Production of Probiotic Kiwifruit Juice Containing Lacticaseibacillus paracasei B۳۱-۲: Investigation of Probiotic Viability, Physicochemical Properties, and AI Predictive Insights

This study examined the viability, physicochemical characteristics, and sensory qualities of kiwifruit juice containing Lacticaseibacillus paracasei B۳۱-۲. To analyze the data, Gaussian Process Regression (GPR) and Multi-Layer Perceptron (MLP) models were used to predict various factors, including pH, acidity, viable cell counts of L. paracasei B۳۱-۲, color differences (ΔE), and overall acceptance. Probiotic L. paracasei B۳۱-۲ was added to the kiwifruit juice at different concentrations (۰, ۱ and ۲%) and stored at ۴°C. The probiotic juices showed fewer changes in pH, acidity, and color compared to the control juice during storage at room temperature. The sample with a ۲% probiotic concentration exhibited the highest viable cell count (۷.۹۸ log CFU mL-۱) and received the most sensory scores among the tested samples. A strong correlation between the predictions made by the GPR model and the actual observed data further validated its effectiveness in similar experimental contexts. This suggests that GPR could offer strategic benefits by lowering laboratory costs and improving analytical efficiency. The GPR model's precision in closely matching real-world data demonstrates its potential as a cost-effective and expedited tool for scientific inquiries. Overall, these findings indicate that kiwifruit juice serves as a promising substrate for carrier of L. paracasei B۳۱-۲.

کلیدواژه ها:

Gaussian Process Regression ، Microbial load ، Multi-layer perceptron ، Sensory analysis