AI can now beat superbugs: Tool would suggest alternative treatment options for antimicrobial-resistant bacteria

When 21-year-old Sudhanshu Kar (name changed) suffered from an unrelenting bout of cough followed by persistent fever, his family initially thought it was a routine infection. But after weeks of no improvement, and multiple courses of antibiotics that failed, the doctors confirmed his worst fear—a drug-resistant strain of Mycobacterium tuberculosis. 

 

Resistance to drugs meant that treatment options were fast running out; that too for a disease like TB which was otherwise completely curable.

 

Stories like Kar's are becoming alarmingly common across India and other low- and middle-income countries, where antimicrobial resistance (AMR) is growing at a terrifying pace. 

 

When bacteria stop responding to antibiotics that once worked, even minor infections can spiral into life-threatening conditions.

 

Now, a pioneering collaboration between Indian and French researchers might turn the tide in this battle. A team led by Dr Emilie Chouzenoux of Inria Saclay and Dr Angshul Majumdar of IIIT-Delhi has developed an artificial intelligence (AI) tool that can recommend alternative treatment options when standard antibiotics fail.

 

“This isn’t just about technology—it’s about giving patients a fighting chance,” says Dr Majumdar.

 

The new algorithm learns from real-world clinical data and intelligently suggests existing antibiotics that might work—even when first-line treatments don’t.

 

Doctors are turning to drug repositioning, that is, using old drugs in new ways, especially since developing a new drug involves both, a long time and high cost.

 

But picking the right drug is often a shot in the dark. This is where the new AI model comes in. It blends hospital treatment records from leading Indian institutions with molecular data about bacterial genomes and the chemical properties of antibiotics.

 

“It’s like giving the AI a doctor’s intuition and a scientist’s toolkit, all at once,” says research engineer Stuti Jain, who was part of the development team along with graduate students Kriti Kumar and Sayantika Chatterjee.

 

To see if it really works, the team tested the model on three dangerous, drug-resistant bacteria: Klebsiella Pneumoniae, a common cause of hospital-acquired pneumonia and sepsis; Neisseria Gonorrhoeae, which causes gonorrhea and is rapidly resisting standard treatments; Mycobacterium Tuberculosis, still one of India’s deadliest infectious diseases.

 

In each case, the AI successfully identified antibiotics that were either known to work or showed strong potential based on data patterns. These suggestions were cross-checked with known resistance profiles and validated by experts.

 

“It’s not magic—it’s a method,” says Dr Chouzenoux. “The power lies in how we’re integrating data from the lab and the clinic.”

 

This could help doctors choose effective treatments faster. In public health programs, it could guide policy on antibiotic use and prevent the spread of resistant bacteria.

 

For patients, it could mean, more targeted treatment, and a better chance of recovery.

 

“We’ve seen too many families lose loved ones because they didn’t get the right antibiotic in time,” says Dr Majumdar. “This tool helps change that.”

Health