Truveta Language Model unlocks research-ready cancer staging data from millions of clinical notes
Published in JCO Clinical Cancer Informatics, new Truveta research demonstrates scalable, high-precision extraction of critical cancer staging information
BELLEVUE, Wash., July 29, 2026 (GLOBE NEWSWIRE) -- Truveta recently published a peer-reviewed study in JCO Clinical Cancer Informatics demonstrating how large language models (LLMs) can extract complex cancer staging information from unstructured clinical documentation with high precision and at unprecedented scale. The publication, “Large Language Models to Extract Cancer Staging Data from Clinical Documentation at Scale,” highlights the capabilities of Truveta Language Model for Oncology (TLM-Oncology), an AI model developed by Truveta to unlock critical oncology variables from the clinical notes that contain some of the deepest clinical information in cancer care.
One of the most important variables in oncology research is a patient's stage of cancer, or how far the disease has progressed when they are diagnosed. Unfortunately, patients' cancer stage is often captured in free-text physician notes, pathology reports and imaging reports—rather than structured electronic health record fields—making it difficult to access.
Now, with the new technical avenues made available through AI, researchers may have found a way at last to get this critical clinical context quickly and accurately. "Some of the most clinically meaningful oncology information exists only within unstructured clinical documentation," said Sujatha Sagiraju, Senior Vice President of Engineering, Clinical Intelligence, and AI Evaluation at Truveta and co-author of the study. "This work demonstrates how deep clinical expertise, advanced AI, and scalable engineering can unlock those insights and transform them into research-ready data.”
In the study, Truveta researchers developed and validated TLM-Oncology to extract cancer staging information across bladder, breast, cervical, colorectal, and prostate cancers. The model analyzed more than 2 million clinical notes and extracted more than 2.7 million staging records representing more than 217,000 patients with different types of cancer. The results demonstrated high precision across cancer types and note formats, including the model's ability to generalize successfully to cancer types it had not previously encountered during training.
The publication reflects a multidisciplinary effort across Truveta's clinical, AI, and engineering teams. Clinicians guided the interpretation of complex cancer documentation, AI scientists fine-tuned large language models to extract nuanced clinical concepts, and engineers built the infrastructure needed to process millions of notes while maintaining quality, consistency, and scalability.
The resulting data can be combined with structured electronic health record information in Truveta Data to create deeper, more complete real-world oncology datasets for research. Unlike prior approaches that focused on a single cancer type, pathology reports alone, or limited staging elements, TLM-Oncology extracts comprehensive staging records, including overall stage, stage type, staging method, associated cancer diagnosis, and relevant timeframes.
"The success of AI in healthcare depends on more than powerful models—it depends on deep clinical expertise," said Swapna Abhyankar, MD, physician informaticist and Director of Clinical Informatics, Clinical Intelligence and Evaluation, at Truveta and corresponding author of the study. "Our clinicians and AI scientists worked together to teach the model the nuances of how cancer staging is documented in real-world settings. By combining clinical judgment with rigorous validation, we've shown that AI can reliably unlock critical oncology variables from clinical notes and make them available for research at a scale that was previously impossible."
The study also underscores the broader potential of Truveta's AI-powered data platform. Since the research was conducted, Truveta has expanded its extraction methods beyond staging to additional oncology variables —including histologic subtype, tumor markers, metastasis, performance status, treatment response, and grade—and has scaled these capabilities across all solid tumors.
The publication of this new body of evidence advances Truveta's mission of Saving Lives with Data by helping researchers access the detailed, real-world oncology information needed to accelerate evidence generation, improve understanding of disease progression, and ultimately help patients benefit from faster scientific progress.
About the Study
The study, "Large Language Models to Extract Cancer Staging Data from Clinical Documentation at Scale," was published in JCO Clinical Cancer Informatics.
About Truveta
Truveta transforms healthcare into intelligence. Built with and owned by US health systems united by a mission of Saving Lives with Data, Truveta powers breakthrough discoveries, regulatory-grade evidence, and AI that improves patient care.
Truveta membership includes: Providence, Advocate Health, Trinity Health, Tenet Healthcare, Northwell Health, AdventHealth, Baptist Health of Northeast Florida, Baylor Scott & White Health, Bon Secours Mercy Health, CommonSpirit Health, Hawaii Pacific Health, HealthPartners, Henry Ford Health System, HonorHealth, Inova, Lehigh Valley Health Network, MedStar Health, Memorial Hermann Health System, MetroHealth, Novant Health, Ochsner Health, Premier Health, Saint Luke’s Health System, Sanford Health, Sentara Healthcare, Texas Health Resources, TriHealth, UnityPoint Health, Virtua Health, and WellSpan Health.
Alexis Hauk Truveta news@truveta.com
Legal Disclaimer:
EIN Presswire provides this news content "as is" without warranty of any kind. We do not accept any responsibility or liability for the accuracy, content, images, videos, licenses, completeness, legality, or reliability of the information contained in this article. If you have any complaints or copyright issues related to this article, kindly contact the author above.
