This study introduces a deep learning architecture for brain tumor segmentation using multimodal MRI data. An adaptive attention mechanism enhances feature extraction, resulting in improved delineation of tumor boundaries, aiding radiologists in planning targeted interventions and tracking tumor progression.

Charlotte Vivian Hayes, Henry Theodore Morrison, Alice Margaret Rivers, Lucas Edward Brown, Georgia Nicole Armstrong, Thomas Julian Clarke

This research presents an explainable AI framework for diagnosing diabetic retinopathy through fundus imaging. Saliency heatmaps enable transparent model predictions, helping ophthalmologists understand key visual patterns while maintaining diagnostic accuracy and fostering clinical trust in computer-aided screening systems.

Daniel Scott Whitman, Isabelle Rebecca Carrington, Leo Patrick Doyle, Amelia Charlotte Richards, Nathaniel George Walker

This paper introduces a federated learning approach for collaboratively training models on distributed electronic health records. It ensures patient privacy while enabling predictive insights across hospitals without data sharing, supporting decentralized healthcare AI development with strong data protection measures.

Edward Martin Sullivan, Olivia Frances Whitmore, Felix Jonathan Brooks, Victoria Hazel Cunningham, Adam Wesley Norman, Chloe Elise Graham

This research proposes a hybrid deep learning model combining transformer and CNN architectures for early detection of Parkinson’s disease. By fusing voice and movement sensor data, the system enhances detection accuracy and offers new diagnostic pathways for neurodegenerative disorder monitoring.

Frederick Owen Baxter, Penelope Iris Lawson, Hugo Alexander West, Daisy Eleanor Kirkland, Zachary Louis Henderson

This paper explores unsupervised autoencoder models for detecting anomalies in wearable health devices. The system identifies deviations in physiological patterns, enabling real-time alerts and personalized recommendations that contribute to proactive and preventive digital health solutions in wearable ecosystems.

Harriet Rose Donaldson, Sebastian Lee Forrester, Eloise May Harrington, Callum Frederick Watts, Phoebe Alice Merton

This study employs gradient boosting to predict heart failure readmissions. The model integrates explainable risk scores, enabling clinicians to identify high-risk patients proactively. It aids personalized care plans and reduces readmission rates through interpretable predictions based on clinical and demographic data.

Clara Evelyn Matthews, Theodore James Huxley, Imogen Daisy Preston, Hugo Maxwell Blake, Abigail Louise Rowland

This paper presents a spatio-temporal deep learning framework for real-time seizure prediction from EEG data. The model extracts temporal and spatial features simultaneously, offering a non-invasive, reliable early-warning system that improves patient safety and enables responsive care interventions in epilepsy monitoring.

Oscar Vincent Hammond, Amelia Rose Thatcher, George Benjamin Faulkner, Eliza Mae Wilton, Thomas Edward Hollingsworth