Chemometric Approach Based on Explainable AI for Rapid Assessment of Macronutrients in Different Organic Fertilizers...

Area of Science:

• Agricultural Chemistry
• Analytical Chemistry
• Spectroscopy

Background:

• Traditional methods for analyzing macronutrients like potassium (K) and phosphorus (P) in organic fertilizers are costly and time-consuming.
• Established techniques such as ICP OES, FAAS, GF AAS, and ICP-MS require significant resources and expertise.

Purpose of the Study:

• To develop a rapid and efficient method for detecting potassium (P) and phosphorus (K) in organic fertilizers.
• To evaluate the combined potential of laser-induced breakdown spectroscopy (LIBS) and visible and near-infrared spectroscopy (Vis-NIR) for macronutrient analysis.
• To utilize Explainable AI (XAI) for feature extraction and enhance predictive model performance.

Main Methods:

• Combined LIBS and Vis-NIR spectroscopy for data acquisition.
• Employed Explainable AI (XAI) with Shapley additive explanation values (Shap values) for feature extraction.
• Integrated spectral data (spectra fusion) from both sensors.
• Utilized Support Vector Regression (SVR), Partial Least Squares Regression (PLSR), and Extremely Randomized Trees (Extratrees) for PK determination.

Main Results:

• Spectra fusion of LIBS and Vis-NIR significantly improved the detection accuracy for P (R²p = 0.9946, RMSEp = 0.0649%, RPD = 13.26) and K (R²p = 0.9976, RMSEp = 0.0508%, RPD = 20.28).
• XAI-extracted features enhanced the predictive capabilities of the models.
• The combined approach demonstrated superior efficiency compared to single-sensor detection methods.

Conclusions:

• The fusion of LIBS and Vis-NIR spectroscopy, enhanced by XAI, offers a powerful and efficient solution for rapid PK analysis in organic fertilizers.
• This integrated approach provides a viable alternative to traditional, labor-intensive analytical methods.
• The study highlights the potential of advanced spectroscopic techniques and AI in agricultural nutrient management.