Back

Clinically applicable rapid susceptibility testing of multi-drug resistant Staphylococcus aureus by mass spectrometry and extreme gradient boosting machine

Wang, Z.; Wang, H.-Y.; Pang, Y.; Chung, C.-R.; Horng, J.-T.; Lu, J.-J.; Lee, T.-Y.

2021-10-06 bioinformatics
10.1101/2021.10.05.463151 bioRxiv
Show abstract

Multi-drug resistant Staphylococcus aureus is one of the major causes of severe infections. Due to the delays of conventional antibiotic susceptibility test (AST), most cases were prescribed by experience with a lower recovery rate. Linking a 7-year study of over 20,000 Staphylococcus aureus infected patients, we incorporated mass spectrometry and machine learning technology to predict the susceptibilities of patients for 4 different antibiotics that can enable early antibiotic decisions. The predictive models were externally validated in an independent patient cohort, resulting in an area under the receiver operating characteristic curve of 0.94, 0.90, 0.86, 0.91 and an area under the precision-recall curve of 0.93, 0.87, 0.87, 0.81 for oxacillin (OXA), clindamycin (CLI), erythromycin (ERY) and trimethoprim-sulfamethoxazole (SXT), respectively. Moreover, our pipeline provides AST 24-36 h faster than standard workflows, reduction of inappropriate antibiotic usage with preclinical prediction, and demonstrates the potential of combining mass spectrometry with machine learning (ML) to assist early and accurate prescription. Therapies to individual patients could be tailored in the process of precision medicine.

Matching journals

The top 8 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.