Digitalization of wheat mold odor based on controlled volatile release: A large-scale study

Area of Science:

• Agricultural Science
• Sensor Technology
• Data Science

Background:

• Accurate assessment of grain spoilage is crucial for food safety and quality control.
• Traditional methods for detecting mold in wheat are often time-consuming and subjective.
• Digitalization of odor profiles offers a promising avenue for objective and rapid quality assessment.

Purpose of the Study:

• To develop a large-scale framework for the digitalization of wheat mold odor.
• To establish a standardized method for gas acquisition and odor detection in wheat.
• To create a machine learning model for rapid and non-destructive mold assessment in wheat.

Main Methods:

• Cultivation of seven wheat varieties across four spoilage levels (normal to severe mildew).
• Development of a graphene-based sensor array for multidimensional odor detection.
• Implementation of a dual pre-treatment strategy (temperature-regulated volatilization and cooling-assisted dehumidification).
• Machine learning modeling using 1491 odor response curves, with external validation on 503 samples.

Main Results:

• The optimized Light Gradient Boosting Machine model achieved high performance for binary classification (normal vs. moldy wheat).
• Accuracy: 95.8%, Sensitivity: 95.5%, Specificity: 96.8%, Area Under the Curve (AUC): 0.984.
• The dual pre-treatment strategy ensured stable and comparable signal acquisition for reliable odor detection.

Conclusions:

• The proposed framework enables rapid and non-destructive assessment of mold in wheat.
• This approach supports standardized grain quality monitoring, enhancing food safety.
• Digitalization of wheat odor profiles using sensor arrays and machine learning is a viable strategy for quality control.