Detection of hospital super bacteria MRSA from the hands of healthcare professionals using machine learning and hyperspectral imaging

Healthcare-associated infections resulting from cross-contamination, particularly from the hands of multidisciplinary staff, significantly impact patient mortality in health units. The prolonged nature of classical phenotypic diagnostic methods underscores the need for faster and more precise alternatives. This study proposes an automatic procedure utilizing hyperspectral imaging (HSI) in shortwave infrared (SWIR) range to detect resistance to oxycillins in hospital bacteria. The automatic procedure employs partial least square with discriminant analysis (PLS-DA) for the classification of antibiotic-resistant and non-resistant bacteria. HSI data were obtained from samples collected from hands of eight intensive care unit workers using sisuCHEMA workstation. Results demonstrated effectiveness of proposed procedure in detecting oxycillin resistance in S. aureus samples and other 14 strains of Staphylococcus spp.