A data fusion approach was applied to a commercial honey data set analysed by (1)H-nuclear magnetic resonance (NMR) 400 MHz and liquid chromatography-high resolution mass spectrometry (LC-HRMS). The latter was performed using two types of mass spectrometers: an Orbitrap-MS and a time of flight (TOF)-MS. Fifty-six honey samples from four monofloral origins (acacia, orange blossom, lavender and eucalyptus) and multifloral sources from various geographical origins were analysed using the three instruments. The discriminating power of the results was examined by PCA first considering each technique separately, and then combining NMR and LC-HRMS together with or without variable selection. It was shown that the discriminating potential is increased through the data fusion, allowing for a better separation of eucalyptus, orange blossom and lavender. The NMR-Orbitrap-MS and NMR-TOF-MS mid-level fusion models with variable selection were preferred as a good discrimination was obtained with no misclassification observed for the latter. This study opens the path to new comprehensive food profiling approaches combining more than one technique in order to benefit from the advantages of several technologies. Graphical Abstract Data fusion between high resolution 1H-NMR and mass spectrometry.
Keywords: Botanical origin; Data fusion; Honey; LC-MS; Nuclear magnetic resonance.