12 Apr 2021
STEM for Britain Award for Clinical AI Researcher
Dr Andrew Creagh is developing smartphone-based technologies and AI for monitoring multiple sclerosis patients
Postdoctoral Research Associate, Dr Andrew Creagh, was announced as this year’s Engineering Bronze place winner at the annual STEM for Britain awards event.
Dr. Creagh was credited for his work developing smartphone-based technologies and artificial intelligence (AI) strategies to monitor disease progression for people with neurodegenerative diseases, such as multiple sclerosis.
The STEM for Britain Awards
The STEM for Britain awards, usually held at the Houses of Parliament, is a major scientific poster competition organised annually by the Parliamentary & Scientific Committee since 1997. The aim of the event is to give members of both Houses of Parliament an understanding into the research being undertaken by early-career researchers across UK universities.
Initially selected as one of the top 10 engineering finalists from 85 applicants, Andrew was invited to present his research at the online Royal Academy of Engineering sponsored event to MPs and academic experts from around the UK. The judging panel awarded the STEM for Britain bronze award to Andrew in recognition of the original research exhibited in his thesis, “The Development of Digital Biomarkers for Multiple Sclerosis from Remote Smartphone- and Smartwatch-Based Assessments”.
We apply advanced signal processing, machine- and deep-learning techniques to improve our understanding of diseases
“Our group specialises in ‘AI for healthcare’, where we apply advanced signal processing, machine- and deep-learning techniques to improve our understanding of diseases, to develop better therapeutic strategies, and enhance patient care in the UK and the developing world”, he says. The CHI group has access to some of the world's largest, curated, anonymised healthcare datasets, and includes work with wearables and hospital data, across scales from the massively multivariate (including anonymised genomics) to the time-series data acquired from medical devices.
Developing Digital Biomarkers for Disease Progression
Andrew also undertook his doctoral degree at the department of Engineering Science in collaboration with multinational pharmaceuticals company F-Hoffmann La Roche, in order to develop novel methods for discovering biomarkers for neurodegenerative diseases using smartphones and smartwatches.
“Nowadays, everyone has a smartphone, and more and more people have smartwatches. These devices contain loads of different sensors which we can use to take measurements about a person’s daily life and health”, says Andrew, discussing his research. “We can then use these digital measurements to develop digital markers of disease: digital biomarkers”.
This is essentially giving clinicians an extra pair of eyes
The problem is that if you have a neurodegenerative or autoimmune disease, you might only see a clinician twice, or three times a year, Andrew points out. “This means we miss all the disease-related changes that occur when patients are not in the clinic. Instead, the idea is to try and measure a person’s disease symptoms using smartphones and smartwatches when they are at home”. Andrew adds, “this is essentially giving clinicians an extra pair of eyes” to measure aspects of a patient’s life they would normally be able to assess.
Clinical applications of machine learning (ML), often termed under the broader tag of artificial intelligence (AI), can act as powerful tools to learn complex and unseen digital patterns of disease. Andrew’s research utilises the latest deep learning algorithms to learn a patients’ disease severity from smartphone and smartwatch measurements, which he has shown can accurately monitor a patient’s MS symptoms over a 6-month period.
AI algorithms can transform smartphone measurements to predict multiple sclerosis patient severity.
Transparency in Clinical AI
Despite the capabilities of AI to learn unseen digital disease patterns, Andrew concedes that a lack of transparency in these so-called “black-box” models remains the largest stumbling block to the wider acceptance of AI for clinical applications. His research therefore aims to interpret the digital biomarkers developed through powerful deep networks using advances in Explainable AI (XAI). Emphasising the importance of interpretability in healthcare AI, Andrew explains that “clear communication of the data, the methods used, and the results obtained will play a vital role in developing our understanding of the capabilities of AI, improving our models, and building fundamental trust in their predictions”.
Research at the CHI lab
My research aims to build upon the concept of using smartphone and smartwatch technologies to develop digital biomarkers for people with neurodegenerative and autoimmune diseases
Andrew’s postdoctoral research at the CHI lab is sponsored by GlaxoSmithKline with collaboration from Professor Aiden Doherty and colleagues at the Big Data Institute. “My research aims to build upon the concept of using smartphone and smartwatch technologies to develop digital biomarkers for people with neurodegenerative and autoimmune diseases. It’s a really exciting field at the moment and we’re really just starting to see how new facets of habitual disease and disease phenotypes can be understood using smartphone- and smartwatch-based patient monitoring”.
Discussing the STEM for Britain competition, Andrew reflects, “I am honoured to receive the STEM for Britain Engineering Bronze award. I would really encourage all early-career researchers to apply next year. The STEM for Britain competition is not only an amazing way to showcase our research to non-specialists, and the opportunity of a direct line of communication into your elected MPs, but STEM for Britain is also a fantastic way to learn more about the ground-breaking research being conducted across Britain”.