Latest News

How The Diagnostic Community Can Integrate AI Into AMR

Contributed Commentary by Theo deVos, Ares Genetics

March 25, 2022 | As the global COVID-19 battle wages on, the ‘silent pandemic’ of antimicrobial resistance (AMR) continues to expand around the world. In 2019 alone, researchers estimated that AMR was associated with 4.95 million deaths globally, with 1.27 million deaths being directly attributed to drug-resistant bacterial or fungal infections. These statistics illustrate the immense magnitude of the AMR crisis currently gripping the world. And as the global health community is gradually able to turn more of its attention previously dedicated to COVID-19 to other health concerns, it is no surprise that the World Health Organization has listed antimicrobial resistance as one of the top ten global health threats facing humanity.

While there are many factors contributing to this crisis, the overuse and misuse of antibiotics throughout all levels of healthcare, compounded by an increasingly limited pipeline of reliable antibiotics to treat common and generally non-life threatening infections are at the top of the list. To mount a serious global fight against AMR will require a comprehensive approach to address these issues, including new technologies that complement antibiotic treatment and support anti-microbial stewardship efforts.

As the medical community continues to evaluate new technology-based approaches to fight AMR, microbial genome sequencing coupled with artificial intelligence (AI) are among the first topics brought up for discussion, and many laboratories are already embracing these new technologies.

How Artificial Intelligence (AI) Can Inform Outbreak Management

With AI and machine learning (ML) driven approaches, laboratories can more quickly analyze the genetic profiles of pathogens. Two of the most developed foci of AI in the AMR lab setting are bioinformatics tools and cloud-based databases. When combined smartly, the technologies allow rapid genome analysis and support sharing of information on key AMR determinants from an increasingly growing number of sources. Ultimately, these strategies allow for rapid interpretation, prediction and annotation of AMR determinants, empowering laboratory and clinical professionals to prioritize combatting AMR and advancing antimicrobial stewardship.

AI-driven cloud-based data platforms allow laboratories to access unprecedented amounts of data including AMR determinants, antimicrobial susceptibility test data (AST) and isolate genome sequences. With the proliferation of sequence data, labs can go even further than just public health databases, incorporating insights from multiple sources. With real-time updates on outbreaks and infectious disease progression, laboratories can better monitor, manage and analyze trends crucial to understanding changes within key pathogens.

In addition to access to key databases, AI-based tools can also provide faster identification of bacterial isolates than traditional cultures of native patient samples, and support antimicrobial stewardship efforts by allowing pharmacists and clinicians to guide optimal antibiotic treatment regimens. And when it comes to combating AMR, speed of actionable information and antimicrobial stewardship are crucial factors. Though this is currently only happening in the laboratory setting, new technology, broader access and education on these tools and processes will prove vital in managing future outbreaks and the growing threat of drug-resistant superbugs.

Obstacles to AI Integration

For the past few years the global health community has been understandably focused on the COVID-19 pandemic. In turn, solutions and tests not directly related to the latest coronavirus have been slow to be addressed and cleared by the FDA. That being said, there have already been precedents set for using research use only (RUO) and lab-developed tests (LDTs) in lieu of their FDA approved counterparts. For example, when it comes to MALDI-TOF for microbial identification, many microbiology labs have chosen to use the RUO database library as it offers more comprehensive coverage of microorganism genera and species compared to the IVD database. The diagnostic community can apply this same approach when it comes to gene panels and databases, incorporating the newer iterations after sufficient testing and validation.

At a more fundamental level, there needs to be greater emphasis placed on spreading awareness of these new technologies, as well as training for healthcare professionals to deploy them effectively. Healthcare workers globally need education and support to understand the impact these tools can have in both individual patient care settings and for larger transmission events and disease patterns.

Looking Ahead

To address the growing AMR challenge, better implementation of existing methods through anti-microbial stewardship, in combination with the developing technologies for AI powered microbial sequence analysis will allow us to prepare for future outbreaks of resistant microbes.  While many of these tools are already available, healthcare professionals are regularly updating and enhancing existing solutions, in addition to creating new ones. The key aspect of AI integration is ensuring that these technologies are future-proof and able to evolve with the changing needs of society. Hospitals and other patient care settings will not be able to incorporate AI solutions if there are no clear roadmaps and processes in place to gradually integrate and educate employees. Routine collaboration and conversation with technical partners, leadership teams, clinicians and all other team members and stakeholders in patient care settings will be vital to ensure technology can be updated and run smoothly.


Theo deVos is Senior Vice President, Corporate Development and Operations at Ares Genetics, an OpGen Group Company, developing DNA-based infectious disease products and services for drug response prediction to improve the way antimicrobial resistant (AMR) infections are diagnosed, prevented or treated. He can be reached at