AI model helps some patients get diagnoses after years of uncertainty, study finds

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

Researchers from OpenAI and Boston Children’s Hospital took existing genetic data from 18 pediatric patients, most of whom are now adults, and examined them through a newly developed artificial intelligence model, solving cases that baffled doctors for years.

The team hopes their model will help thousands of affected American children. Researchers noted that one in every 10 Americans (more than 30 million people, half of whom are children) has a rare disease.

Americans increasingly use AI bots even to learn about their own health: Surveys

The researchers believe their model could help diagnose the rarest diseases and also speed up diagnoses.

The study was published on Thursday. New England Journal of Medicine.

One of the study patients, 28-year-old Kyra, was diagnosed with the extremely rare Myofibrillar Myopathy (MFM) after nearly two decades of uncertainty. MFM is a group of genetic diseases that cause progressive muscle weakness.

“It felt so surreal to me at the time because I never expected to hear back for the life of me, and I don’t think my family expected it either,” Kyra told ABC News.

Although the condition is currently incurable, “we have finally come to this clarity and conclusion,” he said. “It’s nice to at least have a name.”

The study does not suggest that artificial intelligence can replace doctors or geneticists. While the model suggested possible answers, experts made the final diagnosis, and each diagnosis was confirmed by a certified clinical laboratory before being told to families.

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

In this study, artificial intelligence served as an extra set of eyes for experts, helping them review large amounts of complex information in about six to 10 minutes per case, according to the researchers.

Doctors view AI authors increasingly positively, but problems remain

Catherine Brownstein, one of the study’s principal investigators and a research assistant professor in the department of genetics and genomics at Boston Children’s Hospital, told ABC News that by using artificial intelligence, “we can spend our human time on more specific things like reviewing data rather than going down rabbit holes chasing things that might be diagnostic possibilities.”

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

“I think this can be a very useful tool to assist researchers in their efforts, especially when they are very complex and complex, as in this study,” he said. “But I also think we need to be very careful.”

According to Brownstein, protecting privacy is central to the use of such technologies.

“We’re not removing any human guardrails here,” Brownstein said. “A human needs to review everything the AI ​​does.”

The study also shows why old genetic test results may be worth revisiting. Because science changes rapidly, a result that didn’t make sense years ago may become clearer as researchers discover new genes, improve methods for searching genomic data, and learn more about how genetic changes affect health.

“A negative genetic test that is negative now may not be negative in the future,” Brownstein told ABC News.

Brownstein said the rapid pace of genetic discoveries could make it difficult for clinicians to double-check old unsolved cases, but artificial intelligence could help.

“The genome is being decoded more and more every day,” he said. “But AI is really good at this.”

In Kyra’s case, only 60 cases of MFM patients have been published so far. TreatmentMFM13A charity dedicated to finding solutions to the challenges of disease.

The study had limitations, the team said. The researchers looked at existing cases, so the study cannot prove that 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, reduced costs, or changed patients’ care.

The next step is to test the approach in larger prospective studies at multiple medical centers, the authors wrote.

How is AI being used to detect and treat cancer and identify potential risks for patients?

For Kyra, the diagnosis didn’t eliminate years of uncertainty, but she said it gave her and her family something they didn’t have before: a name, a sense of closeness and a connection to others living with the same rare condition.

While artificial intelligence can help researchers find answers faster, the human side of medicine is still most important, he said.

“When it comes to health issues that truly change your life, you want that human touch to be present,” Kyra said. “You want to feel like people care about you and are listening 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.