Developing a prediction model for HELLP syndrome through data mining

Abstract Background HELLP syndrome represents three complications of hemolysis, increased liver enzymes, and low platelet count. Since the causes and pathogenesis of HELLP syndrome are not yet fully known and well understood, distinguishing it from other pregnancy-related disorders is complicated. Furthermore, late diagnosis leads to a delay in treatment, which challenges the disease management. In this paper we aimed to present a machine learning attitude for diagnosing of HELLP syndrome based on non-invasive parameters. Method We conducted this cross-sectional study on 384 patients in Tajrish Hospital, Tehran, Iran, during 2010–2021 in four stages. In the first stage data elements were identified using literature review and Delphi method. Then patient records were gathered and in the third stage the dataset was preprocessed and prepared for modelling. Finally, machine learning models including network-based algorithms (Multilayer Perceptron, Deep Learning), ensemble algorithms (Random Forest and Adaboost) and classic algorithms (Decision Tree, Support Vector Machine and K-Nearest Neighbor) were implemented and their evaluation metrics were compared. Results 21 variables were included in this study after the first stage. Among all the machine learning algorithms MLP and Deep Learning had the best performance with the F1-Score of more than 99%. Based on the modeling output, some variables such as Platelet, Gestational-age, and ALT, were found to be the most important on diagnosis of HELLP syndrome. Conclusion This study showed that machine learning algorithms can be used very successfully in the development of HELLP syndrome diagnosis models. This study also showed that Biomarker features have the greatest impact on the diagnosis of HELLP syndrome.