How AI Is Helping Researchers Fight Superbugs Faster Than Ever Before

Physicians and scientists have been warning us of the dangers of antibiotic resistance since penicillin became widely available—and widely misused—during and immediately following World War II. In the 78 years since, things have only gotten worse. The most recent Global Research on Antimicrobial Resistance (GRAM) report estimated deaths and disabilities linked to 23 treatment-resistant pathogens and 88 pathogen-drug combinations.¹ That report wasn’t inclusive of every treatment-resistant disease out there, and critically pointed to a disturbing trend: An increase in the number of superbugs and no current treatments available to treat patients for them.

When treatments for infections are misused (eg, a patient doesn’t finish a course of antibiotics or self-medicates a parasitic infection with anti-fungal treatments, or a viral infection is treated with antibiotics) pathogens learn to recognize the medicine and mutate to ensure their own survival, rendering available treatments less effective, and in many cases, useless. As a result, treatment-resistant infections are linked to more than 5 million deaths globally each year.² In an industry where it can take 10 to 12 years for a novel therapeutic—that pathogens haven’t yet met—to be approved for use, tiny pathogens become a big problem very fast. Artificial intelligence (AI) has already proven to exponentially speed up the work done by researchers without sacrificing, and in many cases increasing, accuracy. If applied correctly to drug development, AI could significantly speed up drug discovery and development, and help researchers end the treatment-resistant infection crisis before it’s too late.

As a result, treatment-resistant infections are linked to more than 5 million deaths globally each year. Image Credit: © nobeastsofierce - stock.adobe.com

If Superbugs Are so Small, Why Are They Such a Big Problem?

Most people equate “superbug” with Methicillin-resistant Staphylococcus aureus (MRSA), which is a treatment-resistant bacteria commonly linked to living in crowded or unclean places, sharing personal hygiene items, or having open wounds tended during an extended hospital stay. But while the spotlight often shines on infections like MRSA, there are dozens of superbugs lurking in the shadows.

In June 2023, the World Health Organization (WHO) outlined 40 research priorities on antimicrobial resistance (AMR) to address growing global concerns over drug-resistant fungi and bacteria, with a focus on Mycobacterium tuberculosis—or bacterial tuberculosis, which killed 1.6 million people in 2021 and is the 13th leading cause of death globally—and Candida auris (C auris), which is a fungal yeast infections that can kill 1 in 3 patients.³ 

The CDC has also been tightly tracking C auris since outbreaks were first reported in 2016. The increasing rate of infection—from 173 clinical cases in 2017 to 2377 in 2022—combined with the fungi’s ever-growing resistance to echinocandins, the treatment most prescribed for it, has the medical community rightly on edge. And that’s just one disease. The CDC is tracking a total of 18 AMR germs right now, which will infect more than 2.8 million Americans and result in more than 35,000 deaths, before the year is out.

Critically, these reports and guidelines only account for bacteria and fungi. They do not include the alarming increase in treatment-resistant parasitic or viral infections, including certain strains of giardia, a parasite that causes diarrhea, or hepatitis C, a virus that causes inflammation of the liver.

Researchers and Their AI Tools Are Closing in on a Solution

Between 2017 and 2021, the FDA approved just one new antibiotic that could be used to treat some of the world’s most serious superbugs, like Enterobacteriaceae and Pseudomonas aeruginosa, 2 bacteria that often result in serious to critical infections in multiple body systems including the brain, spinal cord, lungs, blood, and urinary tract. As of this year, there are just 27 antibiotics in clinical trials.⁴ But only about 10% of candidate drugs ever make it to market, and if that statistic holds true, then without help, only 3 or fewer of these drug candidates will reach patients a decade from now, and these drugs will only help patients with certain bacterial infections.

Advancements in drug development using AI, however, could help to not only speed up the process and get life-saving treatments into the hands of patients, but also ensure that more useful drug candidates make it through the approval process. Recently, my research team and I applied AI to develop safe and effective bacteriophage therapies targeting the microbiome and pathogenic bacteria. Phages are viruses programmed to seek out and destroy bacteria and are already proven to effectively fight treatment-resistant bacteria. AI was applied to the development of this potential new therapy to help more quickly identify which phages would work, and which wouldn’t, significantly reducing time spent in research and discovery. Importantly, our AI-driven BIOiSIM platform can also test treatments and compounds against real human data to find out whether they’ll be successful and safe before developers ever give the treatment to a patient or clinical trial participant.

When it Comes to Beating Superbugs, AI Has an Overlooked Superpower

Researchers have recently found that AI can support the push for personalized medicine. This means that physicians could personalize treatments for patients infected with AMR pathogens and find the right treatment without the guesswork.

We’ve already put this type of personalized medicine into practice. Earlier this year, researchers completed a study on the use of AI to develop patient-specific therapy regimens for the treatment of COVID-19.5 Although the virus is currently treatable, like every other pathogen, SARS-CoV2 can mutate, and may eventually develop a resistance to antiviral medications, particularly if they’re being misused.

We used AI to analyze large-scale omics and clinical datasets for patient stratification and were able to develop a machine learning model that both demonstrates that clinical features are enough of an indicator for COVID-19 severity and survival, and also demonstrates which clinical features have a more significant impact. Ultimately, we were able to create a guide for physicians to help prioritize specific treatments based on patients’ clinical features.

The goal of AI in fighting superbugs is 2-fold. The first goal is to help physicians treat infections and disease successfully and without the risk of more pathogens becoming treatment resistant. The second goal is to reduce the amount of time it takes for novel, life-saving treatments to get to patients so that they can battle the superbugs that are already here. We’re not yet at the end of the antibiotic—antiparasitic or antiviral—era, as some may believe; in fact, with AI on our side, we’re just at the beginning.

About the Author

Jo Varshney is the CEO and founder of VeriSIM.

References

1. Global Research on Antimicrobial Resistance. Global burden of bacterial antimicrobial resistance in 2019: a systematic analysis [Report]. February 2, 2022. Accessed November 7, 2023. doi:10.1016/S0140-6736(21)02724-0
2. Centers for Disease Control and Prevention. National infection and death estimates for antimicrobial resistance. December 13, 2021.Accessed November 7, 2023. https://www.cdc.gov/drugresistance/national-estimates.html
3. World Health Organization. Global research agenda for antimicrobial resistance in human health [Report]. June 22, 2023. Accessed November 7, 2023. https://www.who.int/publications/m/item/global-research-agenda-for-antimicrobial-resistance-in-human-health
4. European Society of Clinical Microbiology and Infectious Diseases. Not enough new antibiotics in the pipeline, concludes WHO review— especially those targeting deadly drug-resistant microbes [Press Release]. March 15, 2023. Accessed November 7, 2023. https://www.eurekalert.org/news-releases/982758
5. Bello, B. et al. Integrating AI/ML Models for Patient Stratification Leveraging Omics Dataset and Clinical Biomarkers from COVID-19 Patients: A Promising Approach to Personalized Medicine, Int J Mol Sci. 2023;24(7):6250. doi:10.3390/ijms24076250