Near-Infrared Spectroscopy (NIRS) coupled with machine learning/chemometrics for screening pesticide residues in whole fruits and vegetables: systematic review of performance, chromatographic reference standards, and deployment readiness (2016–2026).

This protocol outlines a systematic review to evaluate the performance of near-infrared spectroscopy (NIRS) combined with machine learning/chemometrics for the detection, classification, and/or quantification of pesticide residues in whole fruits and vegetables. We will include peer-reviewed primary studies published in English between 01 Jan 2016 and 10 Feb 2026 that use validated chromatographic methods (GC/LC, with MS when applicable) as the reference standard. Searches will be conducted in Scopus, Web of Science, Embase and ScienceDirect, followed by study selection by two independent reviewers with adjudication by a third reviewer. Start will be used for reference management, deduplication, screening, and data extraction. Primary outcomes will be sensitivity/specificity and AUC (classification) and RMSEP and predictive R² (regression). Risk of bias will be assessed using QUADAS-2 for classification studies and PROBAST for regression/predictive modeling studies. Evidence will be synthesized narratively, with meta-analysis performed when data are sufficiently comparable, and will include assessment of operational deployment readiness (e.g., portability, time-to-result, and feasibility for food safety monitoring).