Pre-eclampsia affects 5 out of 100 pregnant women, and the outcome can be fatal for both mother and baby. According to recent studies, AI can help in prevention. We talked to Barbara Costantini, Associate Professor of Gynaecology and Obstetrics at UniCamillus University
Pregnancy is a complex and delicate period, in which joy at the imminent birth of a baby is combined with fear, doubt and, unfortunately, health risks for both mother and foetus.
One of the most feared health risks is the so-called pre-eclampsia, i.e. the ex-novo onset or worsening of a hypertensive condition, with the presence of proteinuria. Generally, pre-eclampsia occurs after the 20th week of gestation. If left untreated, pre-eclampsia can develop into eclampsia—the presence of generalised convulsive seizures in pregnant women.
Pre-eclampsia occurs in 4.6% of cases, and eclampsia in 1.4%. Women all over the world can be affected, and the outcome can be fatal, so much so that it is considered a major cause of perinatal and maternal morbidity and mortality.
In addition to the most feared event, there are also other complications related to pre-eclampsia: among them, placental abruption, pre-term delivery, neurological problems, intracranial haemorrhage, rupture of the liver capsule, renal failure or cerebral venous thrombosis.
But how do we know if an individual is at risk of being affected by pre-eclampsia? Diagnosis is based on measuring blood pressure and protein in the urine, and other tests to assess whether there is organ damage (liver, kidneys, lungs). Symptoms that should be investigated include headaches, nausea, visual flashes, hand tremors, excessive weight gain.
Causes, prevention and treatment
The causes of pre-eclampsia are not yet fully known, but there seem to be risk factors. “These factors include the presence of pre-existing conditions in the mother (diabetes, autoimmune diseases, chronic hypertension, kidney disease), the presence of twin pregnancy, obesity or advanced maternal age”, says Barbara Costantini, Associate Professor of Gynaecology and Obstetrics at UniCamillus University.
This shows how crucial it is to anticipate the extent of the risk of incurring pre-eclampsia. “Prevention is absolutely crucial, considering that this condition can lead to even serious complications for mother and child”, Prof. Costantini recommends. “Indeed, there is an increased risk for the foetus of growth restriction or reduction of amniotic fluid, while the mother is at greater risk of developing other severe pregnancy conditions, such as eclampsia, HELLP syndrome (H: haemolysis, hypertransaminasemia; EL: elevated liver enzymes; and LP: low platelet count) or placental abruption, up to and including possible lethal outcomes for both mother and foetus”.
Of the currently known treatments, Professor Costantini states that “antihypertensive therapy is of paramount importance, but optimisation of lifestyle and nutrition and careful maternal-foetal monitoring should also be considered”.
AI contribution
According to the latest e-Health news, Artificial Intelligence can help, also thanks to the excellence of University College Dublin and the technology of the ITC company SAS.
The team of researchers from University College di Dublino, funded by Science Foundation Ireland, researched the diagnostic techniques for pre-eclampsia. They also discovered important predictive blood biomarkers which, when combined with other clinical data, can provide a model that instructs AI to compile a stratified risk profile of pregnant women. This data analysis and correlation work was therefore entrusted to new AI software: the tool used to develop a risk stratification prototype is called AI_PREMIe, and uses SAS Viya’s cloud analytics platform, hosted on Microsoft Azure.
AI_PREMIe communicates the risk by means of red, yellow or green flags, thus predicting whether the patients analysed will develop a severe, medium or mild form of pre-eclampsia. It will be piloted in Dublin maternity hospitals by the end of the year and then rolled out globally in the future.
What are the biomarkers that could be predictive? “Let’s say that, during the first trimester, it is possible to carry out both ultrasound evaluations in association with the use of Doppler, and serological evaluations (with the dosage of various anti-angiogenic and angiogenic factors and numerous other serological indices). These have aided the development of several screening programmes with multifactorial algorithms that are especially useful in identifying individuals at risk of developing early-onset pre-eclampsia. This gives us the opportunity to treat the condition, even with antiplatelets”, comments Prof. Costantini.
We also asked Prof. Costantini whether, apart from AI_PREMIe, there are other AI technologies used in gynaecology that have proven to be effective. “AI in medicine is finding increasing use in terms of experimentation. In the gynaecological field, in recent years, AI has been used mainly in the development of management algorithms, software, and in the elaboration of diagnostic hypotheses on the basis of ultrasound and radiological images acquired. At present, compared to its potential, there is still limited use of AI, but studies in this regard are continuously being designed”, Prof. Costantini concludes.