These parents didn't know why their daughter was sick. They used AI to help solve the mystery.

Hilary Eaton was in a work meeting when her phone lit up with a notification. New lab results had arrived for her 5-year-old daughter, Olivia. For most of Olivia's life, doctors had been unable to explain why she was always getting sick. So Eaton opened an AI chatbot and started looking for answers herself.

Olivia had blue eyes, chubby cheeks, and a playful smile. But from the time she was a baby, something seemed wrong. She couldn't sit up as an infant, never slept through the night, and cuts on her skin wouldn't heal properly. When she got colds, they lasted for weeks instead of days.

The family visited many doctors but got few clear answers. Doctors diagnosed Olivia with a connective tissue disorder, which affects how the body's tissues hold together. That explained some of her early physical problems, but it still didn't explain why she kept getting sick so often.

In January 2025, the new lab results showed that Olivia had far fewer antibodies than a healthy child should have. Antibodies are proteins that help the body fight off germs and infections. Having too few of them can make it very easy to get sick. Eaton was shocked and texted her husband, Matt, right away.

Since she couldn't reach Olivia's doctor quickly, Eaton decided to try something new. She and Matt both had science degrees and already used AI tools in their jobs in the biotech industry. She opened ChatGPT, typed in Olivia's symptoms and lab results, and asked what diseases might explain what she was seeing.

Within seconds, ChatGPT gave her a response. The chatbot suggested a condition called CVID, or common variable immunodeficiency, where the body doesn't make enough antibodies. Though CVID turned out not to be Olivia's actual diagnosis, having a starting point gave Eaton a sense of control. She felt ready to go back to the doctor with real questions.

Using AI for health questions has become very common. A Pew Research Center survey found that one in four Americans now uses AI chatbots to look up symptoms or understand health information. Every week, more than 300 million people ask ChatGPT health-related questions, according to OpenAI.

But experts warn that AI is not perfect. Chatbots can sometimes produce wrong or made-up information, which is called a hallucination. Experts say AI works best when people use it carefully, double-check results, and always talk with their doctors about what they find.

Hilary and Matt Eaton first met as graduate students at Duke University. Hilary earned a PhD in cancer biology and did research at Harvard Medical School, while Matt earned his PhD and did research at MIT. Their strong science backgrounds made them well suited to tackle a complicated medical mystery using AI tools.

In 2023, the Eatons spent about $4,500 to have the complete genetic information of all four family members analyzed. After putting their daughters to bed each night, they stayed up late searching through the data with ChatGPT. They were looking for gene mutations — tiny errors in DNA — that could explain Olivia's health problems, and they saved everything in a folder called Genome Detectives.

Because they knew AI could make mistakes, they told the chatbot about their science training and asked it to provide links to real research papers. They double-checked every piece of information they found. Even with those safeguards, there were frightening moments along the way.

At one point, the chatbot flagged a mutation linked to a very serious condition. Children with that mutation often died between the ages of 7 and 10, and Olivia was 6. The parents panicked — but it turned out the genome data itself had errors, and Olivia did not have that dangerous mutation. They cleaned up the data and kept going.

Their detective work eventually led to real results. A chatbot suggestion helped them ask doctors about a monthly IV treatment that gives Olivia healthy antibodies from donors. Doctors agreed it was a good choice, and Olivia started the treatment last fall. Before the treatment, Olivia was sick almost constantly. Afterward, she missed only three days of school the entire year.

Digging deeper with AI also helped find the root cause of Olivia's problems. Eaton discovered a mutation that could explain why Olivia's skin healed so slowly. That discovery led doctors to order more genetic testing, which pointed to a rare condition called Okur-Chung neurodevelopmental syndrome, a disorder that affects about 1 in 100,000 people and can cause developmental delays and immune system problems.

This past winter, Olivia began making strange, uncontrolled facial movements. To prepare for a visit to a brain specialist called a neurologist, Eaton asked ChatGPT what questions to bring up. The chatbot suggested asking about an EEG, a test that measures electrical activity in the brain. Eaton brought it up at Olivia's appointment, and the doctor agreed it was worth trying.

Less than an hour after the EEG, while Olivia and her mother were shopping for fuzzy pajama pants as a reward, Eaton's phone rang. It was the neurologist, Dr. John McLaren. He asked if she was sitting down. The results showed abnormal brain activity that meant Olivia might be prone to seizures, and he ordered anti-seizure medicine right away.

Once Olivia started taking the medicine, something remarkable happened. For the first time in her life, she began sleeping through the night. Now 7 years old, Olivia still goes to the hospital once a month for her antibody treatment, but she is no longer sick nearly as often. Her family says they finally feel like they are getting real answers.

The Eatons are now using ChatGPT to look into the health data of their older daughter, Fiona, as well. Olivia's doctor, Dr. Wendy Chung, said the family shows how AI can work best — as a partner with doctors, not a replacement for them. Eaton agrees, and she says she still has plenty more to learn. "I'm not an expert yet," she said, "because I haven't spent enough time with ChatGPT."