Noel Kennedy (VetCompass, Royal Veterinary College, UK)

Can LLMs answer research questions about veterinary clinical notes?

Veterinary clinical notes contain narratives about individual patient care. In aggregate, these notes can tell us about the health of entire populations of dogs and cats but they are unstructured and currently require expert veterinary researchers to read and make sense of them. This is a time-consuming process that can take individual researchers months to process a few thousand cases. For example, if a researcher wanted to how Boxer dogs' health and disease patterns differed from the general canine population, they would need to read and annotate the notes of thousands of Boxer dogs and then to repeat this in a large group of control dogs. In these kinds of studies, the vast majority of the time of the researcher is spent reading notes of individual dogs rather than in analysis. In this talk I will discuss my recent work in evaluating if LLMs can rapidly answer clinical questions about patient care in a fraction of the time with the current manual method. This approach has the potential to vastly reduce the amount of manual labour required to study patients at a population level.

Noel Kennedy is a Senior Research Fellow in AI at the Royal Veterinary College, where he works on applying natural language processing to clinical veterinary data. With 25 years of experience in software development, including two decades in the veterinary domain, he has contributed to a range of impactful systems. These include the VeNom veterinary coding system, widely used across academia and industry, and the VetCompass application, which has supported over 100 peer-reviewed research papers in veterinary epidemiology. He now focusses on fine-tuning NLP models using one of the world’s largest datasets of veterinary clinical records, covering around 30% of UK veterinary practices. He holds a master’s degree in Advanced Natural Language Processing from the University of Cambridge.

Manoj Kulshrestha (Consultant Ophthalmologist, NHS Professionals, University of Buckingham)

How AI is Changing Medicine: a Simple Matter of Time?

In UK, the National Health Service (NHS) in England aims for most health and social care services to have a core level of digitalisation. All NHS hospitals are managed by trusts. However, not all 202 NHS trusts have the same level of digital maturity. This talk presents an action research study driven by an NHS trust. The study aimed to gain an understanding of information sources and digital health technologies used at the point-of-care by clinicians, particularly junior doctors (a.k.a. resident doctors). The study considered two points in time: before and after the arrival of ChatGPT on 30th November 2022. The study also investigated two AI approaches for transforming a body of evidence into knowledge graphs (symbolic AI) supporting question answering from junior doctors.

Mr Manoj Kulshrestha is an experienced consultant ophthalmologist with over 30 years experience. His expertise covers various areas of expertise, including medical retinal disease including age related macular degeneration, uveitis, retinal vein occlusion and diabetic retinopathy. Mr Kulshrestha is a fellow of both the Royal College of Ophthalmologists (FRCOpth), the Royal College of Surgeons of Ireland (FRCSI). He also holds a Bachelor of Medicine and Bachelor of Surgery (MB ChB) from the University of Birmingham. Mr Kulshrestha has actively contributed to medical education as deputy undergraduate tutor at Leighton Hospital. He also previously lectured in ophthalmology at the Bristol Eye Hospital. His research contributions are notable with over 85 publications in peer-reviewed journals, book chapters, and newspaper articles. He has delivered lectures at various levels, including international training sessions for optometrists in Oslo, Copenhagen, Stockholm, and Helsinki. Additionally, Mr. Kulshrestha has participated as a co-investigator in research studies concerning intravitreal medications and dry eye disease. In his career, Mr. Kulshrestha has held significant roles, such as Regional Director and Chair of Ophthalmology, Lead Clinician and Population Health Chair for Head, Neck, Eye, and Skin for the Hywel Dda University Health Board. His commitment to teaching, research, and patient care reflects his dedication to excellence in the field of ophthalmology. He is currently the Clinical Lead and Quality Improvement Lead for Ophthalmology at Leighton Hospital.

Hugo Lebredo (Universidad de Oviedo, Spain)

Towards Patient Autonomy: A Decentralized, Standards-Based Approach to Medical Data Sharing

This talk presents a proof of concept for medical data exchange across institutions through a decentralized architecture that places patients at the center of control over their health information. Without relying on central nodes or regulatory authorities, the model enables individuals to decide with whom to share their data. By leveraging medical standards and semantic technologies, the proposal promotes interoperability among heterogeneous systems while ensuring patient privacy, data security, and autonomy.

Hugo Lebredo is a lecturer in the Department of Computer Science at the University of Oviedo. His work focuses on medical data interoperability, with a special interest in the application of semantic technologies within the personal health record (PHR) paradigm, prioritizing the protection of patient privacy. He has presented his work at various conferences in the field and has published academic articles related to the exchange of medical-patient clinical information. He is currently conducting research on interdisciplinary projects aimed at improving clinical data integration through open standards and patient-centered architectures.

Dr Mercedes Arguello Casteleiro (BCS SGAI)

Hands-on Tutorial: Neuro-Symbolic AI

Current Artificial Intelligence (AI) "cannot reliably deal with facts, perform complex reasoning, or explain its conclusions". Healthcare organisations rely on "unstructured data such as doctors’ notes, X-rays, CT scans and research articles". Most of healthcare data (estimated 80%) is unstructured, which can be made more meaningful by classifying the data and normalising the data with internationally recognised standards, such as SNOMED CT as clinical coding standard.

Transforming the ever-changing body of scientifically sound research (evidence) from human-readable to machine-interpretable requires both biomedical/clinical knowledge and reasoning capabilities. Instead of using Large Language Models (LLMs) from neural AI as all-in-one generative AI solution, this talk investigates the plausible benefits from neuro-symbolic AI, combining neural AI (to process and extract patterns for health issues from unstructured data) with symbolic AI (explicit representations of background knowledge).

Come along to this hands-on tutorial if you are interested in:
* A comparison between open-source small-size biomedical/clinical LLMs and general-domain LLMs (DeepSeek, Groq, Grok3, QWEN2.5-MAX, Gemini, Claude, and ChatGPT4).
* How to lower the technical skills overhead (understanding of AI and programming code) needed to use open-source small-size LLMs.
* Representing diseases in an actionable (machine-interpretable) disease knowledge, which has a long-standing tradition in biomedical research.

Dr Mercedes Arguello Casteleiro has a PhD in Physics and is an elected committee member of BCS SGAI (the Specialist Group on Artificial Intelligence of the British Computer Society). She is investigating the benefits and drawbacks of combining Symbolic AI with Neural AI, including the fine-tuning of small/medium-size LLMs with specialised datasets of unstructured data (e.g. written narratives and biomedical images). She has worked as a researcher and lecturer in Electronics and Computer Science at the University of Southampton, and in the Department of Computer Science at the University of Manchester. She has also carried out research as part of the Bio-Health Informatics Group at the University of Manchester. The interest shown by her undergraduate and postgraduate students in AI has triggered her current work in low-code/no-code AI with LLMs, such as multimodal LLMs that can produce content (text, image, video, or audio/speech) as output (generative AI).

Prof Grigoris Antoniou (Leeds Beckett University, UK)

The use of AI in clinical decision support for neurodevelopmental conditions

In this talk, we cover our work on using AI as a clinical decision support tool for neurodevelopmental conditions, with a focus on adult ADHD and adult ASD. These works are done collaboratively with the one of UK's NHS Trusts and have resulted in developing new technological solutions. In addition, we review further AI technologies that we are currently exploring to enhance clinical decision support. These include speech analysis, video analysis, serious games, domain-specific ontologies and Generative AI.

Grigoris Antoniou is Professor of AI at Leeds Beckett University and Visiting Professor at the L3S Research Centre in Germany. His research interests lie in semantic technologies, particularly knowledge representation and reasoning and semantics for big data, and its application to health, law and smart cities. He has published over 200 technical papers in scientific journals and conferences. His research has attracted over 15,000 citations. He is member of the European Academy of Sciences and Arts, Fellow of IEEE, Fellow of the European Association for Artificial Intelligence and Fellow of the Asia-Pacific AI Association.

Dr Tannia Gracia (Spatial Genomic Atlas of Glioblastoma, Wellcome Sanger Institute, UK)

A Spatially Resolved Cellular Trajectory Framework for Glioblastoma Heterogeneity

Glioblastoma is an aggressive brain tumour with extreme cellular heterogeneity. Using deep single-cell and spatial multi-omic profiling across 12 tumours, we identify a conserved cancer cell trajectory from developmental-like to hypoxia- and injury-associated states. This progression defines spatial tissue niches and is consistent across patients and genetic subclones. A deep learning framework reveals fine-scale mixing of subclones within these niches and links cancer cell states to regional myeloid environments. We will leverage AI to model cell-to-cell communication within these dynamic spatial niches by integrating spatial transcriptomics data with histopathological features.

Dr. Tannia Gracia is a Senior Staff Scientist at the Wellcome Sanger Institute, where she leads data production for the Human Spatial Genomics Atlas of Glioblastoma. With a PhD in Toxicology from Michigan State University and a BSc in Pharmaceutical Chemistry from Universidad de Cartagena, her career spans over two decades of experience in cancer biology, transcriptomics, and translational research. Dr. Gracia has pioneered assay development and integration of spatial transcriptomics methods, driving biomarker discovery and target validation in oncology. She has held key scientific roles at the Milner Therapeutics Institute, Cancer Research UK, and the Cambridge Institute for Medical Research. Her current focus is on generating high-quality spatial omics data in brain cancer and neurodegenerative diseases, integrating multiple omic layers with clinical information. She collaborates closely with bioinformatics experts, providing clinically annotated datasets to enable among other initiatives, AI-driven modelling of cell-to-cell communication within the tumour microenvironment.

Prof. Juan Carlos Augusto (Middlesex University London)

Intelligent environments supporting health and wellbeing

We will overview the role of systems combining ICT infrastructure, with emphasis on sensing and actuation combined with Artificial Intelligence, in facilitating contextualized and personalized healthcare and well-being services. We summarize the state of the art and highlights challenges to widespread adoption.

Prof. Juan Carlos Augusto leads the Research Group on Development of Intelligent Environments within the Computer Science Department at Middlesex University London. The group has worked for more than a decade in the improvement of the processes to create context-aware systems with a range of applications, mostly, although not exclusively, in relation to health and well-being applications. Professor Augusto has led several participated and led projects funded by the EC and UK funders, and provided a range of services to the scientific community as journal and books Editor, organizing events and publishing research. For a list of publications see: Google Scholar entry, many available from JCAugusto-MDXResearchRepository