Single-Cell Nanomotion and Machine Learning for Parallel Bacterial Identification and Antibiotic Screening

Rapid and accurate identification of bacterial infections and their resistance to antibiotics is critical to effective clinical decision-making and combating antimicrobial resistance. However, current diagnostic approaches are typically segmented: techniques such as MALDI-TOF provide species identification, but cannot assess antibiotic susceptibility, while standard antimicrobial susceptibility (AST) tests are time-consuming and lack concurrent identification capability. In this study, we overcome these limitations by integrating single-cell nanomotion detection using graphene drums with machine learning (ML) algorithms to perform both tasks simultaneously within a single measurement. Nanomotion signals, nanoscale vibrations from single living cells, are recorded in real-time and transformed into time-frequency spectrograms, which serve as inputs to ML models trained for robust pattern recognition. Our framework enables the differentiation of Escherichia coli, Staphylococcus aureus, and Klebsiella pneumoniae, while simultaneously distinguishing resistant and susceptible strains with 98% precision. By coupling highly-sensitive graphene nanomotion sensors with advanced ML tools, our approach delivers a label-free bacterial diagnostics, offering both identification and susceptibility profiling at the single-cell level within a couple of hours.