Effects on quality characteristics of ultrasound-treated gilaburu juice using RSM and ANFIS modeling with machine learning algorithm

Abstract

Gilaburu (Viburnum opulus L.) is a red-colored fruit with a sour taste that grows in Anatolia. It is rich in various antioxidant and bioactive compounds. In this study, bioactive compounds and ultrasound parameters of ultrasound-treated gilaburu water were optimized by response surface methodology (RSM) and adaptive neuro-fuzzy inference system (ANFIS). As a result of RSM optimization, the independent ultrasound parameters were determined as an ultrasound duration of 10.7 min and an ultrasound amplitude of 53.3, respectively. The R2 values of the RSM modeling level were 99.93%, 98.54%, and 99.80%, respectively, and the R2 values of the ANFIS modeling level were 99.99%, 98.89%, and 99.87%, respectively. Some quality parameters of gilaburu juice were compared between ultrasound-treated gilaburu juice (UT-GJ), thermal pasteurized gilaburu juice (TP-GJ), and control group (C-GJ). The quality parameters include bioactive compounds, phenolic compounds, minerals, and sensory evaluation. Bioactive compounds in the samples increased after ultrasound application compared to C-GJ and TP-GJ samples. The content of 15 different phenolic compounds was determined in Gilaburu juice samples, and the phenolic compound of UT-GJ samples increased compared to TP-GJ and C-GJ samples, except for gentisic acid. Ultrasound treatment applied to gilaburu juice enabled its bioactive compounds to hold more in the juice.

Keywords

• Gilaburu
• Ultrasound
• ANFIS
• Bioactive compounds
• Machine learning