TrialMatchAI: Advancing Precision Oncology with Open-Source Innovation

We are proud to announce the publication of TrialMatchAI in Nature Communications (https://www.nature.com/articles/s41467-026-70509-w), a groundbreaking project developed as part of the EU EOSC4Cancer initiative. This work addresses a critical challenge in precision oncology: matching patients to relevant clinical trials using heterogeneous clinical data, including structured information, biomarkers, and free-text clinical notes.

Key Innovations

End-to-End Open-Source System: TrialMatchAI is designed with transparency, reproducibility, and practical deployability in mind, ensuring trustworthiness in real-world clinical settings.

Modular and Traceable: The system provides traceable eligibility reasoning, enabling clinicians and researchers to understand and validate the matching process.

Performance and Explainability: Our approach delivers robust results on both real-world and benchmark datasets, with explainable criterion-level assessments to support clinical decision-making.

Collaboration and Impact

This milestone is the result of a collaborative effort, with special recognition to Majd Abdallah for leading the development and Mikaël Georges for his contributions to system evaluation. TrialMatchAI represents a significant step forward in leveraging AI to bridge the gap between
patients and life-saving clinical trials.

Explore Further

For more details, read the full publication in Nature Communications (https://www.nature.com/articles/s41467-026-70509-w) and visit the Bordeaux Bioinformatics page of the project for additional context.