Automatic Blood Glucose Classification for Gestational Diabetes with Feature Selection: Decision Trees vs. Neural Networks

Rigorous assessment of data mining algorithms in gestational diabetes mellitus prediction

Gestational diabetes mellitus (GDM) is the type of diabetes that affects pregnant women due to high blood sugar levels. The women with gestational diabetes have a chance of miscarriage during pregnancy and having chance of developing type-2 diabetes in the future. It is a general practice to take proper diabetes test like OGTT to detect GDM. This test is to be done during 24 to 28 weeks of pregnancy. In addition, the use of machine learning could be exploited for predicting gestational diabetes. The main goal of this work is to propose optimal ML algorithms for effective prediction of gestational diabetes mellitus and there by avoid it’s side effects and future complications. In this work different machine algorithms are planned to be compared for their performance in predicting GDM. Before analysing the algorithms they are implemented using 10 fold cross validation technique to obtain better performance. The algorithms implemented are Linear Discriminant Analysis, Mixture Discriminant Analysis, Quadratic Discriminant Analysis, Flexible Discriminant Analysis, Regularized Discriminant Analysis and Feed Forward Neural Networks. These algorithms are compared depending on performance measures accuracy, kappa statistic, sensitivity, specificity, precision and F-measure. Then feed forward neural networks and Flexible Discriminant Analysis are obtained as optimal in this work.

Prediction model comparison for gestational diabetes mellitus with macrosomia based on risk factor investigation

PURPOSE

To establish a feasible prediction model for gestational diabetes mellitus (GDM) with macrosomia based on risk factors analysis.

METHODS

A total of 1981 GDM pregnant women with macrosomia were enrolled in this retrospective study. The potential risk factors were revealed between the GDM women with and without macrosomia based on questionnaire and clinical data analysis. Then, prediction models including logistic regression (LR), decision tree (DT), support vector machine (SVM) and artificial neural networks (ANN) were constructed using these risk factors. Effect evaluation was performed based on model forecasting ability and model practicability such as accuracy, true positive (TP) rate, false positive (FP) rate, recall, F-measure, and receiver operating characteristic curve (ROC).

RESULTS

The risk factors analysis showed that factors such as triglyceride (TG), high-density lipoprotein-cholesterol (HDL-c) and ketone body were risk factors for GDM with macrosomia. Then, the forecasting model was constructed, respectively. Based on these risk factors as variables, the overall classification accuracy of the four forecasting models was 86%. DT model had the highest overall classification accuracy. SVM model had advantages over the other three models in terms of TP rate. Among the comparison parameters including overall ROC curve, ANN model was the highest, followed by LR model.

CONCLUSION

Among four forecasting models, ANN might be the optimal predication model, which had a certain practical value for the clinical screening of GDM women combined with macrosomia. Furthermore, HDL-c, TG, and ketone body might be potential risk factors for GDM with macrosomia.