‘AI can only solve real-world problems with the right data’


12 Nov 2024

Indrakshi Dey. Image: Patrick Browne

People think AI can solve problems instantly, says SETU’s Dr Indrakshi Dey, but the reality is much more complex.

Increasingly, scientists are coming to understand the impact of modern environments on human health. For Dr Indrakshi Dey, studying the links between factors such as pollution and artificial light, and the contraction of non-communicable diseases is “urgent” in today’s society.

Her research “addresses the growing and often overlooked link between environmental stressors and non-communicable diseases, such as ocular and dermatological conditions”, she explains.

“By linking the outcomes of my research to tangible societal benefits – such as improving healthcare outcomes through non-invasive diagnostics or understanding the effects of pollution on human health – I make the work more relatable to the general public.”

Dey is head of the division for programmable autonomous systems in the Walton Institute at South East Technological University (SETU).

She completed a MSc in wireless communications at the University of Southampton in the UK and a PhD in electrical, electronics and communications engineering at the University of Calgary in Canada.

She has worked at a number of institutions including Trinity College Dublin and Maynooth University, and was a Marie Skłodowska-Curie Fellow at the Connect Research Centre for Future Networks and Communications.

‘My passion lies in pushing the boundaries of science and technology while fostering collaboration and mentorship within the academic community’

Tell us about your current research.

Currently, I am beginning a project called ENACT funded by Horizon Europe, which investigates the impact of air and light pollution on pre-clinical markers of non-communicative ocular and dermatological diseases.

This research emerged from growing concerns about environmental stressors such as artificial blue light and UV radiation, and how they contribute to conditions such as retinal degeneration, skin cancer and other non-communicable diseases.

The 42-month project will focus on how prolonged exposure to these pollutants accelerates oxidative stress and inflammation in both ocular and skin cells.

The project will incorporate AI-driven approaches to predict and monitor the progression of these conditions non-invasively. We are using cutting-edge techniques such as Bayesian modelling and Transfer Learning through Space-Time Graph Neural Networks (ST-GNN) to analyse the causal pathways between environmental factors and disease progression.

The research aims to offer early detection methods that can be more accessible and predictive, providing insights not only to healthcare professionals but also to public health planners and insurance providers.

The interdisciplinary nature of this project, combining data from environmental monitoring with healthcare diagnostics, has allowed it to expand into a comprehensive study on preventing disease through better understanding of environmental impacts.

In your opinion, why is your research important?

The potential impact of this research is significant. By developing non-invasive, AI-powered prediction and monitoring tools, we can shift from reactive to proactive healthcare. Early detection means patients can receive timely interventions, potentially preventing the progression of debilitating conditions.

Additionally, the environmental data collected can inform public health policies and urban planning, helping mitigate the root causes of pollution-related health issues.

This research could lead to more effective strategies for reducing the incidence of these diseases, ultimately improving the quality of life for many people while reducing healthcare costs globally.

What inspired you to become a researcher?

My journey into research was fuelled by a deep curiosity about how technology can solve real-world problems and improve people’s lives. One specific memory that stands out is from my time as a student, working on a project involving wireless communication systems and how communications systems form the backbone of our society. No one can live today without their smartphone. I was fascinated by the complexity of signals and how data could be transmitted across distances with minimal interference. It was then that I realised the power of research in creating solutions that, while technical, could have profound impacts on society.

Another pivotal moment came when I saw the direct impact that advanced algorithms and predictive models could have on healthcare systems. The idea that technology could not only drive efficiency but also save lives sparked a passion in me to push the boundaries of what is possible, particularly in areas such as AI, data modelling and network optimisation.

From that point, I knew I wanted to devote my career to exploring how these technologies could address larger challenges, such as improving health outcomes and managing environmental risks.

What are some of the biggest challenges or misconceptions you face as a researcher in your field?

One of the biggest challenges I face as a researcher is the misconception that advancements in AI and data science are quick, easy or automatic. Many outside the field believe that with enough data, AI can solve problems instantly, but the reality is much more complex. Developing reliable models, especially for sensitive areas like healthcare, requires significant effort to ensure data quality, model accuracy and ethical use. It’s not just about having vast amounts of data; it’s about having the right data and carefully training models to avoid bias, ensure privacy and provide meaningful, actionable insights.

Click here to listen to Future Human: The Series.

Another challenge is bridging the gap between technology and its real-world application. Many brilliant technological solutions remain in academic or theoretical stages because of hurdles such as lack of representative data, difficulties in clinical trials or challenges in working with interdisciplinary teams.

Implementing AI-based solutions in healthcare, for instance, requires collaboration between data scientists, medical professionals and policymakers – a process that can be slow and complicated.

Additionally, there’s sometimes scepticism or fear around AI in healthcare, with concerns about replacing human judgment. It’s essential to clarify that AI is a tool meant to support, not replace, healthcare professionals, providing them with deeper insights and more accurate predictive capabilities. Overcoming these challenges involves continuous dialogue and education to build trust in the benefits that AI and data science can bring to healthcare and other fields.

Do you think public engagement with science and data has changed in recent years?

Yes, I believe public engagement with science and data has significantly changed in recent years, especially in the wake of the Covid-19 pandemic. The pandemic highlighted how crucial scientific research, data analysis and real-time information are in managing global crises.

People became more aware of the importance of reliable data for making decisions that affect public health and safety. Concepts like infection rates, vaccine efficacy and statistical modelling became part of everyday conversations, which was rare before.

There has also been a noticeable shift in how the public perceive and interact with science. With the surge in data-driven decision making during the pandemic, more people have come to realise the value of scientific research and the critical role it plays in society.

However, this period also revealed some challenges, such as the spread of misinformation and mistrust in scientific data, which can complicate public engagement. This makes it even more important for researchers and scientists to communicate clearly and transparently, ensuring that complex concepts are accessible and understandable to everyone.

Additionally, the rise of digital tools and platforms has made science more open and interactive. People now have access to scientific reports, data visualisations and real-time research updates, which has created opportunities for citizen science and greater public involvement in scientific discourse.

I believe this increased engagement is a positive development, as it fosters a deeper understanding of science’s role in addressing societal challenges and encourages collaboration between the scientific community and the public.

Don’t miss out on the knowledge you need to succeed. Sign up for the Daily Brief, Silicon Republic’s digest of need-to-know sci-tech news.