A machine learning model developed by Chinese researchers could reduce the risk of maternal and fetal mortality.
- Pregnancy hypertension affects 5 to 10% of pregnant women.
- Thanks to LASSO logistic regression, it would be possible to predict the probability of onset of hypertension in pregnant women in the first trimester.
- Age and never having given birth increase the risk of suffering from abnormally high blood pressure during pregnancy.
It affects 5 to 10% of pregnant women, according to health insurance. Pregnancy hypertension is the most common health problem during pregnancy. This abnormal rise in pressure in the arteries occurs after 20 weeks of amenorrhea. It leads to complications in the future mother (pre-eclampsia, retroplacental hematoma) and in the baby (intrauterine growth retardation, prematurity). This is why it must be identified and taken care of quickly after being diagnosed to limit its consequences.
Hypertension and pregnancy: four algorithms compared
Recently, scientists from Shantou University Medical College (China) revealed that they have developed a machine learning model (a form of artificial intelligence) to predict the probability of onset of high blood pressure during pregnancy and reduce the risk maternal and fetal deaths. For this, they used the information (medical history, lifestyle) of 5,067 pregnant women between 7 and 18 weeks of pregnancy. None of them suffered from chronic hypertension. Next, the researchers compared four different predictive models: LASSO logistic regression, random forest (RF), neural networks and support vector machin (SUM).
Age and the fact of never having given birth would increase the risk of hypertension
According to their work, published in the journal Hypertension Research, the LASSO logistic regression was used to predict the probability of occurrence of hypertension in pregnant women in the first trimester. The results of the machine learning model appeared in the form of a nomogram, namely a graphical calculation tool made up of graduated curves between which a ruler is placed.
The authors revealed that the risk of pregnancy-induced hypertension at an age of 35 or over is five times higher than in younger women. Another observation: 21% of patients who had never given birth developed pre-eclampsia due to an increase in blood pressure. Clearly, this model provides a good prediction. However, the team clarifies that it still lacks external validation.