박종민 (한림대학교)

김재석 (한림대학교 의과대학)

박종민 (한림대학교)

용동은 (연세대학교 의과대학)

박정수 (서울대학교 의과대학)

김경남 (연세대학교 의과대학)

서동건 ((주)엔큐랩)

김동찬 ((주)엔큐랩)

김재석 (한림대학교 의과대학)

박종민 (한림대학교)

Rapid detection of methicillin-resistant <i style="font-family: "Times New Roman", serif; font-size: 12pt;">Staphylococcus aureus</i> (MRSA) is essential to prevent healthcare-associated infections such as bacteremia, pneumonia, and surgical wound infections, and prompt treatment of antimicrobials against MRSA improves treatment outcomes. However, traditional MRSA screening tests based on molecular diagnostics are time-consuming, labor-intensive, and costly. The objective of this study was to develop and evaluate the rapid MRSA screening software based on MALDI-TOF MS with machine learning. AMRQuest software was developed to be able to compare MALDI-TOF mass spectra of <i style="font-family: "Times New Roman", serif; font-size: 12pt;">S. aureus</i> with a database by working on machine learning techniques and was successfully used to screen MRSA and identify the bacterial species simultaneously. From the test, the sensitivity, specificity, percent agreement, and Cohen’s kappa value were calculated to determine the accuracy of the AMRQuest software. The SCC<i style="font-family: "Times New Roman", serif; font-size: 12pt;">mec</i>A gene was detected to compare the discrepancy between the cefoxitin disk diffusion test and the results of AMRQuest MRSA screening. Using the results from the AMRQuest software, MRSA and MSSA were successfully distinguished statistically, and the PPV and NPV were estimated to be 97.4% and 99.3%, respectively. In conclusion, the clinical performance of AMRQuest software for MRSA screening was evaluated to determine if it would be sufficient for use in laboratories.