Mayo Clinic’s researchers develop an AI-powered solution to accelerate heart failure drug discovery

Heart failure is one of medicine's biggest puzzles — and it's not getting easier. More than 6 million Americans live with it, and the toll is staggering: frequent hospital stays, reduced quality of life, and far too many deaths. Even after decades of research, doctors still have too few weapons in their arsenal.

Enter Mayo Clinic researchers, who've built a novel way to fast-track drug discovery. Their approach? "Virtual clinical trials." By mixing advanced computer modeling with the messy but rich world of real-world patient data, they've found a way to predict which existing drugs might work for heart failure — before spending billions on human trials.

"We've shown that with our framework, we can predict the clinical effect of a drug without a randomized controlled trial," says Nansu Zong, Ph.D., lead author of the study published in npj Digital Medicine. "We can say with high confidence if a drug is likely to succeed or not."

How does it work?

Mayo's method is part science fiction, part data science. The team combined two heavyweight tools:

• Computer models that map how drugs interact with biological systems
• Electronic health records (EHRs) from nearly 60,000 heart-failure patients

With those, they created trial emulations — virtual stand-ins for traditional randomized trials. Instead of recruiting participants, they used existing patient data to build comparison groups and track outcomes, like changes in key biomarkers that signal heart failure progression.

To boost accuracy, they incorporated drug-target modeling, utilizing AI to link chemical structures to protein sequences and genes. This bridged the gap between lab predictions and real-world outcomes.

Then came the test: 17 drugs, previously examined in 226 Phase 3 trials. The virtual trials nailed the direction of the results — predicting which drugs helped and which didn't.

"This model has the potential to guide drug development pipelines at scale," says Dr. Zong. "Right now, it can tell us the direction of efficacy — whether a drug will be beneficial — but not yet the level of that effect. That's our next step."

Why does it matter?

Drug development is a marathon, not a sprint. Bringing one new therapy to market often takes over a decade and costs north of $1 billion. Most fail along the way.

Repurposing existing drugs is the shortcut we've all been waiting for — their safety profiles are already known, so researchers can jump straight to testing benefits for new uses. But figuring out which ones to bet on? That's the tricky part.

This framework lets researchers:

• Identify winners faster — so patients see new treatments sooner
• Cut costs — fewer dead-end trials
• Sharpen focus — concentrate resources where the odds of success are highest

Faster progress here could mean fewer hospitalizations, better survival rates, and lighter burdens on healthcare systems.

The context

This is more than just one cool research paper. Mayo Clinic is turning the approach into a full-blown initiative led by Cui Tao, Ph.D., vice president of Mayo Clinic Platform Informatics.

They're exploring three flavors of virtual research:

• Trial emulation — replicating past or hypothetical trials using real-world data
• Trial simulation — running "what-if" scenarios to see how a drug might fare in new populations
• Synthetic trials — blending real-world and modeled patient data to replace or supplement trial arms

These efforts align with Mayo's broader strategy, encompassing Precure (risk prediction and prevention) and Genesis (intelligent transplant care and personalized interventions). It's a signal that virtual trials might soon move from the research sidelines to the clinical mainstream — changing not just heart failure research, but drug discovery as a whole.