AI-PREMie
A smart approach to predicting pre-eclampsia
Suppose there was a condition that arose during pregnancy and killed around 50,000 pregnant mothers and half a million babies around the world each year. Suppose that doctors had no more sophisticated ways of diagnosing this condition than taking the mother’s blood pressure and looking for protein in her urine, a sign of kidney damage.
There’s no need to suppose, because this is the case with pre-eclampsia, a dangerous and sometimes life-threatening condition that affects one in 12 pregnancies. In severe cases, babies need to be delivered far ahead of their due date, sometimes with profound implications for their health. And in the worst cases, babies and mothers may lose their lives.
If doctors could predict the course of the condition, it would help them to support the mother and developing baby on that journey, improving health and saving lives. Developing such a test is the aim of the Science Foundation Ireland-funded AI-PREMie project, which is using Artificial Intelligence to analyse blood samples from mothers with suspected pre-eclampsia.
“At the moment there is no way for doctors to be able to predict how pre-eclampsia will affect a mother and baby,” says biomedical scientist Professor Patricia Maguire, who works on AI-PREMie at University College Dublin with Consultant Haematologist and Full Clinical Professor Fionnuala Ní Áinle.
With seed funding from the SFI Future Innovator Prize, the AI-PREMie team began recruiting pregnant women onto the study in maternity hospitals in Ireland. As part of the work, the researchers spoke to many women who had experienced the condition, recalls Professor Maguire, whose own research focuses on platelets, a component of blood.
“I have been studying blood and platelets for almost two decades as a scientist, but this was the first time I was speaking to patients as part of my work,” she says. “We heard the stories of mothers who had lost their babies because of pre-eclampsia, and it had a profound effect on me and on the team in general, we knew we needed to do more.”
The project gathered pace, bringing together experts in obstetrics across the three maternity hospitals in Dublin, including Professor Mary Higgins at the National Maternity Hospital at Holles Street as a societal champion on the project; lab diagnostics, biochemistry and Artificial Intelligence. Industry partners include Microsoft and SAS Institute.
“We are using machine learning to analyse the blood samples from women with suspected pre-eclampsia,” explains Professor Maguire. ”Then by monitoring changes in their blood, particularly their platelets, and correlating them to the clinical data, we are finding patterns that could underpin new tests.”
AI-PREMie won a special prize in the SFI AI for Societal Good Challenge, and received €500,000 to continue the work. “We can’t thank SFI enough for the support, and not only for the funding,” says Professor Maguire. “We also had invaluable feedback from our reviews for the Prize, including the insight that we needed to carry out economic analysis of the impact of pre-eclampsia and the technology to predict it.”
That led to UCD's Professor Gerardine Doyle joining the team, as an expert in value measurement in health care delivery and the measurement of health literacy. Gerardine is now measuring the value of AI Premie, by measuring the health outcomes of the mother and infant alongside the measurement of the financial and societal costs of the lived experience of pre-eclampsia. Such costs include the lost work hours of the mother and her partner, the cost of travelling to hospital for extra tests and consultations and the tangible and intangible costs associated with the mother and infant needing hospitalisation, transfer to a Dublin maternity hospital and the costs to the wider family of such hospitalisations.
“In just three years we have gone from an idea to this multi-disciplinary project that is looking not only at the test that can help doctors but also the impact on the patient and economies too,” says Professor Maguire.
And it’s not stopping there. The AI-PREMie team is also looking to uncover the signals in blood of pre-eclampsia earlier in pregnancy, long before the standard tests could pick up on changes in the mother’s blood pressure or urine. “We are lucky in that our colleague Dr Neil O’Gorman at the Coombe has access to samples donated by pregnant women early in pregnancy and we are starting to apply the technology there,” says Professor Maguire. “If we can pick up very early in pregnancy who is at risk and who needs special care to safeguard their pregnancy, the impact will be enormous.”