This research introduces a residual attention network for automatic segmentation of retinal layers from fundus images. The model identifies pathological patterns associated with early diabetic retinopathy, supporting ophthalmologists with faster, more accurate diagnosis and reducing manual intervention in medical image analysis workflows.

Isabelle Charlotte Redmond, Frederick Lucas Bentley, Georgina Mae Cartwright, Stanley Owen Hadley, Penelope Rae Wickham

This paper introduces a transformer model with temporal embeddings to predict hospital readmission using EHR data. The approach captures long-term dependencies in medical history, improving healthcare outcomes by identifying at-risk patients and enabling timely intervention strategies for chronic disease management.

Charlotte Elise Langdon, Gregory Henry Mansfield, Violet Aurora Prescott, Dominic Rhys Kensington, Beatrice Jane Standish

This study presents a federated ensemble learning approach to leverage distributed biomedical datasets for rare disease research. It maintains privacy while aggregating learning signals across institutions, increasing model robustness and supporting collaborative diagnostics without compromising patient confidentiality or regulatory compliance.

Julian Edward Blackwell, Claudia Ruth Barnett, Jasper William Radcliffe, Madeleine Faye Rowan, Sebastian Paul Ingram

This paper proposes a graph-based survival analysis framework using temporal event sequences from multimodal patient health records. The model predicts patient survival by modeling interdependencies among clinical features, offering explainable predictions for personalized treatment planning and outcome forecasting in medical applications.

Harvey Quentin Bellamy, Francesca Rose Underwood, Hugo Maxwell Pennington, Clara Louisa Fordham, Benedict Arthur Moreland

This research presents a deep variational embedding framework for multi-omics data integration in cancer subtype discovery. The model identifies latent representations that capture molecular diversity, enabling precise stratification of patients and supporting targeted therapeutic interventions in precision oncology workflows.

Imogen Thea Cartwright, Edward Blake Montrose, Tabitha Eleanor Stratton, Rufus Leonard Whitaker, Georgiana Iris Mansfield

This study introduces a semi-supervised deep ensemble method with consistency regularization for brain tumor classification in MRI scans. The model leverages unlabeled data to improve diagnostic performance, reducing annotation costs and supporting early tumor detection in resource-constrained clinical environments.

Dexter Anthony Rowland, Eliza Madeleine Thornton, Rupert Julian Blackstone, Harriet Louise Grantham, Oscar Vincent Kingsley

This research introduces an end-to-end pipeline for histopathology image classification using deep learning. The system automates feature extraction and clustering, enhancing accuracy in detecting cancerous tissue patterns and reducing reliance on manual interpretation in digital pathology diagnostics.

Matilda Florence Bexley, Tristan Edward Hollingsworth, Louisa Celeste Ackerman, Barnaby Felix Redgrave, Eloise Harriet Vickers