Several TRUSTroke consortium partners participated in a dedicated session at HEALTHINF 2026, the 19th International Conference on Health Informatics, organized by INSTICC. The session provided a platform to discuss the latest developments in trustworthy and explainable artificial intelligence and their application to healthcare, with a particular focus on stroke prevention, management, and follow-up.

The session opened with an inspiring keynote by Sara Zullino (EATRIS), who presented a multidisciplinary roadmap for the successful adoption of trustworthy AI in clinical practice. This was followed by a presentation from Luis Marte (Eurecat), who shared recent advances in explainable AI models for predicting stroke recurrence, highlighting the importance of transparency and interpretability in clinical decision support systems.

In the afternoon, attention turned to federated learning and emerging AI-driven approaches for healthcare. Luigi Serio (CERN) introduced the CAFEIN federated learning platform developed within TRUSTroke, while Andrea Protani (CERN) presented innovative privacy-preserving methods for brain tumour localisation using federated AI models. Pietro Caliandro (Fondazione Policlinico Universitario Agostino Gemelli IRCCS) explored new perspectives on AI-supported stroke management and its potential impact on clinical decision-making.

The session also featured contributions addressing a broad range of topics, including machine learning approaches combining flow cytometry and clinical data for early immunodeficiency classification, interpretability in large language models for digital mental health applications, the robustness of federated learning approaches for medical image analysis, and cybersecurity challenges associated with model sharing in federated environments.