AI model helps some patients get diagnoses after years of uncertainty: Study

An artificial intelligence (AI) model is helping some patients get diagnoses after years of unexplained illness, according to a new study.

Researchers from OpenAI and Boston Children's Hospital took the existing genetic data of 18 pediatric patients -- many now adults -- and reviewed it through a newly developed AI model, cracking cases that stumped doctors for years.

The team hopes its model will help thousands of American kids who are impacted. Researchers pointed out one in 10 Americans -- more than 30 million people, half of whom are children -- have a rare disease.

The researchers believe their models may help diagnose the rarest diseases and also expedite diagnoses.

The study was published Thursday in the New England Journal of Medicine.

For 28-year-old Kyra, one of the patients in the study, the ultra-rare diagnosis of Myofibrillar Myopathy (MFM) came after nearly two decades of uncertainty. MFM is a group of genetic disorders that causes progressive muscle weakness.

"It felt very surreal at the time because I just didn't expect to get an answer in my lifetime, and I think my family didn't expect it either," Kyra told ABC News.

Even though the condition is currently incurable, "we finally got that clarity and closure," she said. "At the very least it's nice to have a name."

The study does not suggest AI can replace doctors or genetic specialists. While the model proposed possible answers, experts made the final diagnosis and each diagnosis was confirmed by a certified clinical lab before families were told.

That distinction is important because AI tools can make mistakes and misread information, the researchers said.

In this study, the AI served more as an extra set of eyes for specialists, helping them sift through large amounts of complex information in about six to 10 minutes per case, according to researchers.

Catherine Brownstein, one of the study's lead researchers and a research associate in the division of genetics and genomics at Boston Children's Hospital, told ABC News that by utilizing AI, "we can apply our human time to more specific things, like reviewing the data, rather than going down rabbit holes chasing things that might be possibilities for a diagnosis."

Kyra, the patient, said she sees the promise of AI in cases like hers, but also believes it needs careful oversight.

"I think for the purpose of assisting researchers in their efforts, especially when they're so complicated and complex, like this situation in this study, I think it can be a very useful tool," she said. "But at the same time, I think we do have to be very cautious."

According to Brownstein, privacy protections are central to the use of this kind of technology.

"We're not removing any human guardrails here," Brownstein said. "A human has to review everything that the AI does."

The study also shows why old genetic test results may be worth revisiting. Because science changes quickly, a result that did not make sense years ago may become clearer as researchers discover new genes, improve how they search genomic data and learn more about how genetic changes affect health.

"A negative genetic test that's negative right now might not be negative in the future," Brownstein told ABC News.

Brownstein said the rapid pace of genetic discovery can make it hard for clinicians to keep rechecking old unsolved cases, but AI may help.

"The genome is being decoded more every day," she said. "But AI is really, really good at that."

In Kyra's case, only 60 patient cases of MFM have ever been published, according to CureMFM13, a charitable organization dedicated to addressing the challenges of the disease.

There were limits to the study, the team said. Researchers looked back at existing cases, so the study cannot prove the tool would work the same way in real time.

Additionally, the number of new diagnoses was small, and the study did not measure whether the AI tool saved time, lowered costs or changed patients' care.

The next step, the authors wrote, is to test the approach in larger forward-looking studies across multiple medical centers.

For Kyra, the diagnosis did not erase the years of uncertainty, but she said it gave her and her family things they did not have before: a name, a sense of closure and a connection to others living with the same rare condition.

She said that while AI may help researchers find answers faster, the human side of medicine still matters most.

"When it comes to health matters that really change your life, you kind of want that human touch present," Kyra said. "You want to feel like people care, and that they listen to you, and you're not just a condition."

Joshua Anthony, MD, MBA, is a psychiatry resident at Creedmoor Psychiatric Center and a member of the ABC News Medical Unit.