been performed using electrogastrogram (EGG). In this study, EGGs were measured with and without exposure to the lavender odor, and power spectra were obtained from the frequency analysis of the EGGs. Principal component analysis (PCA) was conducted to extract dimension-compressed data in the frequency space. A linear support vector machine (SVM) was then applied to classify the EGGs into those observed in the odorless condition and those recorded with exposure to the lavender odor. Binary classification by the SVM was evaluated by the correct rate and F value that were 100% resulted from this artificial intelligence analysis for all participants.