Determination of quality levels of fish oils recovered from trout waste using machine learning and odor sensors

In this study, chemical analyses and low-cost gas sensor measurements were conducted to determine the oxidative stability of fish oils obtained through enzymatic hydrolysis from rainbow trout (Oncorhynchus mykiss) waste. The initial EPA and DHA contents of the extracted oils were 1.31% and 4.94%, respectively, with oleic acid (C18:1n9) being the most abundant fatty acid at 38.59%. During a 20-day storage period, changes in free fatty acids, peroxide value, conjugated dienes, conjugated trienes, p-anisidine, and thiobarbituric acid values were monitored, along with alterations in fatty acid composition. Additionally, odor intensity was assessed using an electronic nose system. Sensor responses exhibited significant variations at different stages of the oxidative process, with MQ131 and MQ138 sensors demonstrating sensitivity to primary oxidation, while MQ3 and MQ135 sensors provided strong signals associated with the increase in secondary oxidation products. Overall, the study demonstrates the feasibility of using low-cost gas sensors for real-time oxidation assessment, contributing to improved fish oil stability management.