AI Clinical Trial Matching Study in 3,804 Cancer Patients by Massive Bio
Massive Bio, a global precision oncology company, has announced the publication of a major prospective study demonstrating the real-world effectiveness of its artificial intelligence (AI) platform for matching cancer patients to clinical trials. The study, published in ESMO Real World Data and Digital Oncology as part of a special issue focused on Artificial Intelligence in Clinical Oncology, represents one of the most comprehensive and rigorously validated evaluations of AI-driven trial matching conducted to date.
Titled “Transforming oncology clinical trial matching through neuro-symbolic, multi-agent AI and an oncology-specific knowledge graph: a prospective evaluation in 3,804 patients,” the study moves beyond theoretical or retrospective analyses and instead provides prospective, real-world evidence collected during routine oncology care. This distinction is critical, as most previous research in this space has relied on simulated datasets or retrospective chart reviews rather than evaluating AI systems in live clinical environments.
The study followed 3,804 patients with metastatic cancer throughout 2024, during which Massive Bio’s platform processed more than 157,000 pages of clinical documentation. From this data, the system generated over 17,000 clinical trial matches that were subsequently confirmed by oncologists. These results highlight the platform’s ability to function at scale while maintaining clinical accuracy and relevance.
One of the most significant findings of the study is the dramatic improvement in efficiency. The AI system reduced the time required to match a patient to suitable clinical trials from approximately 120 minutes using traditional methods to just 30 minutes. This fourfold increase in speed has important implications for clinical workflows, where time constraints often limit the ability of physicians to thoroughly evaluate trial eligibility.
In addition to speed, the platform demonstrated strong performance in terms of accuracy. The study reported an F1 score of 0.82, with both sensitivity and specificity measured at 0.84. These metrics indicate a high level of agreement between the AI-generated matches and oncologist decisions. Notably, the system outperformed several baseline approaches, including zero-shot GPT-4, chain-of-thought prompting methods, and GPT-4o-based models, underscoring the value of its specialized architecture.
At the core of Massive Bio’s platform is a neuro-symbolic, multi-agent AI system. Unlike traditional AI models that rely solely on probabilistic language processing, this approach integrates multiple components designed to handle the complexity of oncology data. The system uses agentic decomposition, dividing the task of trial matching into distinct steps such as data extraction, normalization, and reasoning. This modular design allows each component to be optimized for its specific function.
A key element of the architecture is an oncology-specific knowledge graph, which provides structured, domain-specific information that enables deterministic and auditable decision-making. This is particularly important in clinical trial matching, where eligibility criteria often involve complex relationships between biomarkers, disease progression, treatment history, and temporal constraints. By grounding its reasoning in a knowledge graph, the system can deliver transparent and explainable results rather than relying solely on statistical inference.
Another important feature is the inclusion of a human-in-the-loop safety layer. This ensures that oncologists remain actively involved in the decision-making process, reviewing and confirming matches before they are acted upon. This hybrid approach combines the efficiency of AI with the clinical judgment of physicians, addressing concerns about reliability and trust in automated systems.
The study also places a strong emphasis on equity, a critical consideration in the deployment of AI in healthcare. The evaluation framework was specifically designed to assess performance across different demographic and disease subgroups. Results showed that no subgroup experienced a performance gap greater than 10 percentage points, suggesting that the system maintains consistent effectiveness across diverse patient populations. This “equity-by-design” approach aligns with broader efforts to ensure that digital health technologies do not exacerbate existing disparities in access to care.
Leadership at Massive Bio emphasized the significance of the study’s findings. According to Dr. Arturo Loaiza-Bonilla, Co-Founder and Chief Medical AI Officer, the research marks a turning point in the field. Rather than debating whether AI can be applied to clinical trial matching, the focus is now on demonstrating how it can be implemented effectively at scale. He highlighted the importance of the system’s architecture, noting that neuro-symbolic, multi-agent models grounded in domain-specific knowledge graphs represent a foundational infrastructure for modern oncology.
Dr. Selin Kurnaz, Co-Founder and CEO, pointed to the study’s prospective design and rigorous validation as key differentiators. By comparing AI-generated matches directly with real oncologist decisions in live clinical settings, the research provides a level of evidence that goes beyond curated benchmarks. She also stressed the importance of proactively identifying and addressing equity gaps during the development and evaluation process.
From a technical perspective, Çağatay M. Çulcuoğlu, Co-Founder and CTO/COO, highlighted the challenges of building AI systems that can operate reliably in real-world conditions. Clinical data is often fragmented, incomplete, and highly variable, making it difficult for conventional models to perform consistently. The platform’s architecture was specifically designed to handle this complexity, with the knowledge graph serving as a backbone that ensures robustness, auditability, and resilience.
The study’s publication in ESMO Real World Data and Digital Oncology is particularly significant because it aligns with the organization’s broader vision for integrating AI into clinical practice. ESMO has identified three key pillars for the responsible adoption of AI in oncology: real-world data, transparency in AI systems, and equity in healthcare delivery. Massive Bio’s research addresses all three areas.
First, the use of prospective, real-world data ensures that the findings are directly applicable to everyday clinical practice. Second, the system’s architecture provides transparency and auditability, which are essential for building trust among clinicians and regulators. Third, the focus on equity helps ensure that the benefits of AI are distributed fairly across patient populations.
The implications of this work extend beyond technical performance metrics. One of the most persistent challenges in oncology is the low rate of clinical trial participation. In the United States, only about 3% to 5% of adult cancer patients enroll in therapeutic trials. This is not due to a lack of available studies but rather the difficulty of identifying eligible patients in a timely manner.
Clinical trial eligibility criteria have become increasingly complex, often requiring detailed analysis of patient records, biomarker profiles, and treatment histories. This process is further complicated by the fragmented nature of healthcare data, which is frequently spread across multiple systems and formats. As a result, many potentially eligible patients are never matched to appropriate trials.
Massive Bio’s platform aims to address this gap by functioning as an integrated component of clinical infrastructure. By embedding AI directly into oncology workflows, the system enables real-time trial matching at the point of care. This allows oncologists to make informed decisions more quickly and ensures that patients have access to potentially life-saving treatment options.
To date, the company’s platform has onboarded more than 200,000 cancer patients and supports matching across over 19,000 active interventional oncology and hematology trials worldwide. This scale demonstrates the platform’s potential to transform how clinical trials are accessed and utilized.
In conclusion, the study represents a significant advancement in the application of AI to oncology. By combining speed, accuracy, transparency, and equity, Massive Bio’s neuro-symbolic, multi-agent system offers a practical solution to one of the most challenging problems in cancer care. As the field continues to evolve, this work provides a strong foundation for the broader adoption of AI-driven clinical decision support systems, ultimately helping to bridge the gap between patients and the trials that could improve their outcomes.
About Massive Bio
Massive Bio, co-founded by Selin Kurnaz, Arturo Loaiza-Bonilla, and Çağatay Çulcuoğlu, transforms the pharmaceutical value chain with AI-driven solutions. As an AI-enabled real-world data company, Massive Bio streamlines patient journeys, improves access to cutting-edge treatments, and optimizes clinical trial operations across 17 countries. A recipient of the DiMe Seal, the Digital Medicine Society’s independent quality certification covering clinical evidence, privacy, security, and usability, Massive Bio is listed in the CMS Medicare App Library, connecting its platform to more than 68 million Medicare beneficiaries. A founding member of the CancerX public-private partnership and participant in the White House Cancer Moonshot initiative, the company continues to lead the way in ethical AI and data-driven innovation.
