BACKGROUND: Rapid and accurate identification of intracellular pathogenic Brucella species and biovars is essential for effective public health surveillance, outbreak control, and preservation of animal and human health. While traditional biotyping remains the gold standard for biovar classification, it is time-consuming, technically demanding, dangerous, and costly. We validated artificial intelligence (AI)-coupled Fourier-transform infrared (FTIR) spectroscopy for Brucella biotyping as a faster, automated, and cost-effective alternative. RESULTS: One hundred and sixty-three strains of different species and biovars, originating from humans and animals, were assessed. The FTIR achieved high accuracy across all assessed species, achieving 95-100% sensitivity and specificity. On the biovar level, differentiating also vaccine from wild type strains, it achieved 80-100% sensitivity and specificity, depending on the species and the number of available strains for different biovars. CONCLUSIONS: Our research shows the identification of the FTIR as a highly robust and automated high-throughput method. This approach generates real-time data that shows potential for real-time epidemiological surveillance and allows early detection of outbreaks caused by zoonotic pathogens. By integrating classical microbiological methods with AI, the method shows great promise for its applicability in routine laboratory work and a high-throughput, fast, simple, safe, and cost-effective strategy for epidemiological surveillance of intracellular pathogenic Brucella bacteria on a global scale.
Duvnjak et al. (Thu,) studied this question.