Working at speeds 100,000 times faster than the average laptop and developed by the University of Bristol in close partnership with HPE and NVIDIA, Isambard-AI is able to process in one second what it would take the entire global population 80 years to achieve.
Our University of Bristol colleagues are already harnessing the power of Isambard-AI to progress incredible breakthroughs, propelling them to results faster than they ever imagined possible.
Below, we speak to three researchers based at University of Bristol who are using Isambard-AI to address critical healthcare challenges, across dementia, skin cancer, cardiac disease and more.
Professor Dima Damen – aiding independent living for dementia patients
My name is Dima Damen, I’m a Professor of Computer Vision at the School of Computer Science. Now, I lead a [research] group of 10 PhD students and two postdocs – all working on analysing footage from wearable cameras.
Tell us a bit about your project
Three years ago, we gave people cameras and told them to record their daily lives – which led to over 3,670 hours of footage from 900 participants all around the world. All shot point of view (POV).
Someone told me ‘Oh, this is like Peep Show’. They do a lot of POV.
We then wanted to look at all this data and try to predict what someone would do next, based on what they had already done – which would be very helpful for people with dementia.
They kind of slow down when they can't remember what they're supposed to do next. But, if you show them footage, from their own POV, of them doing the task previously, it really triggers and strengthens their memory to independently complete the task.
How are you using Isambard-AI?
We had the footage already, but had only been able to use a little bit as video is a lot of data – one minute is about a 100,000 pages of text.
Isambard-AI offers very powerful computation (millions of calculations at the same time) but also very powerful storage – so we can look a whole video of someone’s daily life.
We did something called digital twinning – where AI would take the POV footage and create a 3D digital model of the space and track where all the objects are in the space. All generated just from the footage.
So, you can answer questions like: 'Where did I leave my keys? Where did I put this? Did I put the remote control in the fridge?'
What are your planned next steps?
One of first things we would like to trial is applications for people with dementia. Filming their day with tech glasses – then using an AI model to find and show them relevant footage, in their glasses, when they need to complete a task.
With predicting what someone does next, the current research we're doing is on a much smaller data observes for two seconds and generates a prediction for the next one second. But now with more data, we wonder how much further we can predict – and would like understand anomalies or things that are not expected.
Dr. James Pope – understanding bias in skin cancer detecting AI
I’m James Pope and Senior Lecturer in Data Science. My research largely involves artificial intelligence and machine learning.
Tell us a bit about your project
The idea is that you might take a picture of a mole or skin lesion and then AI model would tell you whether it might be cancerous or not.
We wanted to see if the model was better for predicting cancer on darker skin or lighter skin, or if it was equally the same. Simply, was there tone bias which could affect the result?
We used about 4,000 images to train our model – that involved trying lots of configurations and combinations to make sure it works as well as possible. And you need a big machine to do this...
How are you using Isambard-AI?
It is a really big computer that’s much faster than a regular computer. Things that might have taken us five years to do, maybe Isambard can do in a couple of weeks.
I was surprised how fast it was when I first started using it. My first thought was something had gone wrong – but, I already had results and I hadn’t even finished my coffee!
We used Isambard-AI to tune the best possible model and then we went through lots of images of skin to see if it was biased or not. And the answer was yes – it did better for lighter skin tones than darker skin tones.
What are your planned next steps?
It’s clear next steps should involve addressing bias in AI models. I think work that we've done here, can hopefully, help us put mitigations in place – such as including a test for bias.
I don't know that we can always remove it because it may simply be that it's harder to diagnose a dark mole on a dark skin, but we need to be aware of it.
I would like for apps where I can easily share my health concerns with my doctor. In this case, taking a picture of a mole and getting a screening result. I realise there could be some errors, and I would still want to see doctor, but it would help my life.
Dr. Jon Lees and Dr. Danielle Paul – understanding the cell to understand diseases
My name is Dr. Jon Lees and I'm a lecturer in Bioinformatics. I've been collaborating with Danielle and lots of other experimentalist groups.
I'm Dr. Danielle Paul. I am a Senior Research Fellow, funded by the British Heart Foundation.
Tell us a bit about your project
Jon: We’ve been working together, with lots of other researchers, to understand how human cells works and are wired - in terms of their network of protein interactions.
Danielle: Proteins are the building blocks of the cell and have specific roles and functions.
Jon: They essentially link up together to do certain jobs and when that goes wrong, you get certain diseases. We’ve been looking at Alzheimer's, heart disease and cancer to understand the key proteins that cause them.
Before we can understand diseases, or how to fix them, we need to work out how the cell actually works. If you looked at a car engine, for example, you can't really fix cars unless you understand the mechanics and how the bits work together.
How are you using Isambard-AI?
Danielle: We can use Isambard-AI to predict protein structures and study changes that occur when you have a [disease causing] mutation.
Jon: Isambard-AI is kind of the computer of your dreams. It allows us to scale everything up and just try out new things that you couldn't before. We can construct more of the whole cell, not just bits, to properly understand it.
Danielle: The scale of what we can actually look at and what we can model in one go is much bigger. We can see larger protein complexes and more cell interactions with Isambard-AI.
Jon: It would take like 50 years on my own computer at home. These models are so big and, that's why they're so good, but we needed Isambard-AI's power – without it, we wouldn’t be doing the project.
What are your planned next steps?
Jon: We're looking at translating it to therapeutics (the treatment of diseases). Developing peptide inhibitor treatments, like Ozempic (for weight loss and diabetes), but for various other diseases. Also, how viruses interact with the human genome – which might be good for future Pandemic preparedness.
Danielle: I’m focusing on an inherited heart disease called Hypertrophic cardiomyopathy, which is the leading cause of sudden adult death in the young. There's a new class of drugs that reached the clinic last year which has been a real game changer. Eventually, I would love to have targets for therapeutics directed at other proteins central to the disease.
Jon: Isambard-AI has given us lots of interesting new avenues of research and we’re building a network of researchers across Bristol... were very excited for the future possibilities Isambard-AI will open up for health and wellbeing.
The Isambard-AI system has a great user experience. The team at the Bristol Centre for supercomputing (BriCS), who built and manage Isambard-AI, have done a terrific job and managed to deliver this amazing project really fast.