Artificial intelligence (AI) continues to reshape how clinical research is conducted, and a new framework called TRIALSCOPE may mark the next step in leveraging electronic medical records (EMRs) for regulatory-grade evidence. Published in NEJM AI on September 22, TRIALSCOPE demonstrated the ability to extract unstructured EMR data and simulate cancer clinical trials with results that mirrored real-world outcomes.
What is TRIALSCOPE?
Developed by researchers at Microsoft and academic partners, TRIALSCOPE combines multiple AI capabilities—including biomedical language models, probabilistic modeling, and causal inference—to transform noisy, incomplete EMR data into usable real-world evidence (RWE).
The framework has five main components:
- Data structuring pipeline to extract and organize EMR information.
- Probabilistic modeling to address missing data and reduce noise.
- Patient triaging system to identify eligible patients for target trials.
- Causal inference model to simulate trial designs.
- Validation tests to ensure robustness and transparency.
This end-to-end process enables researchers to simulate clinical trial outcomes without requiring new patient enrollment, potentially saving both time and resources.
Testing Against Cancer Trials
To validate TRIALSCOPE, researchers applied it to cohorts of advanced non-small cell lung cancer (NSCLC) patients drawn from a large U.S. health network. Eleven published trials were replicated using the framework:
- In nine of the eleven trials, TRIALSCOPE’s simulated hazard ratios for overall survival matched published trial data.
- In one trial (EMPHASIS-lung) that was discontinued due to lack of patient enrollment, TRIALSCOPE was able to simulate outcomes and suggest treatment efficacy, passing multiple diagnostic checks.
- Beyond lung cancer, the framework was also applied to pancreatic cancer trials, where results were consistent with clinical findings—demonstrating its potential generalizability.
Benefits and Challenges
Potential Benefits
- Scalability: EMRs can be converted into trial-quality evidence at scale.
- Efficiency: Trials that fail due to recruitment challenges may still produce valuable evidence.
- Broader insights: By removing trial eligibility criteria, TRIALSCOPE can also model real-world patient populations, offering insights beyond tightly controlled study designs.
Ongoing Challenges
Experts caution that the technology is not a full replacement for randomized controlled trials (RCTs). Editorials accompanying the study highlighted concerns about unmeasured confounding variables and the difficulty of applying TRIALSCOPE in therapeutic areas where diagnosis and outcomes are more subjective (e.g., asthma, neurological disorders).
Instead, the framework may be best positioned as a complementary tool—helping to inform trial design, augment evidence bases and provide additional context for regulatory and clinical decisions.
Implications for Industry and Regulators
The ability to generate validated real-world evidence directly from EMRs represents a significant step forward for life sciences companies and regulators alike. For sponsors, tools like TRIALSCOPE could accelerate development timelines, provide early insights into treatment effectiveness, and inform submission strategies. For regulators, it signals a growing need to establish clear standards for the use of AI-generated RWE in product approvals.
EMMA International’s Role
At EMMA International, we closely monitor advancements like TRIALSCOPE to help our clients stay ahead of emerging trends in real-world evidence and regulatory science. Our team supports sponsors in:
- Integrating RWE into development strategies to complement traditional clinical data.
- Engaging with regulators to align on the use of innovative evidence sources.
- Designing compliant processes for AI-enabled data curation, validation, and submission.
As AI-driven frameworks like TRIALSCOPE evolve, EMMA International is here to guide life sciences companies in responsibly adopting these technologies—ensuring that innovation translates into safe, effective, and timely therapies for patients.
For more information on how EMMA International can assist, visit www.emmainternational.com. Contact EMMA International at (248) 987-4497 or by email at info@emmainternational.com to learn more..com. Contact EMMA International at (248) 987-4497 or by email at info@emmainternational.com to learn more.ternational.com. Contact EMMA International at (248) 987-4497 or by email at info@emmainternational.com to learn more.8). FDA finalizes guidance on therapeutic protein biosimilars.
