In September 2024, University of Bristol was crowned ‘AI University of the Year at the inaugural National AI Awards at Adastral Park science campus in Suffolk. It also won the AI Award for High-tech and Telecom at the same awards, for the REASON Open Networks Project.  

Such achievements do not come easily - the University of Bristol has long been considered at the forefront of research into Artificial Intelligence - a position consolidated recently by the University’s establishment of the Europe’s most powerful supercomputer, Isambard-AI.

The Jean Golding Institute, a multidisciplinary centre for data science and data-intensive research at the University of Bristol, coordinated the University’s nomination. It was managed by Patty Holley:

“This award represents the collective efforts of hundreds of researchers across diverse fields,” she said, “all of whom are working toward the goal of responsible AI innovation. Leading the charge in AI research is not just about breakthroughs; it's about advancing technology with a deep commitment to ethics and interdisciplinary collaboration, ensuring that progress benefits society as a whole."

The field of healthcare is one in which Artificial Intelligence shows considerable promise. The University of Bristol’s Elizabeth Blackwell Institute has funded a number of projects which are advancing the capabilities of AI in healthcare, and serve to help the University remain uniquely buoyant in this crowded research field. 

Powerful perspectives on drug development

One of the most immediate ways in which AI can facilitate healthcare is in drug development. Elucidating the incredibly complex interactions of possible novel therapeutic agents with cellular processes requires the vast computational resources that AI is capable of marshalling. 

Dr Luca Shytaj is the project lead on ‘Assessment of the integration of molecular docking with AI-based design of personalised vaccines’. The project uses a model the team developed during Covid called Custommune. Different algorithms can be used with Custommune to enhance its function, and the project assesses these. Complex computational models such as this could help develop personalised treatments much more effectively.

“Future medicine is envisioned as personalised, with most medicines or vaccines tailored to individual needs,” explained Dr Shytaj. “Our immune system identifies diseased cells through small protein fragments, akin to a game of Wordle. However, not all fragments are equally informative, and recognition varies based on genetics, like the information provided by letter combinations differs in various languages. We aim to combine AI computation and laboratory testing to design informative fragments for optimal immune recognition, using HIV as a model. If successful, our project could help in the development of more effective, personalised vaccines and immune-based treatments.”

Deep learning for Alzheimer’s

Of course, the potential for Artificial Intelligence in biomedical research isn’t restricted to identifying and improving vaccines. AI can also prove invaluable when attempting to determine the course of disease by looking at patterns in cell morphology. Lecturer in Data Science Dr Qaing Lui, from the School of Engineering, Mathematics and Technology, is investigating the implications for AI in Alzheimer’s disease, using AI to identify disease phenotypes in Alzheimer’s disease by looking at the shape and structure of neurons.

“Our research has the potential to make a significant impact, particularly in the realm of neurodegenerative diseases like Alzheimer’s,” said Dr Lui. “This deep learning-based cell approach offers a more accurate and efficient method for detecting subtle changes in neuronal structure caused by by toxic agents, which could, in time, have a variety of important societal and healthcare benefits.”

Dr Lui continued, “The model's effectiveness in detecting specific neuronal markers associated with disease states may lead to the identification of novel therapeutic targets and strategies, contributing to more effective interventions. Its accurate analysis of neuronal changes, in the presence of both disease agents and potential therapeutics, could streamline the drug discovery process and lead to safer and more effective therapies, which could improve patient outcomes. And the reduction in time and resource requirements may result in significant cost savings for research institutions and pharmaceutical companies.”

Cardiovascular disease and AI

Around 690 million people globally live with cardiovascular disease and the need for effective treatments has never been greater. Understanding the protein structures involved in these diseases is phenomenally difficult, but it could provide mechanisms for understanding disease states as well as normal, healthy functioning. Dr Danielle Paul and her team are using deep learning models to investigate the problem. 

“We are looking to use new advances in deep learning, in particular the University of Bristol’s new Isambard-AI,” said Dr Paul. “It will help us decipher 3D protein structures at the whole genome level. A key component of many inherited cardiovascular diseases are mutations in the sequence of amino acids used to build proteins which result in changes to the molecular structure.

“We’re using modifications of a deep learning model called Alphafold2, to assess the impact of common disease-causing mutations - and to develop new hypotheses about the inflammatory processes that are associated with heart disease.  

“Once established, we hope that these techniques will be applicable to other researchers; we hope to highlight how Deep Learning and the Isambard-AI resource can be used to support their work.”

Ethics of AI

As well as in treatment research, it is easy to envision a future where AI is used more in other key roles in the healthcare setting. Patient interactions, diagnosis, treatment, image analysis and basic administration; the potential uses of AI are vast. However, with such ubiquity come other, ethical challenges. 

Dr Emanuele Rati, Lecturer in Philosophy of Science and Artificial Intelligence, is investigating such issues with his team:

“Because AI tools increasingly shape how equitable and sustainable healthcare systems are, research on Responsible AI (RAI) has become a priority for governments, industry, and academia,” he said. “Despite significant advances, however, challenges remain. In particular, RAI considerations are often tailored to high-income countries, which neglects the host of social and cultural differences in different countries that impact how medical AI could be used.

New discipline

“Our work attempts to reframe RAI in healthcare in order to address this, with a method that will investigate how medical AI tools interact in context with specific personal, social, and environmental factors. It will also assess how AI can make healthcare more or less equitable and sustainable.

“We aim to reframe AI ethics as a discipline, in particular within the medical context, and in the long term, our research will reorganise how ethical considerations shape the design and implementation of medical AI tools."

AI for providers

Dr Helen Smith, Honorary Senior Research Associate in Bioethics, is investigating professional ethical guidance for healthcare AI use.

“Healthcare professionals look to their regulatory bodies (e.g. the General Medical Council, the Nursing and Midwifery Council) to direct their practice,” she said. “Artificial intelligence (AI) is being developed for use in healthcare in the UK. The problem is that there is no guidance to help healthcare professionals safely and fairly use AI in their work. Each of the different healthcare professions could make its own guidance for AI use. This could be confusing, however, as different professions might be given contradictory guidance.  

“Our research will help by identifying what could be in professional ethical guidance for AI use for all healthcare professionals (doctors, nurses, physios, paramedics, etc), taking a first step towards creating unified guidance.”

Professor Jonathan Ives is also working on Dr Smith’s project. “AI is likely to become ubiquitous in healthcare,” he said. “From supporting resource management, through diagnostics and triage, to more futuristic examples such as AI driven nanobots for cancer treatment or AI therapists.

“The challenge is not to prevent is use, but to ensure we can use it in a way that allows us to get the benefits of AI without introducing new risks and problems that are worse than the problems it is here to help us with. This is why the kind of work that we - and others - are doing, alongside the developers of AI technology, is so important.”

Artificial Intelligence is set to permeate our lives in myriad ways. It has vast implications for the betterment of society, and the streamlining of our healthcare systems is key to that. It has the potential to radically overhaul how we find therapies and cures for heretofore hard to treat diseases, but it also has the potential to ease the high burdens placed on healthcare providers. All of these will require careful research and oversight - which researchers at the University of Bristol are well placed to provide. 

AI in Health research community 

Elizabeth Blackwell Institute supports a growing AI in Health research community. We are bringing together researchers at the University of Bristol from different disciplines and external partners with an interest in AI in health.

Join this community and you will be:

• added to a mailing list to receive updates on funding opportunities, jobs, news and events 
• able to chat to other community members on Teams channels; find collaborators, ask questions, share ideas 

Find out more about this community or complete a form to join.