Results achieved

During July 2026, Dr. Natsiavas engaged with Nostrum Biodiscovery and Barcelona Supercomputing Center staff to explore high-performance computing approaches to AI evaluation and biomedical research.

This secondment provided an invaluable opportunity to engage in direct collaborative exchanges, specifically focusing on the intersection of data science, clinical frameworks, and ethical AI. The initial phase of the visit allowed for a comprehensive alignment of methodologies, exploring how different research environments frame data disparities and computational fairness in healthcare.

The core of the work involved a deep dive into potential bias factors that influence AI models within health systems. Through active collaboration with local researchers specializing in bioinformatics and real-world data analysis, the joint effort focused on how systemic, algorithmic, and data-level biases manifest in biological datasets and clinical workflows. This specialized exchange highlighted the importance of addressing data representation and confounding variables when training healthcare models.

The primary output of this secondment was the compilation of a detailed report mapping these bias factors and proposing considerations for mitigating them in real-world data settings. This collaborative work successfully combined INAB-CERTH’s domain insights with the host institution’s bioinformatics expertise, establishing a strong foundation for ongoing joint initiatives and contributing directly to the broader goals of the AHEAD project.