Effects of drying time on physicochemical properties and volatile organic compounds of pork jerky revealed by GC-IMS and interpretable machine learning.

Drying time affects dehydration, color development, Maillard reactions, and volatile organic compounds (VOCs) in pork jerky. Pork jerky was dried at 80 °C for 4-6 h, with marinated undried samples as the RAW control. Physicochemical properties, free amino acids, and volatile fingerprints were evaluated. Gas chromatography-ion mobility spectrometry (GC-IMS) detected 53 volatile signals, whose profiles changed markedly during drying. Support vector regression (SVR) and extra trees regression (ETR) were developed to predict moisture content, absorbance at 420 nm, fluorescence intensity, and L*, a*, and b* values from GC-IMS. Both models achieved strong test-set performance, with R2 values of 0.996-0.998 for SVR and 0.989-0.995 for ETR. SHapley Additive exPlanations (SHAP) analysis identified shared and model-specific volatile markers, including pentanal, furaneol, ethyl propanoate, heptanal, 2-hexenal, 2-butanone, and 3-isobutyl-2-methoxypyrazine. These results demonstrate the potential of GC-IMS combined with interpretable machine learning for quality prediction and volatile-marker screening during pork jerky drying.