Extensive evaluation of prediction performance for 15 pork quality traits using large scale VIS/NIRS data.

Area of Science:

• Agricultural Science
• Spectroscopy
• Animal Science

Background:

• Visible and near-infrared spectroscopy (VIS/NIRS) is a valuable tool for assessing meat quality in livestock and food industries.
• Accurate prediction of meat quality traits is crucial for optimizing pork production and consumer satisfaction.

Purpose of the Study:

• To evaluate the predictive performance of seven different models for 15 meat quality traits in pigs using VIS/NIRS data.
• To develop and validate a novel method, spectral-wide association analysis (SWAS), for selecting informative wavelengths for meat quality prediction.

Main Methods:

• Collected VIS/NIRS spectral data from 1206 pigs' longissimus muscle.
• Measured 15 key meat quality traits.
• Applied seven established chemometric models for trait prediction and developed the SWAS method for feature wavelength selection.

Main Results:

• The best predictive models achieved R_cv² values above 0.9 for most meat quality traits.
• The SWAS method identified significant association wavelengths, demonstrating a positive correlation between the number of selected wavelengths and prediction performance.
• Prediction accuracy using selected wavelengths was comparable to using the full spectral range, validating the feature selection approach.

Conclusions:

• VIS/NIRS, combined with appropriate modeling, offers high accuracy in predicting pig meat quality traits.
• The developed SWAS method is an effective tool for selecting crucial wavelengths, potentially simplifying and improving the efficiency of spectral-based meat quality assessment.