Harnessing AI to Empower the Community Oncology Workforce

Each May, National Cancer Research Month offers a moment to celebrate the achievements of oncology clinical research while also recognizing the critical work still ahead. Thanks to numerous successes in preclinical cancer research, the field has reached a pivotal juncture where progress is increasingly constrained not by discovery, but by the capacity to conduct and enroll patients in clinical trials. To fully realize the promise of these scientific discoveries and ensure patients have access to potentially life-changing treatments, the clinical trial workforce must be strengthened. As oncology data expands at an unprecedented rate, clinicians and research staff face escalating demands. This burden is felt acutely in community oncology settings, where resources are often limited and operational pressures are high.

A key challenge facing community oncology sites is the complex, time-consuming process of matching patients to appropriate clinical trials. Oncologists and their teams are expected to evaluate thousands of patients against the eligibility criteria for hundreds of active trials each year. This task requires deep clinical knowledge, attention to detail, and, traditionally, labor-intensive manual chart reviews. Given these constraints, it’s not uncommon for patients to miss out on potentially lifesaving clinical trials—not because they are ineligible, but because the opportunity was not recognized in time.

At the recent 46th Annual Meeting of the Society for Clinical Trials (SCT) in Vancouver, a forward-looking session focused on a potential solution: the use of artificial intelligence (AI) to support and streamline patient-trial matching in real-world clinical environments. This technology offers a scalable, intelligent way to reduce workload and expand access to cancer clinical trials in the community setting.

Integration of AI-Powered Trial Matching Into Clinical Operations 

In place of overhauling active IT systems and infrastructure, sessions at SCT highlighted AI solutions, such as PRISM, that are designed to integrate within existing clinical workflows, rather than altering or adding to them. PRISM is embedded directly within commonly used electronic health record (EHR) systems. 4 core AI-powered applications were highlighted:

• Patient-to-Trial Matching: Smarter Screening Before the Visit
  This function proactively scans upcoming patient appointments and identifies all locally available trials specific to individual patients. By surfacing opportunities before the patient even walks into the clinic, providers can have timely, informed discussions about trial participation.  

• Trial-to-Patient Matching: Boosting Enrollment for New Studies 
  When a new trial opens, cancer centers can run the reverse search: identifying patients across the system who meet eligibility criteria. This can meaningfully shorten the time between trial activation and first enrollment, allowing promising therapies to reach patients faster. 

• Continuous Monitoring: Keeping an Eye on Future Matches AI systems can also track previously unmatched patients and alert staff when new trials (or changes in a patient’s condition) make them eligible. This longitudinal tracking ensures that patients aren’t screened once and forgotten but are continuously considered for relevant studies over time. 

• External Referral StreamliningBeyond the walls of a single cancer center, AI tools can help screen and refer patients from (or to) affiliated practices or regional networks. This creates a larger, more diverse pool of potential participants while helping smaller or more rural clinics provide access to research without duplicating effort.  

Together, these capabilities improve efficiency, support equity, and create a scalable model for trial recruitment that fits community oncology's complex landscape. For trial sponsors, the benefits are equally compelling: increased enrollment, reduced timelines, and broader, more representative data. 

Guardrails for Trust: Building Transparent, Verifiable AI Tools 

A known challenge of generative AI is the potential for “hallucinations,” or inaccurate suggestions. To mitigate this risk, the SCT session emphasized transparency: each AI-generated match is linked directly to the patient’s medical record, with clear explanations for why a patient does or does not meet inclusion/exclusion criteria. This enables real-time, human-in-the-loop validation, reinforcing both trust and accountability in patient care. 

Humans and Technology: A Partnership Model for Implementation 

Recognizing that successful adoption of new tools requires human expertise and support, the session emphasized an end-to-end implementation model to support the rollout of AI tools like the PRISM platform. AI solutions can (and should) be deployed alongside clinical experts who ensure that the tool is a good fit for the realities of day-to-day clinical practice. The human component of this team includes the following: