Validation of APAS Independence (CCS) AI-Algorithms to detect MRSA in a routine setup
Culture-based MRSA screening is an important part of infection control in everyday clinical practice where fast and reliable results are essential. Artificial Intelligence (AI)-based systems may increase the sample throughput, decrease time to result and increase test accuracy. This study was conducted to validate the use of AI-based MRSA detection in combination with an automated inoculation system in a routine setup.
A total of 5,122 MRSA screening samples were examined in three west German hospitals over a 3-month period. The samples were inoculated on Columbia agar with Sheep Blood and chromogenic MRSA agar (biplate) by an automated inoculation system. APAS Independence was then used to classify growth after 24 hours and all samples were additionally checked by experienced medical laboratory technicians and microbiologists. The results of AI-based classification and conventional plate reading were compared.
From 5,122 samples, 65 (1.27%) were positive for MRSA. APAS Independence AI-based reading showed no false negative results leading to a sensitivity and negative predictive value (NPV) of 100% under routine conditions to detect MRSA. The establishment of AI-based MRSA detection in routine microbiology laboratory can significantly reduce the number of samples that must be processed manually by medical laboratory technicians and microbiologists. Thus, sample throughput can be upscaled with no loss of precision or accuracy.
Poster Presentation: by Labor Dr Wisplinghoff
Conference: ECCMID 2021, Online
Date: July 2021
Authors: Giglio S, Jazmati N, Krienke S, Nowag A, Wirth S, Wisplinghoff H
Citation: Krienke S, et al. ECCMID 2021. Validation of APAS Independence (CCS) AI-Algorithms to detect MRSA in a routine setup.