Sexaje de huevo fértil de gallinas reproductoras Hy-Line mediante Espectroscopía de Infrarrojo Cercano (NIR) portátil y herramientas quimiométricas
No Thumbnail Available
Date
2024
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Universidad Nacional de Trujillo
Abstract
La determinación del sexo in ovo de embriones de gallinas ponedoras antes de que perciban dolor resolvería el problema ético y de bienestar animal asociado con el sacrificio masivo de pollitos machos recién nacidos. Además, permitiría ahorro de costos y espacio en el proceso de incubación. El objetivo de esta investigación fue determinar el sexo de huevos fértiles de gallinas reproductoras Hy-Line mediante un espectrómetro (NIR) portátil y herramientas quimiométricas. Se utilizó 200 huevos incubables, 130 se destinaron al desarrollo del modelo predictivo y 70 para su validación externa. Se adquirieron datos espectrales NIR en los días de incubación 0, 1, 3, 5, 7, 8 y 10. Los datos espectrales fueron preprocesados con las técnicas MC (Media Centralizada), SNV (Normalización de Variabilidad Estándar) y 2° derivada Savitzky-Golay. El análisis de los datos se realizó en el software Matlab R2020a, mediante un análisis de componentes principales (PCA) y análisis discriminante de mínimos cuadrados parciales (PLS-DA). En la fase de calibración, los modelos PLS-DA con mejor desempeño fueron en los días 0 y 5 con una precisión de predicción de 66,2% para machos y 65% para hembras. En la fase de validación externa, el modelo con mejor rendimiento fue en el día 0 con una precisión de predicción de 47,6% para machos, y 66,6% para hembras. En conclusión, un espectrómetro (NIR) portátil y herramientas quimiométricas, pueden sexar in ovo huevos de gallinas Hy-Line, con una precisión baja, por ser un área en desarrollo, es fundamental mejorar la precisión y robustez del modelo a través de la integración estratégica de herramientas de Inteligencia Artificial (IA) que mejoren el preprocesamiento y análisis de los datos espectrales.
ABSTRACT Determining the in ovo sex of laying hen embryos before they perceive pain would solve the ethical and animal welfare problem associated with the mass slaughter of newborn male chicks. It would also allow cost and space savings in the incubation process. The objective of this research was to determine the sex of fertile eggs from Hy-Line breeder hens using a portable (NIR) spectrometer and chemometric tools. 200 hatching eggs were used, 130 were used for the development of the predictive model and 70 for external validation. NIR spectral data were acquired on incubation days 0, 1, 3, 5, 7, 8 and 10. The spectral data were preprocessed with the MC (Centralized Mean), SNV (Standard Variability Normalization) and 2nd derivative Savitzky techniques. -Golay. Data analysis was performed in Matlab R2020a software, using principal component analysis (PCA) and partial least squares discriminant analysis (PLS-DA). In the calibration phase, the PLS-DA models with the best performance were on days 0 and 5 with a prediction accuracy of 66.2% for males and 65% for females. In the external validation phase, the model with the best performance was on day 0 with a prediction accuracy of 47.6% for males, and 66.6% for females. In conclusion, a portable (NIR) spectrometer and chemometric tools can sex in ovo eggs from Hy-Line chickens, with a low precision. As it is a developing area, it is essential to improve the precision and robustness of the model through integration. strategy of Artificial Intelligence (AI) tools that improve the preprocessing and analysis of spectral data.
ABSTRACT Determining the in ovo sex of laying hen embryos before they perceive pain would solve the ethical and animal welfare problem associated with the mass slaughter of newborn male chicks. It would also allow cost and space savings in the incubation process. The objective of this research was to determine the sex of fertile eggs from Hy-Line breeder hens using a portable (NIR) spectrometer and chemometric tools. 200 hatching eggs were used, 130 were used for the development of the predictive model and 70 for external validation. NIR spectral data were acquired on incubation days 0, 1, 3, 5, 7, 8 and 10. The spectral data were preprocessed with the MC (Centralized Mean), SNV (Standard Variability Normalization) and 2nd derivative Savitzky techniques. -Golay. Data analysis was performed in Matlab R2020a software, using principal component analysis (PCA) and partial least squares discriminant analysis (PLS-DA). In the calibration phase, the PLS-DA models with the best performance were on days 0 and 5 with a prediction accuracy of 66.2% for males and 65% for females. In the external validation phase, the model with the best performance was on day 0 with a prediction accuracy of 47.6% for males, and 66.6% for females. In conclusion, a portable (NIR) spectrometer and chemometric tools can sex in ovo eggs from Hy-Line chickens, with a low precision. As it is a developing area, it is essential to improve the precision and robustness of the model through integration. strategy of Artificial Intelligence (AI) tools that improve the preprocessing and analysis of spectral data.
Description
Keywords
Sexaje in ovo, Espectroscopía de infrarrojo cercano, Quimiometría, Avicultura, Huevo fértil