Originally published on September 24th, 2020,
The UCD Institute for Discovery, in partnership with the Analytics Institute of Ireland and in association with UCD Research, hosted a webinar entitled, “The Future of AI in Healthcare”. This event was co-hosted by Professor Patricia Maguire, Director UCD Discovery and Lorcan Malone, Chief Executive at the Analytics Institute and featured expert speakers and panellists from academia and industry. This event also launched UCD Discovery’s new AI Healthcare Hub. In case you missed it, here are our top takeaways.
Peter Keeling is founder and CEO of diagnostics commercialisation company, Diaceutics Ltd.
Peter has driven this hugely successful Belfast-based company from a start-up fifteen years ago to IPO last year. With a data lake of some 300m people, Diaceutics has a “laser focus” on diagnostics throughout the “patient journey” where there are still “significant hurdles” to overcome.
The diagnostic journey for a lung cancer patient “going from cough to cancer medication” can take up to ten tests over a four year period The prevailing “myth” that diagnostics are less valuable than treatment means more than half of patients on lung cancer treatment in the US have not been given a lung cancer diagnosis. Peter believes better testing will be “either equal to or greater than” treatment in terms of delivering radically better patient outcomes.
The pandemic has seen cancer testing plummet by 30% “in all of the key markets from China to the US”. He predicts “Covid-19’s third wave” will be a healthcare system challenged further by a backlog of cancer admissions.
Dr Paulina Szklanna is Manager of UCD AI Healthcare Hub, an interdisciplinary incubator whose mission is the democratisation of AI to accelerate projects and deliver meaningful, translational outcomes across UCD, Ireland and globally in both academia and industry.
UCD AI Healthcare Hub is up and running with the aim of putting impactful algorithms into production on a number of exciting projects. One such project, in collaboration with Dublin’s Mater Misericordiae hospital, involved gathering the data of all severe Covid-19 patients admitted during the first wave of the virus. After preparing this data and subjecting it to a cloud-based platform for analysis, the Hub identified the “three most important variables” for predicting disease severity.
“Almost 97% of the time we can predict how severe Covid-19 may get which is an amazing achievement and we are now working to get more data from Covid patients so that we can strengthen this algorithm and hopefully put it into production so that it can make a real-world difference.”
Other projects underway at the AI Healthcare Hub include AI_PREMie, a tool to predict preeclampsia severity in pregnant women, and, in collaboration with UCD Centre for Physics in Health and Medicine, a risk-stratification calculator for hamstring injury in athletes.
Pearse Keane is Consultant Ophthalmologist at Moorfields Eye Hospital, London and Associate Professor at UCL Institute of Ophthalmology.
In 2017 ophthalmology surpassed orthopaedics as the busiest medical speciality in the NHS. In 2016 Pearse initiated a formal collaboration between Moorfields and the AI company DeepMind, sharing 1.2m anonymised retinal scans with them. Together they are developing a deep learning system that will pinpoint and prioritise the most severe cases. “What we have done is published a proof of concept in a research journal and now we’re on what I’ve come to learn is the much more challenging part of the journey, which is seeing how we can implement that algorithm”. He describes this next stage as “going from code to clinic” and was “very excited to hear Paulina talk about democratisation of machine learning and AI in healthcare” because “actually it has to work in the real world to really matter”.
Panellists in discussion with Prof Patricia Maguire:
Prof Walter Kolch, Director of Systems Biology Ireland (SBI) at UCD said there are “numerous obstacles” on the route to turning an algorithm into an actual application because “the medical curriculum doesn’t include any training in AI or data analytics. As human beings we often find it difficult believing in something we don’t understand. So there is a psychological barrier”.
In a discussion about ethics, Dr Ashwini Mathur, Head Data Science & Artificial Intelligence Hub Dublin at Novartis, likened AI today with the invention of the x-ray in the 1950s. “So the ethics are pretty straightforward. All the responsibility and accountability is yours. Do not even think about blaming it on the data.”
Oonagh O’Shea, Managing Director at Accenture Applied Intelligence believes that people are the key to implementing AI in healthcare because they “...want the benefit of personalised experiences but they’re sceptical of the methods of delivery. [But] we’ll see them owning the experience. We’ll equip patients to manage their treatments and their health.”
This article was brought to you by UCD Discovery - fuelling interdisciplinary research collaborations.