Triomics' AI Flags Trial Patients, Iu Health Tests If It Scales

Clinical trials in oncology keep missing patients, not from lack of interest, but from messy, manual screening. IU Health, Regenstrief, Triomics, and Eli Lilly are running a live test of Triomics PRISM inside two community oncology clinics to see if AI can actually help - without adding work or chaos.

AI finds patients: PRISM sits in the clinic workflow and suggests likely trial matches, rather than being another dashboard nobody checks. Quick facts:

• The test runs in two satellite sites chosen for messy, real-world conditions - staffing limits and inconsistent notes included.
• The study is a hybrid effectiveness-implementation design, meaning it measures accuracy and whether clinicians will actually use it.
• The goal is trust and fit first, scale second - not a flashy sales demo but a pragmatic pilot.

Why does this matter?: Clinical trial enrollment tanks because screening is slow and fragmented. If AI can reliably flag eligible patients in routine care, trials get faster and more representative. That’s huge for drugs and for patients who live outside academic centers.

But don’t buy the hype blind. Vendor claims and retrospective tests rarely show how tools behave in live workflows. Real problems lurk: variable documentation, alert fatigue, and workflow disruptions that quietly kill adoption. This study’s smart move is starting small - test trust, measure clinician burden, then decide on scaling. That approach reduces deployment risk and forces evidence over enthusiasm.

If PRISM actually passes these checks, it’s a blueprint: AI evaluated inside the mess it’s supposed to help, not in sanitized demos. If it fails, the system learns where and why - which is better than rolling out a tool that looks great on paper and creates more work in practice. Either way, this is the kind of real-world test AI needed, not another vendor press release.