Barretts esophagus books 2

Publications

Getting Faster & Smarter: AI and Automation Improve Accuracy for MRSA Screening

Surveillance for key pathogens of epidemiological significance is important for infection prevention in hospital settings. At the Johns Hopkins Hospital and its affiliates, anterior nares cultures are routinely obtained from patients admitted to intensive care units and oncology and plated on chromogenic agars for the detection of Staphylococcus aureus and methicillin resistant S. aureus (MRSA). This publication describes a prospective study of an automated imaging station (APAS Independence, LBT Innovations, Adelaide AU) linked with interpretive software. The software uses artificial intelligence (AI) to analyze the images and presumptively designates the cultures on specific solid agar media as positive or negative and sorts them into those categories. Over a 5-month period, 5, 913 samples obtained from 3,847 patients were cultured onto CHROMagar MRSA and were read in parallel manually and by the APAS Independence instrument. Likewise 774, S. aureus chromagar plates were inoculated and included in the study. The study demonstrated high sensitivity (100%; 96.5%) and negative predictive values (100%, 99.0%) of digital image analysis for MRSA and S. aureus, respectively, such that negative cultures can be reliably reported without technologist intervention. Five low-level positive cultures, missed by manual reading, were detected by the APAS Independence. Because of high false-positive rates, a technologist is still required to verify the positive results. This is the first publication evaluating the accuracy of the APAS MRSA analysis module on BD BBL CHROMagar and subsequent application to both MRSA and S. aureus CHROMagar. Data from this study supported the manufacturer’s 510 K submission for FDA clearance. In a high-throughput workplace, triaging samples in this manner consolidates the workload and reduces the need for technologist intervention on samples where critical thinking is not required, i.e., negative and no-growth plates. Advanced automation incorporating AI (“smart incubation”) is making its way to clinical microbiology laboratories and will improve laboratory efficiency.

View Full Article