Rapid identification of MRSA using mass spectrometry and machine learning from over 20000 clinical isolates

<p>Rapidly&nbsp;identifying&nbsp;methicillin-resistant&nbsp;Staphylococcus&nbsp;aureus&nbsp;(MRSA)&nbsp;with&nbsp;high&nbsp;integration&nbsp;in&nbsp;the&nbsp;current&nbsp;workflow&nbsp;is&nbsp;critical&nbsp;in&nbsp;clinical&nbsp;practices.&nbsp;We&nbsp;proposed&nbsp;a MALDI-TOF MS based&nbsp;machine&nbsp;learning&nbsp;model&nbsp;for rapidly MRSA prediction,&nbsp;the model&nbsp;was&nbsp;evaluated&nbsp;on&nbsp;a&nbsp;prospective&nbsp;test&nbsp;and&nbsp;four&nbsp;external&nbsp;clinical&nbsp;sites.&nbsp;On&nbsp;the&nbsp;dataset&nbsp;comprising&nbsp;20359&nbsp;clinical&nbsp;isolates,&nbsp;the&nbsp;area&nbsp;under&nbsp;the&nbsp;receiver&nbsp;operating&nbsp;curve&nbsp;of&nbsp;the&nbsp;classification&nbsp;model&nbsp;was&nbsp;0.78&ndash;0.88. Our MALDI&ndash;TOF MS-based ML model for the rapid&nbsp;MRSA&nbsp;identification can&nbsp;be&nbsp;easily&nbsp;integrated&nbsp;into&nbsp;the&nbsp;current&nbsp;clinical&nbsp;workflows&nbsp;and can further support&nbsp;physicians&nbsp;prescribe&nbsp;proper&nbsp;antibiotic&nbsp;treatments.&nbsp;</p>