Pre-eclampsia is one of the leading causes of maternal mortality and morbidity on a global scale, with an incidence during pregnancy of between 2% and 4%.
A new model, called PIERS-ML, is based on the use of easily accessible data and can accurately predict adverse outcomes within two days of the initial pre-eclampsia assessment.
This model was developed through the use of machine learning, multiple imputation and ten-fold cross-validation techniques on a development dataset that included more than 8,800 patients from 11 countries with different incomes.
The predictive capacity of the PIERS-ML model was assessed using the area under the R eceiver curve Operating Characteristic (AUROC). Compared to the currently used logistic regression model, the PIERS-ML has demonstrated greater accuracy, with an AUROC of 0.80 (5% confidence interval 0.76-0.84) compared to 0.68 ( 95% confidence interval 0.63-0.74) of the logistic regression model.
This model is able to classify women into risk categories, ranging from very low to very high, resulting in percentages of adverse maternal events within 48 hours of 0%, 2%, 5%, 26% and 91%, respectively.
In addition, the PIERS-ML model was validated using a dataset of 2,901 women diagnosed with pre-eclampsia at two hospitals in south-east England, confirming its ability to accurately classify women based on risk.
According to the study authors, the implementation of the PIERS-ML model could improve the identification of women with pre-eclampsia who are at risk of serious adverse events within two days of the initial assessment, thus providing more precise guidance to women, their families and maternal care providers.
The development of machine learning-based maternal risk assessment models could explore the inclusion of additional variables in the PIERS-ML model for further improvements in maternal risk assessment.
The study was funded by the Diversity in Data Linkage Centre for Doctoral Training at the University of Strathclyde, the Fetal Medicine Foundation, the Canadian Institutes of Health Research and the B ill & Melinda Gates Foundation.
read also: What if the theory of exponentials applied to health?
