Microbial Classification by Multimodal Atomic Force Microscopy with Machine Learning

Abstract

The development of high-purity sterile environments is essential in healthcare, and the maintenance of environments. Monitoring microbial conditions is a critical process during disinfection, especially for highly resistant endospores, which can induce re-contamination under suitable triggers. Traditional optical-based microbial detection techniques struggle with resolution and effectiveness and lack the multi-parametric information. We introduce automated AFM with multimodal ML integration to enhance the accuracy and efficiency of microbial detection. This approach leverages detailed mechanical data from AFM, coupled with the analytical capabilities of ML, to provide a more robust method for identifying microorganisms in complex environments. Overall, the work aims to improve the efficiency and accuracy of AFM through robotic technology and artificial intelligence, and freeing researchers from the labor of time-consuming measurements and data analysis.