This paper proposes a hybrid architecture combining graph-convolutional networks and transformers for modeling patient pathways. The model leverages relationships across modalities in electronic health records to accurately forecast treatment outcomes and optimize care strategies for chronic disease management.

Clara Marianne Beckett, Isaac Benjamin Rowland, Lydia Frances Whitmore, Hugo Leonard Westbury, Martha Sophia Kellerman

This research introduces a federated learning framework enabling collaborative brain imaging analysis across hospitals. It ensures data privacy and enhances model generalization for neuroscience applications, including disease detection and cognitive state prediction from fMRI and PET scan datasets without centralized data pooling.

Maxwell George Ellison, Amelia Violet Carter, Nathaniel Charles Fleming, Zoe Isabelle Hamilton, Henry Thomas Westcroft

This study presents a reinforcement learning approach for automated insulin dosing based on continuous glucose monitoring. The system dynamically adjusts insulin recommendations, improving glycemic control and reducing hypoglycemia risks in Type 1 Diabetes patients using real-time feedback and personalized historical health data.

Frederick Simon Worthington, Lucy Madeleine Cowell, Sebastian Edward Trent, Eloise Charlotte Morton, Oliver Graham Dunstan

This paper explores a fusion model that combines wearable biosensor data and natural language processing from speech transcriptions to detect anxiety episodes. The approach supports early interventions by identifying physiological and linguistic markers in real-time within ambulatory or outpatient care settings.

Charlotte Elizabeth Henley, George Tobias Brigham, Isabelle Rose Falconer, Edward Henry Levington, Sophia Catherine Newstead

This study benchmarks vision transformers and convolutional neural networks for classifying skin conditions using smartphone images. The research evaluates accuracy, robustness, and inference speed, providing insights into deploying lightweight diagnostic tools for dermatological assessments in telemedicine and primary healthcare contexts.

Francesca Louise Godwin, William Edward Tredwell, Hannah Grace Pemberton, Joshua Benedict Carrington, Alexandra May Stafford

This research develops a deep learning system that classifies respiratory disorders using audio signals from wearable chest microphones. It captures subtle patterns in breathing sounds and correlates them with medical conditions to aid early diagnosis in non-invasive and continuous monitoring environments.

Harriet Amelia Bostwick, James Malcolm Featherstone, Eleanor Grace Rowley, Lewis Christopher Norton, Abigail Sophie Redmond

This study introduces a predictive framework that integrates multimodal data from surgical records, imaging, and patient histories to anticipate postoperative complications. The model supports risk stratification and proactive interventions, improving outcomes and reducing readmissions in surgical care across multiple health institutions.

Oscar William Standish, Rebecca Claire Morland, Dominic Lewis Whitaker, Fiona Rachel Langford, Edward James Appleton