Machine learning improves clinical decision-making to fight antimicrobial resistance


A lab technician holds a tube containing a swab sample taken for the Covid-19 serological test at the Leumit Health Services laboratory in the Israeli town of Or Yehuda in this file photo taken July 16, 2020 – Copyright AFP Rostislav NETISOV

The spread of antimicrobial resistant organisms, and the extension of the number of specific species resistant to a wider range of antimicrobials, continues to pose a considerable threat to the hospital environment. Many species are nosocomial infectious agents, increasingly difficult to treat, and posing a particular threat to immunocompromised patients. Several different genes confer resistance to a given antimicrobial agent.

Understanding regional variations in antimicrobial resistance has a double importance. First, it allows scientists to understand the spread of resistance and warn about the loss of effectiveness of a particular agent for a given bacterial species. Second, it helps healthcare professionals decide which antimicrobial to administer to the patient. Often there is little time to characterize infectious species to determine the optimal antimicrobial. By understanding resistance patterns in the community, some antimicrobials may be preferential over others at the local level.

One way to advance region-focused understanding of antibiotic resistance patterns is to use machine learning to make computational predictions. This form of artificial intelligence provides an algorithm, with the ability to predict certain outcomes through a model learned by providing a large amount of experimental data. Part of this data is training data, used to increase the success rate of the predictions made. Once the cross-validation score (“training set”) has reached an acceptable level, real-world clinical data can be reviewed (“testing set”).

Such analyzes can reveal hitherto hidden determinants of antimicrobial resistance by examining metagenomic datasets, environmental microbiome datasets, and their pathogenic potential in humans. Additionally, the pairing of machine learning algorithms and laboratory testing can help accelerate the discovery of new antimicrobials. This final step involves computer-assisted prospecting to align new drugs with alternative mechanisms of antimicrobial action (so-called ‘synergistic drug combinations‘).

Machine learning approaches include logistic regression (LR), support vector machine (SVM), random forest (RF), and convolutional neural network (CNN).

Machine learning algorithms are able to correlate genomic variations with phenotypes and search for resistance patterns against given agents in regions. The examination of these databases involves an algorithm using Boolean functions of conjunction (logical AND) or disjunction (logical OR).

Researchers based at the Jeffrey Cheah Biomedical Centre, Wellcome-MRC Cambridge Stem Cell Institute, University of Cambridge, UK, developed such a model and the results were published in the journal Microbiota. The research undertakes predictive analysis at the very local niche level of the International Space Station (“Machine Learning Algorithm to Characterize Antimicrobial Resistance Associated with the Surface Microbiome of the International Space Station”).


Comments are closed.