AI-Fueled Trial Matching Signals a Clinical Research Reboot

The launch of Mount Sinai Tisch Cancer Center’s AI-powered clinical trial matching platform, PRISM, marks a decisive shift in how large health systems may approach research participation at scale. While artificial intelligence continues to draw headlines across diagnostics, imaging, and documentation, this initiative surfaces a quieter transformation: the operational integration of trial eligibility into real-time care workflows.

Powered by Triomics’ OncoLLM, an oncology-specific large language model, the platform automates the trial-matching process across the Mount Sinai Health System, replacing manual chart reviews with continuous eligibility scanning embedded in the EHR environment. More than a technical upgrade, PRISM represents a systemic recalibration. One in which equitable trial access, long limited to academic hubs, becomes a byproduct of automation logic and standardized data governance.

Mount Sinai is now the first NCI-designated Comprehensive Cancer Center in New York City to operationalize such a tool across its entire system. That milestone reveals two interlinked strategic priorities: (1) improving the consistency and timeliness of trial enrollment, and (2) broadening participation among historically underrepresented patient populations.

From Manual Bottlenecks to Machine Triage

Traditional trial-matching has long relied on specialist coordinators poring over fragmented records to determine eligibility, an approach inherently biased toward patients with better proximity to academic centers and more frequent touchpoints with research staff. At Mount Sinai, this process is now being reengineered to run silently in the background, flagging eligible patients across seven hospitals and 400+ outpatient sites, regardless of their location or care pathway.

The result is a redefinition of “feasibility.” What was once a matter of staff availability and documentation completeness becomes a question of infrastructure integrity. Are clinical notes standardized? Are biomarkers discretely coded? Is the EHR structured enough to support automated querying?

This evolution aligns with broader industry efforts to operationalize computable eligibility criteria for clinical research. The Office of the National Coordinator for Health Information Technology (ONC), for example, has called for greater interoperability and structured data capture to support research recruitment as a byproduct of routine care documentation. PRISM’s deployment shows what such alignment looks like in practice.

Equity by Design or Algorithmic Optimism?

Mount Sinai’s systemwide deployment is particularly notable for its reach. Patients at community-based facilities like Mount Sinai Queens or Mount Sinai Brooklyn are now scanned for trial eligibility using the same AI-driven protocols as patients at its flagship academic center. This theoretically levels the playing field for enrollment, but also raises questions about what equity means in an algorithmic context.

Real-world data is messy. Clinical documentation may vary across sites, and structured fields are often inconsistently populated. An AI model is only as equitable as the data pipeline it ingests. As JAMA Oncology highlighted in 2023, clinical trial matching tools must be evaluated not only for accuracy but also for fairness, particularly when serving racially and socioeconomically diverse populations.

Mount Sinai’s implementation will likely become a test case for whether real-time matching can both scale and democratize trial access. The platform’s outputs may eventually reveal whether patients previously excluded due to geography or data quality are now being surfaced for enrollment, or whether algorithmic gaps replicate old inequities with new efficiency.

From Innovation Showcase to Operational Baseline

Triomics’ partnership with Mount Sinai reflects a growing trend in which academic medical centers seek commercial AI vendors to help embed research into everyday operations. But what begins as innovation risk often becomes baseline expectation.

Other health systems are watching. According to a recent report by Fierce Healthcare, peer institutions such as Duke and Johns Hopkins are also investing in AI-powered trial platforms, each taking a slightly different approach to integration, transparency, and governance.

What Mount Sinai’s approach underscores is that the future of research participation may be less about patient motivation and more about systemic design. Trial enrollment, in this model, becomes less of a campaign and more of a configuration: a matter of ensuring that eligibility flags reliably fire, documentation remains structured, and workflows are built for proactive outreach.

Governance, Not Just Guidance

AI in healthcare continues to draw scrutiny for potential bias, explainability gaps, and overpromised capabilities. In the trial-matching space, those concerns are amplified by ethical stakes. Flagging a patient for a trial they don’t actually qualify for wastes time; failing to flag one who does may delay access to life-saving care.

Mount Sinai appears attuned to this dynamic. The deployment involved coordination among Mount Sinai Research IT, clinical leaders at the Tisch Cancer Center, and Triomics engineers. The emphasis on cross-functional governance suggests that this is not a plug-and-play product drop, but a layered integration effort involving real-time testing, feedback loops, and ongoing model evaluation.

Looking ahead, the health system plans to publish findings from the PRISM deployment in peer-reviewed journals and national conferences. These outcomes will help define what “success” means in the AI-matching space. Is it increased enrollment? Faster screening? Broader demographic reach? Or fewer missed opportunities?

What This Signals for Healthcare Leaders

For CIOs and CMIOs evaluating AI tools, Mount Sinai’s deployment illustrates a critical point: not all clinical AI must deliver diagnostic precision or predictive analytics to be transformative. Sometimes, the most impactful interventions are logistical, streamlining how existing information gets surfaced and acted upon.

This platform is not redefining what trials exist, or even who is technically eligible. It is simply ensuring that eligibility becomes visible early enough to matter. That’s an operational breakthrough with clinical implications.

Healthcare systems seeking to meet national benchmarks for trial enrollment, address diversity mandates, or improve research integration across care sites may find more value in workflow-linked AI than in next-gen modeling.

If PRISM performs as expected, Mount Sinai will not just be expanding access to research. It will also be redefining the infrastructure of access itself.