Speaker
Description
Bacterial infections are among the diseases with high morbidity and mortality rates worldwide, posing a significant threat to global public health. There is an urgent need to develop precise and rapid diagnostic methods for bacterial infections to enable personalized medication and treatment for infected patients, promptly save lives, and reduce the spread of antimicrobial resistance. Bacterial infection diagnosis encompasses two key aspects: bacterial identification and antibiotic susceptibility testing (AST). Current clinical methods for bacterial identification and AST are limited by the time-consuming process of bacterial culture. Bacterial identification is typically performed using MALDI-TOF MS, while AST is conducted with automated biochemical analyzers, requiring an additional step of proliferation testing under antibiotic stimulation, resulting in a delay of 6~24 hours compared to bacterial identification.
To accelerate antibiotic susceptibility testing (AST) and reduce costs, we have developed two rapid AST methods based on MALDI-TOF MS. The first method detects deuterium incorporation into newly synthesized proteins under antibiotic stimulation, allowing for monitoring of protein synthesis and using machine learning to predict bacterial susceptibility. This approach introduces a series of discriminative features, resulting from mass shifts induced by deuterium incorporation, which significantly enhances the performance of machine learning models, especially on small datasets. Additionally, when transferring training results from public datasets to smaller datasets, this method improves the accuracy of antibiotic susceptibility predictions. The second method monitors changes in bacterial metabolites under short-term antibiotic stimulation and, when combined with machine learning, also predicts antimicrobial susceptibility. Both methods reduce AST time to just 0.5 to 1 hour after bacterial identification by MALDI MS. Furthermore, these approaches integrate both AST and bacterial identification into a single mass spectrometer, facilitating faster diagnoses, reducing equipment and labor costs, and demonstrating great potential for broader clinical applications of mass spectrometry.
| User consent | yes |
|---|
