This study presents a federated learning framework for securely training models across multiple medical institutions. It preserves patient privacy while enabling joint predictive analytics on heterogeneous biomedical datasets, supporting collaborative research in diagnostics, treatment planning, and population health management.

Michael Everett Coleman, Noelle Isabella Foster, Sven Henrik Johansson, Claudia Marianne Beck, Emiliano Tomas Gutierrez

This research develops explainable AI systems for analyzing clinical dialogue in mental health evaluations. Leveraging contextual embeddings and sentiment dynamics, the model enables transparent risk assessments for disorders like depression, anxiety, and PTSD, promoting ethical AI in psychiatry applications.

Jeremy Lucas Whitman, Sofia Elisabeth Kern, Theo August Mitchell, Hannah Grace Sinclair, Felix Emanuel Duarte

This study introduces a multi-view ensemble model integrating histopathological image features and genomic biomarkers for early autoimmune disease diagnosis. The approach enhances diagnostic sensitivity and enables early intervention, especially for systemic conditions such as lupus and rheumatoid arthritis.

Luca Francesco Barbieri, Isabella Nora Schmidt, Ethan Bradley Stone, Clara Josephine Delaney, Tobias Nathaniel Walsh

This paper proposes secure biometric-based authentication protocols for wearable and implantable medical devices. The approach ensures fast, energy-efficient identity verification, critical for emergency scenarios where rapid medical access and data integrity are vital for lifesaving interventions and trauma monitoring.

Nathaniel Scott Brewer, Vivienne Livia McAllister, Oscar Benjamin Frick, Helena Camille Durant, Elias Johann Krause

This research introduces a hybrid model combining vision transformers and graph neural networks for tumor segmentation in real-time MRI analysis. The model adapts to complex tumor morphology and outperforms traditional CNN baselines in speed, precision, and tumor boundary accuracy.

Frederik Andreas Mortensen, Amelia Ruth Hopkins, Daniel Leo Wilcox, Beatrice Noemi Ferreira, Julian Oliver Keane

This paper proposes a deep learning system to continuously monitor cardiovascular health using wearable sensors. The model adapts to individual baselines, detects arrhythmias early, and enables real-time alerting for medical intervention. It ensures precision health monitoring with minimal false alarms.

Gavin Isaac Thornton, Elise Victoria Monroe, Benjamin Claude Whitaker, Nora Madeline Cross, Jasper Lionel Forsyth

This research integrates biomedical ontologies and graph attention networks to identify repurposing candidates for existing drugs. The system analyzes molecular, phenotypic, and disease interaction embeddings to suggest new therapeutic uses, accelerating drug development while minimizing risk and cost in clinical trials.

Olivia Frances Vaughn, Christopher Owen Beckett, Madeleine Iris Sharpe, Hugo Felix Lindström, Claudia Rose Donnelly

This study presents a self-supervised learning framework for training biomedical image classifiers using limited labeled datasets. It extracts robust visual features from unlabeled samples, significantly enhancing downstream performance on rare disease datasets and low-resource medical imaging settings.

Henry Samuel Rowley, Abigail Louise MacIntyre, Viktor Emil Johansson, Natalie Fiona Walters, Elliot Patrick Granger