Publication in Nature Communications: AI-powered system matches cancer patients to clinical trials
Precision cancer medicine depends upon getting patients enrolled in the right clinical trials at the right time. A study led by Majd Abdallah and Macha Nikolski (CNRS, University of Bordeaux), in collaboration with Sigve Nakken and Eivind Hovig (Institute for Cancer Research), introduces TrialMatchAI — an AI-powered software system designed to automatically match cancer patients to relevant clinical trials.
Finding the right clinical trial for a cancer patient is often a slow and complex task. Each trial has its own list of requirements that patients must meet to be eligible, and these are often written in dense clinical language. Searching through hundreds of trials manually is neither practical nor reliable, and as a result many trials struggle to recruit enough patients in time — slowing the development of new treatments.
TrialMatchAI tackles this problem by reading and interpreting patient records — both structured data and free-text clinical notes written by doctors — and comparing them automatically against a large database of trials. A strength of the system is its ability to match patients based on molecular characteristics, such as specific mutations in their tumour DNA, to trials that require exactly those features for enrollment. This kind of biomarker-based matching is increasingly central to modern cancer medicine, but is particularly difficult to automate. Importantly, TrialMatchAI also tells clinicians why a particular trial was recommended, not just that it was.
"This work shows that AI-driven patient-trial matching is both feasible and highly promising. At the same time, there is still important work ahead before such systems are ready for routine clinical use, particularly validation across larger and more diverse patient cohorts", says Nakken.
Nakken and Hovig contributed text mining analytics and scientific expertise related to the precision oncology and biomarker focus of the project. The work was funded by the EU under the EOSC4Cancer project. Nakken was further supported by an Åsgard Mobility Program (France-Norway) grant.
Links:
The Nature Communications publication:
TrialMatchAI: an end-to-end AI-powered clinical trial recommendation system to streamline patient-to-trial matching.Abdallah M, Nakken S, Georges M, Bierkens M, Galvis J, Groppi A, Karkar S, Meiqari L, Rujano MA, Canham S, Dienstmann R, Fijneman R, Hovig E, Meijer G, Nikolski M.
Nat Commun. 2026 Mar 25. doi: 10.1038/s41467-026-70509-w.
Computational Biology & Bioinformatics - CNRS / University of Bordeaux