Artificial intelligence (AI), machine learning and big data are all offering breakthroughs in the healthcare and biomedicine sector. According to an Imperial College of London study, Rapid Automated Quantification of Cerebral Leukoaraiosis on CT Images, machine learning technology can improve dementia and stroke diagnosis, as well as offering personalized treatments.
Paul Michael Agapow, lead biomedical data scientist at the Imperial College of London, argues that data science is playing an innovative role throughout the biomedicine sector. Innovation Enterprise spoke to Paul ahead of the Big Data and Machine-Learning for Pharma Summit about the potential of machine learning and big data within the bioscience sector.
Innovation Enterprise: What is the importance of machine-learning technology and big data in the medical sector, and how has the adoption of big data and machine-learning technology disrupted the industry?
Paul-Michael Agapow: It can be shocking to understand how fragile our understanding of health and disease is, and how much of it is based on traditional beliefs and practices and ad hoc reasoning. It's no wonder – disease is complicated. Try to understand why one person is sick compared to another person's illness and you're propelled into untangling the differences in their genomes, life histories and environments.
There are millions of differences between you and the person next to you. Health-wise, what are the important ones?
This is fundamentally a data science and statistical problem. Thus, as computational power has skyrocketed, so has our ability to tackle these questions, turning biomedicine into a data-centric science. There is resistance to this as there is to any revolution. Medical systems and staff have to adapt and acquire new skills and capabilities. Statistical models have to be interpreted usefully into clinical actions. Uncomfortable truths have to be confronted because, under careful analysis, some widely-accepted treatments and drugs offer little or no benefit. If the data shows that patients with the same disease are different, treatment becomes more complex, as they have to be treated differently.
There is great power behind this revolution, data science is a force multiplier across all levels of biomedicine for all sizes of players. Drug design is more powerful and efficient with in silicio generation and testing of candidate molecules. Clinical trials are leaner and more informative. After so many years of talk about 'digital health' and disruption of healthcare, it is actually happening with startups elbowing their way into the market. It is an exciting time.
IE: In relation to the UK government's £1bn AI funding with the Imperial College of London, announced earlier this year, how has the university leveraged the funding and what research developments have been made so far?
PMA: The AI Sector Deal means that the UK government will fund 1,000 PhD studentships in AI. This is vital to keeping the UK and local industry competitive. There is a growing hub of expertise in London, but we have the constant problem of brain drain on two axes: From the university to the commercial sector from the UK and the US. It is too early to see the results yet, but Imperial has been very good at bringing together experts across domains to apply AI to tough real-world problems. We need the deal to keep the PhD "factory" rolling, supporting specialist courses like our Masters in AI and machine learning.
IE: What are the key topics that you will be discussing at Innovation Enterprise's Big Data and Machine Learning Summit?
PMA: I'm keenly interested in the pragmatic use of data science in biomedicine. There are all sorts of approaches and methods that look promising but falter when they meet reality. Human health isn't like building a bridge or solving an equation. Biology is messy and absurdly complex, and our experiments and analysis give us messy, complex, noisy data. I want to understand more about how we move forward despite this complexity and noise, and how to work with uncertainty and incomplete data.
To find out more about the latest data visualization trends and solutions, you can watch Paul-Michael Agapow speak at Innovation Enterprise's Big Data & Machine Learning for Pharma Summit in Madrid on October 10, 2018.
Book your place HERE