Differentiation of closely-related species within Acinetobacter baumannii-calcoaceticus complex via Raman spectroscopy: a comparative machine learning analysis

• Published in: World journal of microbiology & biotechnology Vol. 40; no. 5; p. 146
• Main Authors: Xiong, Xue-Song, Yao, Lin-Fei, Luo, Yan-Fei, Yuan, Quan, Si, Yu-Ting, Chen, Jie, Wen, Xin-Ru, Tang, Jia-Wei, Liu, Su-Ling, Wang, Liang
• Format: Journal Article
• Language: English
• Published: Dordrecht Springer Netherlands 01.05.2024; Springer Nature B.V
• Subjects: Acinetobacter; Acinetobacter baumannii; Algorithms; Applied Microbiology; Biochemistry; Biomedical and Life Sciences; Biotechnology; Environmental Engineering/Biotechnology; human diseases; Learning algorithms; Life Sciences; Machine learning; Microbiology; prediction; Raman spectroscopy; Species; species identification; Spectroscopy; Spectrum analysis; Support vector machines; SERS spectra; Support vector machine; Machine learning; Surface-enhanced Raman spectroscopy; Acinetobacter baumannii
• ISSN: 0959-3993; 1573-0972; 1573-0972
• DOI: 10.1007/s11274-024-03948-6

Bacterial species within the Acinetobacter baumannii-calcoaceticus (Acb) complex are very similar and are difficult to discriminate. Misidentification of these species in human infection may lead to severe consequences in clinical settings. Therefore, it is important to accurately discriminate these pathogens within the Acb complex. Raman spectroscopy is a simple method that has been widely studied for bacterial identification with high similarities. In this study, we combined surfaced-enhanced Raman spectroscopy (SERS) with a set of machine learning algorithms for identifying species within the Acb complex. According to the results, the support vector machine (SVM) model achieved the best prediction accuracy at 98.33% with a fivefold cross-validation rate of 96.73%. Taken together, this study confirms that the SERS-SVM method provides a convenient way to discriminate between A. baumannii , Acinetobacter pittii , and Acinetobacter nosocomialis in the Acb complex, which shows an application potential for species identification of Acinetobacter baumannii-calcoaceticus complex in clinical settings in near future.