Modern medicine produces a huge amount of information, scans, lab results, doctor’s notes, and medical students can feel empowered but also overwhelmed. At the same time, AI is entering hospitals and we do not want students to treat it as a mysterious “black box”.
The project BRAIGED is changing that with xAIDA, a learning assistant that helps future clinicians practice diagnosing with AI - not to replace judgment, but to strengthen it.
The goal: graduates who are confident, careful, and ready to use AI responsibly in real clinics.
We start with kidney health because chronic kidney disease is common and depends on many clues. The result is faster, clearer learning and safer habits: students build confidence without overconfidence, reduce stress from information overload, and learn to spot bias or errors.
Also, we will keep it responsible - no identifiable patient data, strict protections, and a firm rule that humans stay in charge.
So when future clinicians say: “I understand what the AI is pointing to. Here is where I agree, here is where I do not agree, and here is my final decision”. That mindset is the heart of responsible, human-centered AI in healthcare.
Project description
The BRAIGED project aims to develop xAIDA, an explainable AI-driven educational diagnostic assistant tool, to investigate how AI technologies create novel prerequisites, opportunities, and challenges for learning, decision-making, and clinical reasoning in medicine.
Central to this project is the enhancement of AI literacy, enabling medical students and practitioners to critically evaluate and collaborate with AI in real-world clinical settings.
In the context of information-rich curricula, where students often face information overload and the fear of inadequacy, the project seeks to design and implement AI-driven learning tools that foster confident, competent, and safe practitioners, while mitigating adverse effects on learners’ mental health.
The primary objectives include integrating diverse data modalities (imaging, tabular, and textual) to predict and classify Chronic Kidney Disease (CKD), developing and refining machine learning models (e.g., CNN-based architectures and alternative segmentation techniques), and utilizing post-hoc explainability methods to transform AI diagnostics from a black-box process into an interpretable, meaningful, and relevant educational tool.
Key R&D challenges involve identifying structural and functional markers critical for kidney disease assessment, managing the complexities of dynamic and static renal scintigraphy, and overcoming data limitations through advanced transfer learning strategies.
Additionally, the project will explore experiential and simulation-based learning models to enhance clinical reasoning and diagnostic skills, grounding its approach in the paradoxical inquiry framework to embrace inherent contradictions in AI-driven medical training.
Ultimately, BRAIGED seeks to establish new interdisciplinary research directions and contribute to academic discourse on responsible AI integration into clinical practice.
