Geochemical fingerprinting and machine learning for authenticating sparkling wine origins

The global wine market faces persistent threats from counterfeiting, particularly for high-value segments like sparkling wines. Traditional authentication methods relying on supply chain traceability and geographical indications are insufficient to curb rampant fraud, posing economic and health risks. This study pioneers a scalable geochemical fingerprinting framework, combining isotopic and elemental analyses with advanced machine learning, to authenticate sparkling wine origins. Using 75 French sparkling wine samples from Champagne and Burgundy, we achieved 100% classification accuracy with strontium isotopic ratios (⁸⁷Sr/⁸⁶Sr), which uniquely reflect geological characteristics of vineyards. To mitigate high analytical costs, the concentration of rubidium (Rb) was identified as a cost-effective alternative, reducing expenses by 75% while maintaining over 90% accuracy.