Integrating NMR Spectroscopy and Machine Learning for Quality Assessment and Geographic Discrimination of Olive Oils

Olive oil quality is shaped by genetic, environmental, and processing factors, demanding advanced analytical methods for precise characterization. This study introduces an innovative pipeline integrating ¹H NMR spectroscopy, GC-MS/HPLC, and machine learning to address three key challenges: authentication, quality control, and terroir profiling. We developed a rapid, sustainable NMR-based metabolic fingerprinting method that quantifies 50 critical parameters (e.g., fatty acids, polyphenols, tocopherols, sensory traits) in a single analysis while distinguishing olive cultivars and harvest years. Applying this approach to Tuscan EVOOs (monovarietal and poly-varietal), combined with GC-MS/HPLC and chemometrics, we found milling practices contributed to 13% of compositional variation. A two-stage machine learning strategy isolated geographic effects, identifying 33 robust markers (e.g., nonanal, hexyl acetate, oleuropein derivatives, cycloartenol) enabling 80% accuracy in origin classification. The same methodology distinguished filtered vs. non-filtered EVOOs with 90% accuracy. These results establish a dual-purpose framework: 1) a high-precision quality control system for industry, and 2) insights into how genetics, environment, and post-harvest practices collectively shape oil chemistry. The approach supports PDO certification by linking chemical fingerprints to terroir, guides genotype selection for climate adaptation, and optimizes processing to preserve raw material integrity. By decoding genotype-environment interactions, we enable targeted olive breeding for metabolic traits suited to specific growing conditions—a critical tool for climate resilience. Merging advanced analytics with AI, this research bridges analytical chemistry and agricultural innovation, offering transformative solutions for olive oil science and industry.