Summary
This paper presents a non-destructive approach for quantifying egg yolk ratio using visible–near-infrared hyperspectral imaging integrated with machine learning and explainable artificial intelligence. Partial least squares regression with Savitzky–Golay preprocessing achieved the strongest predictive performance (R² 0.68–0.79 across datasets), with Shapley additive explanations identifying water, lipid and protein spectral signatures as key contributors to model predictions. The method offers rapid, interpretable assessment for egg quality grading and nutritional evaluation without sample destruction.
Regional applicability
The technique could support UK egg grading standards and quality assurance in commercial poultry production, though geographic origin of the study and applicability to specific UK egg phenotypes or production systems cannot be determined from the abstract.
Key measures
Coefficient of determination (R²) for yolk ratio prediction; spectral wavelengths (374–1015 nm); model types (PLSR, random forest, extreme gradient boosting, support vector regression)
Outcomes reported
The study developed and validated a visible–near-infrared hyperspectral imaging method combined with machine learning to non-destructively predict egg yolk ratio. Performance was assessed across calibration, validation and independent test datasets, with the optimal model achieving R² values of 0.79, 0.73 and 0.68 respectively.
Topic tags
Dig deeper with Pulse AI.
Pulse AI has read the whole catalogue. Ask about this record, its theme, or how the findings apply to UK farming and policy — every answer cites the underlying studies.