HLTH-2027-02-TOOL-01: BIOMARK-AI: Bridging Biomedical Knowledge and Multimodal Data for Predictive Biomarkers

Our project will bring together clinical researchers, biomedical scientists, data experts and AI developers to accelerate the discovery and clinical validation of multimodal predictive biomarkers for chronic non-communicable diseases. Using established and emerging AI methods, we will integrate molecular, cellular, physiological, imaging and digital biomarkers with longitudinal clinical and real-world datasets to identify combinations that improve the prediction of disease progression and treatment response.

A distinctive feature of our project is its knowledge-driven MedAI approach, which automatically extracts and harmonises quantitative evidence on biomarkers and predictive models from scientific publications, maps this knowledge to variables in longitudinal clinical and real-world datasets, and integrates it with AI-driven analyses to identify and prioritise multimodal biomarker combinations for predicting disease progression and treatment response.

We will also provide executable predictive models that can be deployed locally and benchmarked and validated against independent cohorts. We will evaluate predictive accuracy, reproducibility, generalisability, bias, and clinical utility to support more effective and personalised treatment decisions.