A Multimodal Spectroscopic Approach Combining Mid-infrared\nand Near-infrared for Discriminating Gram-positive and Gram-negative\nBacteria

The\nrapid and accurate identification of pathogenic bacteria is\ncrucial for combating the growing threat of antibiotic resistance,\nnosocomial infections, and food safety concerns. This study presents\na novel and comprehensive comparison of two vibrational spectroscopic\ntechniques - attenuated total reflection Fourier transform infrared\n(ATR-FTIR) spectroscopy and a low-cost miniature near-infrared (NIR)\nspectrometer - for distinguishing Gram-positive and Gram-negative\nbacterial samples grown using the same stock media solution. This\nis the first report of NIR spectroscopy being applied to differentiate\nGram-positive and Gram-negative bacteria, as well as the first direct\ncomparison of ATR-FTIR and NIR for the combined multimodal analysis\nof clinical bacterial isolates. Using a data set of five Gram-positive\nand seven Gram-negative species and recording spectra in triplicate,\nthe study employed advanced data fusion and multivariate analysis\ntechniques to classify the spectra and facilitate NIR band assignment.\n2D correlation analysis revealed strong positive correlations between\nkey spectral markers identified in both modalities. Partial least-squares-\nand support vector machine discriminant analysis models were validated\nusing a methodology based on 100 repeated random sampling of calibration\nand test sets. Models demonstrated that both the standalone ATR-FTIR\nand the combined ATR-FTIR/NIR approach achieved exceptional classification\naccuracy (>98%) in differentiating the two bacterial groups. Differences\nobserved in the spectra were attributed to the distinct cell wall\ncompositions of Gram-Positive and Gram-negative bacteria. Notably,\nthe low-cost NIR technique also showed promising performance, with\nclassification accuracy values above 90%. The findings highlight the\npotential of these rapid, noninvasive, and cost-effective vibrational\nspectroscopic techniques, particularly the NIR method, for point-of-care\napplications in clinical microbiology and food safety monitoring.\nThe combination of ATR-FTIR and NIR data further enhances the robustness\nand reliability of bacterial identification, paving the way for broader\nadoption of these advanced analytical tools in various healthcare\nand food safety settings.