Early Gestational Diabetes Mellitus Diagnosis Using Classification Algorithms: an Ensemble Approach

EasyChair Preprint 10820

Abstract

In the next twenty years, type 2 diabetes may affect over 50% of GDM patients, and infants and adults can acquire the disease. It is critical to consider both the mother's and the children's short-term and long-term challenges. In the current situation, early diagnosis is essential due to maternal morbidity and mortality and fetal problems. Early identification and prevention are inefficient and frequently problematic in developing and underdeveloped nations. Determining GDM requires a well-designed approach, which is urgently needed. This study's primary objective is to forecast GDM in the first trimester. To determine if a pregnant woman is at risk of GDM or not. This research used KNN, LR, and RF for classification and an ensemble (majority vote) model for final models. After an error-free validation, the data was passed to the machine learning pre-trained model file, which in turn returns the predicted value to the frontend-designed with HTML and CSS. Python serving framework was used to connect the frontend code with the model. The ML model file, web codes, and flask codes were uploaded to the Github repository in preparation for final deployment on the server. The codes on Github were connected to Heroku where the web application is hosted. The user's interface with the web application to access the pre-trained model to make predictions.

Keyphrases: Algorithms, Classifier, Diabetes, Ensemble, diagnosis

@booklet{EasyChair:10820,
  author    = {Oluwafemi O. S. Abe and Olumide O. Obe and Olutayo K. Boyinbode and Biodun N. Olagbuji},
  title     = {Early Gestational Diabetes Mellitus Diagnosis Using Classification Algorithms: an Ensemble Approach},
  howpublished = {EasyChair Preprint 10820},
  year      = {EasyChair, 2023}}