For decades, the term “incurable disease” was considered a cold, impenetrable and definitive wall in the medical sciences. In the room where the word ‘incurable’ enters, ‘hope’ disappears through the window.
But as humanity moves through 2026, that wall will finally crack under the massive processing power of artificial intelligence (AI). Ultimately, hope blooms through the destroyed cracks of the wall.
In today’s AI-powered world, researchers are not just treating symptoms; they rewrite the rules of medical science.
From ‘folding’ proteins in seconds, to designing tailor-made treatments for rare cancers, to combating antibiotic resistance, AI is transforming the health laboratory from a place of trial and error into the realm of ‘bio-digital assurance’.
Accelerating the discovery of drugs to combat ‘superbugs’
AI tools are breaking a decades-long impasse in drug discovery, especially to combat antibiotic-resistant superbugs.
Scientists have struggled for years with the slow and expensive traditional drug development process.
According to the BBC, only twelve new antibiotics were approved for use between 2017 and 2022. Worst of all, most antibiotics have remained ineffective against bacterial resistance.
Antibiotic resistance superbugs are responsible for 5 million deaths per year worldwide. It is also estimated that these infections could cause more than A$2.5 trillion in economic disruption globally by 2050.
Here comes the breakthrough role of AI models, which now screen billions of compounds in just a few days and identify entirely new chemical structures.
Researchers at the Massachusetts Institute of Technology (MIT) have designed two new compounds using the AI. According to Collin from MIT, these compounds could be very effective against highly resistant infections, gonorrhea (Neisseria gonorrhoeae) and methicillin resistant (Staphylococcus aureus)MRSA.
AI can also perform ‘freestyle’ molecular design or build on existing molecules, reducing the millions of possibilities into hours rather than years.
Given its unprecedented efficiency, the generative AI tool that Collin and his team used screened more than 45 million different chemical structures for their ability to target Neisseria gonorrhoeae and Staphylococcus aureus.
Previously, AI was also used to discover new and powerful antibiotic compounds Clostridium difficilea common intestinal infection, and Mycobacterium tuberculosisthat causes tuberculosis.
Focused on Parkinson’s disease
Despite 200 years of research, medical researchers have been unable to find an effective treatment that slows the progression of neurodegenerative diseases.
10 million people worldwide have developed Parkinson’s disease. In the US, up to 1 million people live with this disease.
Michele Vendruscolo, professor of biophysics and co-director of the Center for Misfolding Diseases at the University of Cambridge in Great Britain. “There are endless debates about the origins of the condition. If you go to a Parkinson’s conference you will hear dozens of different hypotheses, all of which are being actively investigated.”
In 2024, AI came in handy. Vendruscolo and his colleagues from the University of Cambridge used machine learning, a form of AI, to target Lewy bodies (misfolded protein clumps) that play a role in the early stages of neurodegeneration in Parkinson’s patients.
Scientists are now using AI to find molecules that stabilize proteins before they misfold so they can prevent the disease entirely.
Using the AI, they use AI to screen billions of molecules for a few thousand pounds in just a few days.
The compound proposed by AI proved to be more effective than conventional approaches when tested in the laboratory. Researchers are hopeful that AI could one day help stop Parkinson’s disease completely.
“If we can stabilize the proteins in this form by binding to them, we have prevented Parkinson’s disease – which is better than cure,” Vendruscolo said.
Repurposing existing medicines
David Fajgenbaum, associate professor of medicine at the University of Pennsylvania in the US, has shown that many thousands of already approved drugs can treat diseases for which they were not designed. He saved his own life using a transplant drug to treat Castleman’s disease.
New AI models are being used to ‘map’ the world’s 8,000 approved drugs against 17,000 diseases.
Notable examples include the rare chromosomal disorder Pitt-Hopkins syndrome, the rare inflammatory disease sarcoidosis, and a rare kidney cancer, where AI is being used to repurpose existing treatments.
Simulating disease progression using AI
At McGill University, researchers created a “virtual disease system” for idiopathic pulmonary fibrosis (IPF), a progressive lung disease.
By sequencing lung cells at different stages, AI simulates how a cell goes from healthy to diseased. This approach allows scientists to test the virtual effects of drugs on a subject.
Folding protein
Scientists have spent decades mapping a single 3D protein structure. In 2022, AlphaFold 2, an AI tool developed by DeepMind, solved this protein mystery by predicting the 3D shapes of almost all 200 million known proteins. It was largely limited to the proteins themselves.
AlphaFold 3 is now able to model DNA, RNA and ligands. It also predicts how a drug binds to a specific protein pocket.
Limits
Despite the AI revolution in healthcare, certain challenges remain:
Much of the crucial data on drug absorption and toxicity is hidden in biotech and pharmaceutical companies and not publicly available.
AI currently excels in the initial screening and target identification phases, but drugs still face years of mandatory, slow-moving human clinical trials.
“AI is revolutionizing drug discovery, but only in very specific ways,” says Vendruscolo.
