REALM Consortium Meets in Warsaw to Drive Forward AI-Based Medical Device Software Evaluation
The REALM consortium convened for its 3rd Progress Meeting from 23–25 June 2025, hosted by the University of Warsaw in the historic heart of Warsaw. For the direct benefit of patients, the interdisciplinary team of 15 European partners aims to develop a collaborative framework that enables regulatory authorities, software developers, healthcare professionals, and policymakers to create and evaluate innovative medical device software jointly. The aim is to ensure that AI-driven healthcare solutions are not only technically sound but also ethically responsible, clinically useful, and regulatory-compliant from the ground up.
Now 30 months into the project’s lifecycle, the three-day event brought together the team working on the REALM platform to review progress and chart the path toward practical implementation.
Prof. Michel Dumontier (Maastricht University), the project coordinator, opened the meeting, congratulating everyone on their work done so far, highlighting the progress made and milestones achieved. The meeting showcased steady technical progress across the project’s work packages, spanning ethical frameworks, data architecture, use cases, and demonstrators. Updates from WP1 to WP7 highlighted achievements and upcoming deliverables, while also encouraging discussions between technical, ethical, and communication teams.
Partners engaged in a hands-on session with the RIANA Dashboard, REALM’s interface for data access and AI interaction, led by Bart Elen from VITO. A role-play exercise simulated real-world evaluation scenarios, with participants stepping into the roles of GDPR experts, healthcare regulators, patient representatives, and HTA evaluators. The session evaluated the dashboard’s usability, transparency, and compliance, providing valuable feedback for further development.
Spotlight on Sustainability and Exploitation
Day two focused on how REALM’s solution can continue beyond the project’s end. Led by EURICE, the Exploitation and Sustainability Workshop addressed critical questions of integration, governance, and long-term value creation. Consortium members mapped potential pathways for deploying REALM’s tools in clinical, research, and regulatory settings.
“We're not just building prototypes, we're preparing for real-world adoption,” noted Dr Gökhan Ertaylan, REALM's project co-coordinator, during the workshop. “This means strategic planning for ecosystem compatibility, scalability, and trustworthiness from day one.”
Real-World Applications: The Use Cases Take Centre Stage
The REALM platform brings together different stakeholders to collaboratively develop and thoroughly test medical device software. To validate and improve its functionalities, REALM tests it using 5 use cases. These use cases involve the application of AI-based algorithms to address various unmet clinical needs:
- DuneAI (UM)– Automated deep learning segmentation software for non-small-cell lung cancer in CT scans.
- COPowered (COM) – AI prediction of hospitalisation risks in COPD patients using patient-reported outcomes.
- Pharmacogenomics Passports to Practice (PGx2P) – Stratified pharmacogenetic testing to reduce adverse drug reactions and improve prescription accuracy.
- The STAR project (UoL) – ICU decision-support tool for blood glucose regulation in critical care settings.
- AI models for COPD and ASTHMA risk stratification (ASCOPD/TRAQBEAT) – Predictive biomarkers and sensor-based monitoring to improve inpatient outcomes and reduce hospital stays
The meeting included a session on communication, upcoming conferences for networking, and the REALM Academy’s ongoing training initiatives, all key components for widening the project’s impact.
As REALM moves towards the final year, the consortium is focused on strengthening stakeholder engagement, enhancing demonstrators, and finalising exploitation strategies. The goal remains clear: to deliver AI-driven tools that improve patient outcomes while adhering to the highest standards of safety, transparency, and trust.