Automated Machine Learning (AutoML)-Derived Preconception Predictive Risk Model to Guide Early Intervention for Gestational Diabetes Mellitus

Abstract

The increasing prevalence of gestational diabetes mellitus (GDM) is contributing to the rising global burden of type 2 diabetes (T2D) and intergenerational cycle of chronic metabolic disorders. Primary lifestyle interventions to manage GDM, including second trimester dietary and exercise guidance, have met with limited success due to late implementation, poor adherence and generic guidelines. In this study, we aimed to build a preconception-based GDM predictor to enable early intervention. We also assessed the associations of top predictors with GDM and adverse birth outcomes. Our evolutionary algorithm-based automated machine learning (AutoML) model was implemented with data from 222 Asian multi-ethnic women in a preconception cohort study, Singapore Preconception Study of Long-Term Maternal and Child Outcomes (S-PRESTO). A stacked ensemble model with a gradient boosting classifier and linear support vector machine classifier (stochastic gradient descent training) was derived using genetic programming, achieving an excellent AUC of 0.93 based on four features (glycated hemoglobin A1c (HbA1c), mean arterial blood pressure, fasting insulin, triglycerides/HDL ratio). The results of multivariate logistic regression model showed that each 1 mmol/mol increase in preconception HbA1c was positively associated with increased risks of GDM (p = 0.001, odds ratio (95% CI) 1.34 (1.13–1.60)) and preterm birth (p = 0.011, odds ratio 1.63 (1.12–2.38)). Optimal control of preconception HbA1c may aid in preventing GDM and reducing the incidence of preterm birth. Our trained predictor has been deployed as a web application that can be easily employed in GDM intervention programs, prior to conception.

Approach and Method for Bayesian Network Modelling: A Case Study in Pregnancy Outcomes for England and Wales

“…When presented with a broader dataset, these solutions tend to identify the disease of interest in patients who actually have other conditions that may share even only a small number of similar symptoms, or present in academic papers with claims of unreasonably high degrees of accuracy from a very limited number of data points when tested with the author's training dataset. For example, in development of an automated ML model to identify gestational diabetes mellitus (GDM) [22] excluded women of black, white or mixed ethnicities, those who were found to have the clinically similar though not identical Type 2 Diabetes Mellitus (T2DM) in pregnancy, and those whose glucose tests were reported outside a narrow 4-week window from their dataset; and based on analysis of only four data points (HbA1c, mean arterial blood pressure, fasting insulin and the woman's HDL ratio) and testing with their own training dataset of 222 participants, the authors claimed an 'excellent' 93% accuracy for preconception prediction of future GDM.…”

Recent applications of machine learning and deep learning models in the prediction, diagnosis, and management of diabetes: a comprehensive review

“…In order to construct a preconception-based GDM predictor, game theory concepts were merged with genetic programming (GP)-based automated machine learning (AutoML) [ 82 ]. For this purpose, the Shapley additive explanations (SHAP) framework was combined with the GP-based Tree-Based Pipeline Optimization Tool (TPOT) to find significant features and choose optimal supervised machine learning models.…”

Stacking Ensemble Method for Gestational Diabetes Mellitus Prediction in Chinese Pregnant Women: A Prospective Cohort Study

“…e sum of the scores for the seven components was calculated for each participant, with a higher score indicating worse sleep quality. Sleep quality was classified according to the score achieved: poor (PSQI score of [16][17][18][19][20][21], general (PSQI score of [11][12][13][14][15], better (PSQI score of 6-10), and best (PSQI score of 0-5).…”

“…Stacking is a suitable method for integrating different types of base learners, which is less used in GDM prediction. Kumar et al proposed a stacked ensemble model with a gradient boosting classifier and SVM for predicting GDM risk, and their method achieved great performance with an AUC of 0.93 [ 17 ]. Their results demonstrated the effectiveness of stacking methods, but they used biochemical factors such as HbA1c, fasting insulin, and triglycerides/HDL ratio.…”

“…Stacking is a suitable method for integrating different types of base learners, which is less used in GDM prediction. Kumar et al proposed a stacked ensemble model with a gradient boosting classifier and SVM for predicting GDM risk, and their method achieved great performance with an AUC of 0.93 [17].…”