BRAIN: Centre for Responsible Innovation in Big Data and Artificial Intelligence

Developing a proof-of-concept generative AI system to streamline cancer referral pathways for symptomatic bowel patients. This project integrates vision-language models for parsing referrals and risk-stratification to support faster diagnostic standards in cancer care. In collaboration with University Hospitals Birmingham and the West Midlands Cancer Alliance, we are exploring AI-driven automation in oncology diagnostics to improve patient outcomes and healthcare workflows. Read press article here. 

Building "REVEAL," a foundation model for surgical video understanding using self-supervised learning and vision-language modelling. The project aims to improve generalisation in surgical AI, enabling tool detection, segmentation, tracking, and report generation without extensive retraining. In collaboration with University Hospitals Birmingham and NHS Bristol Heart Institute, we are curating large surgical video datasets with semi-automated annotation to advance interventional care and surgical decision support. 

A feasibility study exploring how AI-driven extended reality (XR) technologies can transform surgical planning and postoperative care. In partnership with University Hospitals Birmingham, Bristol Heart Institute, and NHS Digital, this project investigated how a "surgical metaverse" and patient-specific digital twins can enhance clinical decision-making, surgical training, and patient outcomes. The study laid the groundwork for future AI-enhanced digital health solutions. Read news article here.

Developing AI-driven fetal ultrasound analytics for early detection and monitoring of congenital heart disease (CHD). This project leverages vision-language large models to assist sonographers in identifying cardiac anomalies in utero, aiming to improve diagnostic accuracy. In collaboration with the University of Bristol and NHS Bristol Heart Institute, we are exploring automated anomaly detection and predictive analysis for prenatal cardiology. Oracle Cloud resources support secure handling and sharing of 6,000 fetal scans. 

A modified Delphi study developing expert consensus on technical validation metrics, usability and safety measures, and reporting standards for AI-driven Mixed Reality (MR) in dentistry. The project addresses a critical gap: the absence of standardised metrics for geometric accuracy, AI segmentation performance, tracking latency, and human factors assessment hampers comparability across MR dental navigation systems and delays clinical adoption. Working with Prof. Khaled Q AlHamad at Qatar University's College of Dental Medicine, this study will convene 15-30 international experts across prosthodontics, oral surgery, endodontics, orthodontics, XR engineering, medical AI, and human factors. Through structured consensus rounds, we aim to establish validated benchmarks for implant navigation, guided preparation, orthodontic placement, and endodontic access. Reporting recommendations will align with CONSORT-AI, SPIRIT-AI, DECIDE-AI, and MI-CLAIM frameworks to enable trial readiness and regulatory alignment. 

CONCORD leverages large language models to enhance clinical documentation efficiency and guideline compliance in Cancer Multidisciplinary Team (MDT) settings. MDTs are critical for coordinating complex cancer care but are resource-intensive, costing NHS England an estimated £154 million annually. High case volumes and varied documentation practices pose challenges to consistency and efficiency, impacting patient care and waiting times. We are developing an agentic platform that automates three critical MDT tasks: synthesising multimodal patient data into concise overviews, providing real-time conversational analytics to highlight key clinical insights, and documenting actionable decisions as cohesive summaries integrated into hospital Cancer Registries. Covering all stages of the cancer pathway from diagnosis through treatment and surveillance, CONCORD creates a "digital twin" tailored to each patient's cancer journey. The platform will be evaluated against standard-of-care documentation, focusing on quality, time and cost savings, and real-world impact. This project exemplifies co-creation, with clinical leadership from University Hospitals Birmingham and technical development led by BCU's BRAIN centre.