AI Cracks Rare Disease Mysteries in 29 Undiagnosed Cases

When Helene and Mikk Cederroth lost three of their four children to an undiagnosed disease, they refused to let other families face the same heartbreak alone.

Last September, their foundation brought over 100 researchers together for an urgent mission: crack 29 medical mysteries that traditional testing couldn't solve. Their secret weapon? AlphaGenome, Google DeepMind's new artificial intelligence that can read the "dark matter" of human DNA.

Most genetic testing focuses on the 2% of DNA that codes for proteins. But AlphaGenome tackles the other 98%, the non-coding sequences where mutations hide like needles in a cosmic haystack. These hidden variants often cause devastating diseases that leave families desperate for answers.

Dr. Eric Klee from the Mayo Clinic tested AlphaGenome before the hackathon on a real case. A patient's mystery illness seemed connected to a genetic variant, and lab work showed it affected heart cells but not brain cells, matching the symptoms perfectly. AlphaGenome's predictions confirmed what the experiments revealed.

The Cederroth family founded the Wilhelm Foundation, named after their eldest son who died at 16. They organized three hackathons across the globe, refusing to accept that 350 million people with rare diseases should be "left behind" without diagnoses.

Why This Inspires

Traditional genome sequencing helps only a fraction of rare disease patients get answers. Now AI is opening doors that seemed permanently locked, giving researchers new ways to interpret the vast stretches of DNA that were once impossible to understand.

The tool became available to scientists last year, democratizing access to technology that could transform rare disease diagnosis. What once took years of experimental work can now be predicted and verified faster, potentially cutting the agonizing wait families endure.

Each diagnosis means families can finally understand what's happening, connect with others facing the same condition, and sometimes find treatments. For parents like the Cederroths, it means other children might get the answers their own couldn't.

The hackathon approach brings urgency to problems that often languish in research labs for decades. When scientists volunteer to "lock themselves in a room" for three days straight, they're racing against time for real families waiting for hope.

This technology doesn't replace human expertise but amplifies it, letting researchers focus on the hardest puzzles while AI handles the massive computational work of predicting genetic effects across billions of DNA sequences.

The work continues as more researchers gain access to AlphaGenome and apply it to thousands of unsolved cases worldwide.