Climate art and allergy guidance: News from Imperial
Here's a batch of fresh news and announcements from across Imperial.
From climate art for murals across the UK, to allergy guidance for caregivers of those who are at risk of anaphylaxis, here is some quick-read news from across Imperial.
Allergy advice
Doctors have produced a patient information sheet for parents and carers who are looking after someone with a food allergy.
This follows research conducted at Imperial, which found only 60% of patients who had previous hospital treatment for food anaphylaxis – a serious allergic reaction which can be life-threatening – were prescribed adrenaline autoinjector "pens" like EpiPen or Jext.
Dr Paul Turner and Professor Jennifer Quint, from Imperial's National Heart & Lung Institute, examined over 130,000 NHS hospital records surrounding food allergy between 2008 – 2018.
The results prompted them to create the guidance, which they hope will save lives. It urges carers to seek a diagnosis, be aware of the symptoms of anaphylaxis, as well as providing advice on how to use adrenaline pens appropriately.
Dr Paul Turner said: “Anaphylaxis can be life-threatening, and must be treated quickly. The first-line treatment of anaphylaxis is an adrenaline injection into the leg muscle, and a call to the emergency services.
"We stress the importance of people with a food allergy to know how to use adrenaline pens. Our guidelines will help people who care for someone with a food allergy to recognise the symptoms of anaphylaxis, and how to treat it early.”
The data and information sheet were also reported on by BBC News.
AI model bias
An AI chest X-ray foundation model may not be fit for clinical applications due to racial and sex-related biases, a new study suggests.
The study, led by Dr Ben Glocker, aimed to assess the potential risks for using foundation models, which are trained on large datasets for many purposes, in the development of medical imaging AI tools.
To investigate any bias within the model, the researchers randomly sampled a set of 3,000 patients (1,000 samples each from three racial groups). Significant differences were found between male and female and Asian and Black patients related to disease detection.
Dr Ben Glocker, from Imperial's Department of Computing, said: “AI is often seen as a ‘black box’ – unexplainable in the way it works - but that’s not entirely true. With the right tools, we can open the box and inspect its features to prevent bias from creeping in. Model inspection is one way of continuously monitoring and flagging issues that need a second look. The work doesn’t start with the AI model, it starts with the data used to build it.
“As we collect the next dataset, we need to, from day one, make sure AI is being used in a way that will benefit everyone.”
The paper ‘Risk of Bias in Chest Radiography Deep Learning Foundation Models’ was published in Radiology: Artificial Intelligence, of the Radiological Society of North America.
Grantham Climate Art Prize
Three winners of the 2023 Grantham Climate Art Prize have been announced, as well as six runner-up designs, by the Grantham Institute – Climate Change and the Environment and sponsors Octopus Energy.
The three winning designs will be painted by professional artists as outdoor public murals in London, Coventry and Glasgow ahead of the United Nations Climate Conference COP28 starting in November.
Photographs of the finished murals and images of the runner-up designs will also be displayed on billboards across London including in public transport hubs in November.
In addition, a highly commended mural will be displayed in Nine Elms close to the newly revamped Battersea Power station with sponsorship from developer Ballymore.
Linsey Wynton, who managed the art prize for the Grantham Institute, said: “Young people will be most affected by the climate crisis and we asked them to depict their vision for a more sustainable future to inspire others to take steps to achieve this.”
This year’s entrants were asked to draw on the Grantham Institute’s 9 things you can do about climate change for inspiration.
Woman Scientist of the Month
Eleonora Moratto, PhD student in the Department of Life Sciences, has been named Woman Scientist of the Month by the European Platform for Women Scientists (EPWS).
Eleonora uses electric fields to slow the spread and symptoms of root rot in cocoa, durian and oil palm, caused by the Phytophthora palmivora pathogen.
Alongside her studies, Eleonora pursues ballet and combines science and art together – seen in her Dance your PhD choreography, which reached the final in a competition by Science Magazine.
Speaking to EPWS, Eleonora said: “I want to thank Professor Bernadette Byrne for acting as an informal mentor and, more recently, Dr Claire Stanley for believing in my ability to carry out my own line of investigation and showing me the path to independent academic research. I would also like to thank my PhD supervisors Dr Giovanni Sena and Dr Tolga Bozkurt for nurturing my love of science”
Mesothelioma modelling
A new machine learning model, developed by Imperial researchers, could detect subtypes of mesothelioma tumour cells.
Mesothelioma is an aggressive and deadly cancer in the surrounding tissue of the lungs and is usually caused by asbestos exposure. It has a poor prognosis with less than 10% 5-year survival rates as it is difficult to diagnose.
The model, developed by a multi institutional consortium of scientists and the National Heart & Lung Institute was tested on 234 tissue samples. It successfully predicted the presence of dangerous sarcomatoid components in local tissue regions.
Dr Jan Lukas Robertus, study author and Honorary Clinical Senior Lecturer at the National Heart & Lung Institute, said: “Mesothelioma is classified into three histological subtypes each with different prognoses. Being able to subtype mesothelioma is essential for accurate and early prognosis and better treatment. This study is a promising new step which can help us to understand the disease pathology and outcome.”
The paper 'Malignant Mesothelioma subtyping via sampling driven multiple instance prediction on tissue image and cell morphology data' was published in Artificial Intelligence in Medicine.
Article images: Shutterstock
Want to be kept up to date on news at Imperial? Sign up for our free quick-read daily e-newsletter, Imperial Today.
Article text (excluding photos or graphics) © Imperial College London.
Photos and graphics subject to third party copyright used with permission or © Imperial College London.