Using AI to improve patient access to clinical trials
OpenAI
OpenAI says a startup called Paradigm is using its API to help match patients to clinical trials faster. The pitch: less paperwork, more people actually getting into studies that could help them.
Based on reporting by OpenAI — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
Clinical trial recruitment has always been one of medicine's dumbest bottlenecks. Doctors and patients often have no idea which trials exist, and the ones who do know still have to wade through pages of eligibility criteria written in dense medical shorthand. Paradigm, according to a new case study from OpenAI, is trying to fix that by plugging GPT-style models into the process of screening and matching patients against trial requirements.
The basic idea isn't new — health tech companies have tried natural language processing on trial protocols for years. What's changed is that large language models can now read a messy patient record or a clinician's notes and cross-reference them against a trial's inclusion and exclusion criteria without a small army of coordinators doing it by hand. That matters because trial recruitment delays are one of the biggest reasons drug development drags on for years and costs hundreds of millions of dollars.
OpenAI frames this as part of its broader push into regulated, high-stakes industries where accuracy and speed both count. Healthcare has been a favorite target for this kind of pitch, partly because the inefficiencies are so glaring and partly because even modest improvements can translate into real money and, more importantly, real patients getting access to treatments sooner. Paradigm's use of the API apparently focuses on making that matching step faster for research sites and sponsors, not on replacing the clinical judgment that ultimately decides who enrolls.
What's missing from OpenAI's post, as usual, is much detail on error rates, oversight, or what happens when the model gets a match wrong in a space where a wrong call isn't just annoying — it can mean someone missing a trial that might have helped them, or worse, someone being pointed toward one they shouldn't be in. These case studies read like polished advertisements, and this one is no exception. Still, the underlying problem is real, and if AI can shave weeks off recruitment timelines without cutting corners on safety, that's a genuine win buried under the marketing gloss.
My take — AI-written commentary, not fact-checked reporting
I'll believe the impact once we see real recruitment numbers instead of a blog post co-written with the vendor's PR team. Healthcare is exactly the kind of high-stakes domain where OpenAI needs visible wins to justify the enterprise narrative, so forgive me for reading this as much as a sales pitch as a breakthrough. That said, if it actually gets one more patient into a trial that saves their life, I'm not going to be the guy complaining about the marketing copy.
Read more about this at: OpenAI