Finding the right clinical trial has historically depended on which hospital a cancer patient happens to walk into, and whether an overworked research coordinator has time to manually cross-reference that patient’s chart against dozens of shifting eligibility criteria. Mount Sinai’s Tisch Cancer Center is trying to close that gap with a new AI platform called PRISM, launched in January 2026 in partnership with the AI company Triomics, according to Mount Sinai’s newsroom and reporting from Becker’s Hospital Review.
How PRISM works
PRISM is powered by Triomics’ OncoLLM, a large language model pipeline built specifically for oncology. Rather than requiring staff to manually sift through spreadsheets of trial protocols, the system reads a patient’s electronic health record, including diagnoses and clinical characteristics, and automatically flags which open trials that patient may be eligible for, earlier in their care journey than manual screening typically allows.
Why the systemwide rollout is notable
According to Mount Sinai, its health system became the first National Cancer Institute-designated Comprehensive Cancer Center in New York City to deploy an oncology-focused AI matching tool across its entire network, not just at its flagship hospital. That means patients treated at Mount Sinai Queens, Mount Sinai Brooklyn, Mount Sinai South Nassau, Mount Sinai Morningside and Mount Sinai West now have access to the same trial-matching capability as patients at The Mount Sinai Hospital in Manhattan, addressing a longstanding disparity where community-hospital patients had far less access to research studies than those at major academic centers.
The bigger problem this is meant to solve
Clinical trial enrollment in oncology has chronically lagged behind need; many trials fail to hit recruitment targets, and patients who might benefit from experimental therapies frequently never learn a relevant trial exists. Manual chart review to check trial eligibility is slow and inconsistent between institutions, which is part of why several major cancer centers, including Dana-Farber with its own MatchMiner-AI tool launched around the same period, have been racing to build AI-driven alternatives.
Perspectives and limitations
Supporters argue that tools like PRISM could meaningfully expand access to research for patients who would otherwise never be screened for a trial, particularly outside major flagship campuses. But critics of AI-based trial matching caution that these systems are only as good as the structured and unstructured data feeding them, and that algorithmic matching still requires a human research coordinator or physician to confirm eligibility, obtain consent, and manage the logistics of enrollment. There is also the open question of whether expanding the pool of “matched” patients actually translates into more enrollments, or simply more administrative alerts that still get triaged by short-staffed trial offices.
A crowded field of AI trial-matching tools
Mount Sinai is far from alone in this push. Dana-Farber Cancer Institute launched its own tool, MatchMiner-AI, in February 2026, using large language models to match patients to trials based on their full medical record and to generate plain-language summaries of complex eligibility criteria for patients. Separately, peer-reviewed work on an end-to-end system called TrialMatchAI has described automating patient-to-trial matching using fine-tuned, open-source language models within a retrieval-augmented generation framework. Industry analysts tracking this category estimate the market for AI-based clinical trial patient-matching solutions was worth roughly $576 million in 2025 and could grow at a compound annual rate above 30% through the mid-2030s, reflecting how central this problem has become across cancer centers nationally, not just at Mount Sinai. Tempus, another company active in this space, has reported that its own patient-query tool screened out roughly 72% of ineligible patients automatically and saw a 27.31% increase in patients ultimately identified as potential matches after nurse review, figures that suggest the efficiency gains Mount Sinai is chasing with PRISM are not unique to one platform but reflect a broader pattern across the AI trial-matching category.
What’s next
Mount Sinai says the goal is to keep expanding PRISM’s reach and refine its matching accuracy as more real-world data comes in from its satellite locations. If oncology-specific LLM tools like OncoLLM prove reliable at scale, similar systems could plausibly expand into other therapeutic areas beyond cancer, and hospital systems without in-house AI teams may increasingly license platforms like Triomics’ rather than build their own, turning trial matching into a standard piece of cancer-center infrastructure rather than a boutique offering at a handful of research hubs.
