Why this exists
I have lost people I love to cancer. Others close to me are alive right now because a clinical trial reached them in time. And others never found the trial that might have mattered — not because it didn’t exist, but because finding it is harder than it has any right to be.
That gap made me angry enough to build something.
Here is the problem in one sentence: every recruiting trial in the United States is publicly listed on clinicaltrials.gov, and almost nobody can find the one that fits. The registry speaks in eligibility criteria — “ECOG performance status ≤1, prior taxane exposure permitted, no untreated CNS metastases.” Patients speak in plain language: my mother is 58, the cancer spread to her liver, she has tried two drugs. Between those two languages sits a translation layer that has mostly lived inside the heads of research nurses and well-connected oncologists. Patients with networked doctors get the translation. Everyone else gets a keyword search.
Building AI systems is what I do for a living — by day I lead AI strategy at a large technology company; TrialThread is a personal project, built on my own time and not affiliated with my employer. The systems I work on read messy human language and structured data, and connect the two. TrialThread points that same machinery at trial matching: describe the diagnosis in your own words, and it reads the actual eligibility criteria of recruiting trials — starting near you, widening the search when the first pass is thin — then explains in plain English which trials may fit, and what might get in the way.
Three promises
Patients never pay. Not now, not later. If TrialThread earns money, it will come from trial sites and sponsors who pay for qualified referrals — the way trial recruitment already works today — and never from a patient or a family.
Nothing you type is stored. No account, no profile, no database of your medical situation. Your description powers the search in front of you and is then discarded. If we ever add saved searches, it will be opt-in, explained in plain language, and deletable.
Honesty over hope. If a trial probably will not take you, we say so, and we say why. False hope spends the one resource patients do not have.
Open, on purpose
The matching engine is open source under Apache-2.0 — read the code, run your own, make it better. The most useful contribution is not code: it is clinician-reviewed test cases that measure match quality. Trial sites and sponsors who want qualified, consented referrals can open a conversation through GitHub as well; that work would live in a separate service, under standard recruitment agreements, and it would never change the three promises above.
TrialThread will not cure anything. It closes a distance — between public information and the people whose lives depend on reaching it. The trials exist. The patients exist. The thread between them is what has been missing.
— Eric Porres, New York
I write about AI systems at Beyond Reason