UC researchers to develop AI-based delivery type prediction system

^{Translation: Diana Taborda}

A research team from the Department of Informatics Engineering (DEI) of the Faculty of Sciences and Technology of the University of Coimbra (FCTUC), together with the Faculty of Medicine of the UC (FMUC), is working on the development of a system that aims to predict, through computational analysis, the possibility/probability of a vaginal delivery after induction.

Induced labour is increasingly common these days, but it doesn't always result in a vaginal birth. This was the starting point for the research "Predicting Vaginal Delivery After Labour Induction with Machine Learning" by Iolanda Ferreira, a doctoral student in Health Sciences, supervised by Ana Luísa Areia, professor at FMUC, and co-supervised by João Nuno Correia, professor at FCTUC.

"All induced labour has a 30 to 35 per cent chance of ending in a caesarean, so we know in advance that 70 per cent of women will have a vaginal birth. However, if we could determine which of these 30 per cent would actually result in a caesarean section, we would be able to advise appropriately and proactively about the need for a labour induction, which is an arduous process for the mother and the foetus, and which can add to the emotional and economic burden associated with this procedure," explains Iolanda Ferreira.

Because it's such a common procedure that generates a lot of data, we thought we could use a technique to analyse that data to help doctors understand if or when it's worth inducing labour to try a vaginal delivery," Ferreira says.

According to João Nuno Correia, "the idea is to develop a system that, by combining data (tables and images), generates a support tool that provides tailored information on the probability of vaginal delivery after labour induction for each pregnant woman. If the likelihood is high, induction is performed with greater confidence. If not, i.e. if there is a very high probability of a caesarean section, the mother-to-be can be advised otherwise".

Although this is a topic that has been studied before, "the innovation in this research is to predict the type of labour by also using ultrasound image data. Clinicians rely on the woman's clinical history and her characteristics during the course of the pregnancy, and we want to see whether the system is able, by analysing this combination of clinical data and images, to identify a way to help reach a decision at a later stage," the researchers point out.

So far, the research team has analysed data from 2,600 women followed at the Coimbra Hospital and University Centre (CHUC), and the results are promising. The next step is to analyse the collected ultrasound scans and then create a tool with all the data and test it on human subjects.

According to the researchers, "this collaboration between DEI and FMUC is crucial because it will, in the future, support doctors in their decisions before delivery, thus making it possible to improve both neonatal prognosis and women's experience of childbirth".