[Simplification of NIR model for citrus's sugar content based on sensory methods]

Area of Science:

• Agricultural Science
• Food Science
• Spectroscopy

Background:

• Consumer demand for high-quality citrus necessitates accurate sugar content (SC) assessment.
• Non-destructive methods like near-infrared spectroscopy (NIRS) offer potential for rapid SC prediction.
• Validating NIRS predictions against human sensory perception is crucial for market acceptance.

Purpose of the Study:

• To validate NIRS-based SC predictions against consumer sensory evaluations in citrus.
• To simplify NIRS prediction models for citrus SC through variable selection.
• To assess the feasibility of NIRS for real-time citrus quality assessment.

Main Methods:

• Near-infrared spectroscopy (NIRS) was employed for non-destructive sugar content (SC) prediction in citrus.
• Sensory data from human evaluators were collected and analyzed.
• Statistical methods including one-way ANOVA, outlier exclusion, and variable selection (SPA-MLR) were used to develop and refine the NIRS model.

Main Results:

• A significant difference in sensory perception was observed due to individual diversity, not gender.
• The optimized NIRS model (SPA-MLR) demonstrated robustness with a prediction set correlation (Rp) of 0.86 and a root mean square error of prediction (RMSEP) of 0.567.
• The developed NIRS model achieved a prediction time of just 6.8 ms, meeting sensory requirements.

Conclusions:

• NIRS technology can effectively predict citrus sugar content to meet consumer sensory expectations.
• The simplified SPA-MLR model enhances the practicality of NIRS for real-time citrus quality assessment.
• This approach provides a valuable reference for future industrial applications in the fruit industry.