Research trends in spectroscopy combined with chemometrics for cheese analysis: a systematic review

Ensuring the authenticity and quality of cheese remains a major challenge that requires reliable, rapid, and non-destructive analytical tools. The combined use of spectroscopy and chemometrics has emerged as a powerful approach to address these needs. This systematic review analyzes trends and research gaps in their application to cheese analysis, based on 181 peer-reviewed articles retrieved from Scopus (1994–2024). Data were extracted on geographic distribution, milk species, spectroscopic techniques, and chemometric algorithms. Most studies were conducted in France, Italy, and Brazil, reflecting their strong cheesemaking traditions and regulatory attention to quality control. Cow’s milk predominates over sheep, goat, and buffalo milk, due to its broader availability and lower cost. NIR and MIR spectroscopy are the most frequently employed techniques, valued for their simplicity, speed, and suitability for industrial contexts. Regression models dominated analytical tasks, followed by classification and exploratory analyses, primarily implemented through PLSR, PLS-DA, and PCA, respectively. However, the gradual incorporation of machine learning algorithms indicates a shift toward capturing nonlinear patterns in complex spectral data. Common preprocessing methods included hybrid approaches, MSC, and SNV, often integrated with dimensionality reduction methods such as PCA and iPLS. Emerging research trends highlight (a) the integration of deep learning and open science practices, (b) multimodal data fusion and comparative assessment of spectroscopic techniques, and (c) the development of portable systems for in situ monitoring. Overall, this review provides valuable insights for academia, industry, and regulatory agencies, and can serve as a foundation for future, more comprehensive research.