This study explores lightweight deep learning models for anomaly detection in biomedical signals from wearable devices. The approach ensures low-latency processing with high accuracy, enabling real-time healthcare monitoring for chronic conditions and early medical interventions without compromising computational efficiency.

Eliza Margaret Holloway, Thomas Nathaniel Burke, Chloe Isabelle Jennings, Daniel Arthur Maynard, Sophia Evelyn Clarkson

The paper proposes a transfer learning technique to enhance rare disease identification across institutional image repositories. It reduces training data requirements while increasing generalization, ensuring robustness across domain shifts and maximizing diagnostic accuracy for underrepresented medical conditions in diverse clinical environments.

George William Bateman, Fiona Catherine Rhodes, Henry Sebastian Liddell, Olivia Beatrice Henley, Edward Lawrence Drummond

This paper presents a blockchain-based model for securing electronic health records (EHRs) through immutable audit trails. The proposed mechanism enhances trust in medical record systems by logging access transparently and enabling tamper-proof tracking of sensitive patient data across healthcare providers.

Julian Francis Mercer, Harriet Anne Whitmore, Benjamin Luke Cartwright, Alice Joanna Rowley, Oliver David Pemberton

This research develops an ensemble model integrated with explainable AI techniques to interpret electrocardiogram (ECG) signals for cardiac risk assessment. The system enhances transparency in decision-making, allowing clinicians to validate model outputs and improve trust in AI-assisted diagnostic recommendations.

Charlotte Emily Morland, Zachary Edward Kendall, Victoria Rose Fenwick, Frederick James Barnett, Isabelle Grace Harcourt

This work introduces a deep graph neural network framework to model complex drug interaction networks. By capturing relational dependencies among pharmaceutical compounds, the system predicts adverse effects and uncovers novel interactions, enhancing biomedical informatics tools for safe and effective treatment planning.

Sebastian Hugh Templeton, Emily Grace Linton, Charles Matthew Radcliffe, Georgina Alice Middleton, James Arthur Penrose

This study proposes a federated learning model for analyzing biomedical data from decentralized sensors. The approach preserves patient privacy, reduces data transmission overhead, and enables collaborative learning across hospitals while ensuring model performance consistency and compliance with healthcare data protection standards.

Annabelle Lucy Whitfield, Dominic Andrew Holbrook, Eleanor Francesca Baines, Julian Patrick Hargrave, Nathaniel Edward Carrington

This paper introduces a knowledge-based natural language processing system to extract clinical insights from radiology reports. By leveraging medical ontologies, the model ensures accurate identification of findings, aiding diagnostic workflows and enhancing decision-making efficiency in healthcare informatics environments.

Frances Amelia Kingsley, Harrison George Beaumont, Lucy Charlotte Rutledge, Anthony Samuel Redford, Matilda Eleanor Sheffield