How Can Artificial Intelligence Stop the Spread of Antimicrobial Resistance?

For many years, before most of us had even heard of a coronavirus, the main biological event that was thought to pose a global risk to humanity was the spread of antimicrobial resistance (AMR) among bacteria and other microbes.

As we begin to see the light at the end of a long year living with the COVID-19 pandemic, the threat of AMR has not gone away and remains a priority. One voice sounding the alarm is that of precision medicine company OpGen, who use molecular diagnostics and informatics to identify AMR before it can spread. To find out more, Technology Networks caught up with OpGen’s CEO Oliver Schacht.

Ruairi Mackenzie (RM): What are the most significant factors in the spread of AMR?

Oliver Schacht (OS):
 The main cause of antimicrobial resistance (AMR) is antibiotic use. Overly lenient antibiotic stewardship practices, including over-prescription, is one of the biggest factors which can contribute to the increase and spread of AMR. The more antibiotics are used, the more bacteria can become resistant to them. Infection control practices are stricter than ever in healthcare facilities but there are still many factors in hospital, critical care and nursing facility settings that can lead to the spread and transmission of drug-resistant infections, particularly among high-risk patient groups with pre-existing conditions. Invasive medical devices such as catheters and ventilators also heighten infection susceptibility in these populations. Frequent empirical use of broad-spectrum antibiotics, i.e., use without prior diagnosis of the underlying pathogen and its resistance profile, makes patients more vulnerable to suffering and increases AMR.

RM: What are the limitations of currently available AMR detection protocols?

OS: 
While detection of AMR present in bacteria is routinely performed by conventional phenotypic culture-based methods, molecular methods have become an integral part of clinical practice and they are widely used in clinical microbiology laboratories due to the greater speed and accuracy they provide. Phenotypic analysis is too time-consuming and results may be inconclusive or simply unavailable. This leads to prolonged use of empirical antibiotics and delays in targeted antibiotic therapy, consequently worsening the outcomes for the patient and increasing AMR.

Further contributing to the faster time-to-result of molecular methods is the fact that they are performed directly from native specimens, and not reliant on growth in culture.

More good news is that next-generation, rapid multiplex molecular panels are showing comparable clinical sensitivity and specificity when compared to conventional culture methods. For instance, a recent multi-center study published in the Journal of Clinical Microbiology looked at the performance of the Unyvero rapid molecular lower respiratory tract panel in bronchoalveolar lavage (BAL) specimens against standard of care microbiological testing and found Unyvero to have an overall high negative predictive value of 97.2% for pathogen detection on a per sample basis.

Another study found the Unyvero LRT panel capable of detecting 
27% more pathogens than traditional culture, including clinically important pathogens such as K. pneumoniaeH. influenzaeAcinetobacter and S. maltophilia. Turnaround time for results was also reduced to just 5.2 hours for the LRT panel, compared to 2.8 days for the culture method.

These sample-to-answer multiplex molecular panels enable rapid support to ensure that optimal treatment or control strategies can be undertaken in a timely manner.

RM: What artificial intelligence (AI) techniques does OpGen use to improve AMR analysis and why are they useful in this context?

OS: 
OpGen is embracing cloud-based, data-sharing software that enables better insights on the progression of bacteria and viruses in real time. This includes our ARESdb tools, which leverages AI-powered prediction algorithms to predict AMR and antibiotic susceptibility (AST).

With large databases of phenotypic and genotypic profiles available, machine learning and AI tools can help healthcare providers gain deeper, actionable insights to identify both pathogens and potential treatment courses.

RM: Antibiotic development has been largely ignored by pharma and biotech, despite the risk of a global catastrophe if AMR becomes widespread. Will COVID-19 change that?

OS: 
We’re hopeful that recent reports indicating the impact of COVID-19 treatment on prevalence of drug-resistant infections will shed light on what might become the next global pandemic, if left unchecked. For instance, one study found that, even without microbiological confirmation of a bacterial infection, 72% of ICU COVID-19 patients were administered antibiotics, while only 7-14% actually had a confirmed bacterial co-infection.

Adequately addressing this threat requires not only increased research and investment in new antibiotics – it also calls for cutting edge molecular diagnostics solutions that can simplify and facilitate detection of drug-resistant pathogens, enabling smarter use of available antibiotics. Further, cloud-based data sharing platforms add another layer of value to healthcare professionals and epidemiologists, delivering outbreak insights and informing decision-making.

 

Original post: https://www.technologynetworks.com/informatics/blog/how-can-artificial-intelligence-stop-the-spread-of-antimicrobial-resistance-346588

Leave a Reply

Your email address will not be published. Required fields are marked *