Three years ago, Rich Kaplan ran into a kind of medical barrier that rare-disease patients know well: the treatment existed, but the paperwork demanded proof. So he turned to ChatGPT. Based in Seattle, Kaplan has catastrophic antiphospholipid syndrome (APS), a rare autoimmune clotting disorder that can damage organs, including the kidneys. As he puts it, “About half the people who get this don’t make it past the first month.”
His nephrologist prescribed a medication intended to slow the kidney damage. The prescription reached the pharmacy, then stalled. The prior authorization failed. An exception letter written by his doctor failed. It was an expensive medication for a special case. Ninety days passed, then 120. The insurer would not cover the drug without documented evidence that it helped someone with a condition as rare as his.
After decades at Microsoft, Rich fell back on something he knew well: problem solving with software. He opened ChatGPT and asked it to “find every reference to this medication being used for this disease,” to “show the clinical trials, case reports, studies, anything that exists.” What came back was a referenced, five-page synthesis pulling together the available evidence. He prefaced that with a short, plain language introduction: “I’m alive today. I have a rare disease. I’m trying to stay alive longer.” When the packet went through the insurer’s third-party arbitration process, it was finally approved.
Americans like Kaplan are sending nearly 2 million ChatGPT messages every week navigating health insurance claims, billing, and eligibility, according to a new analysis of anonymized message data. But his use goes well beyond that.
Kaplan takes many medications, which can wear down both patients and their clinicians. Now he imports his medication list from Epic MyChart and asks the model to retain it. When he considers an over-the-counter medication, he checks first. Once, the model flagged a stomach remedy as risky because of his kidney disease, so he wisely passed.
He also uses ChatGPT to translate medical records into something he can act on: visit summaries, specialist notes, dense instructions that pile up over decades. He asks for a synopsis, patterns over time, and questions to bring to his next appointment, then uses those responses to have better conversations with his doctors.
Kaplan shares his approach with other APS patients, especially those far from major medical centers. He warns them that some clinicians still prescribe the wrong blood thinners for APS. He tells them to go in prepared, ask why, and ask for alternatives when something feels off.
His bigger message is about agency. “People are scared of AI,” he says. “I get it. But you have to move past the fear.” For Kaplan, ChatGPT is a way to secure better treatments, get coverage approvals, and translate medical information so that he can stay engaged with his own care.
