Prediction of gestational diabetes using machine learning algorithms and deep learning

Project Synopsis & Overview

This project focuses on predicting Gestational Diabetes Mellitus (GDM) using machine learning and deep learning techniques based on clinical data collected during pregnancy. The main objective is to identify women who may be at risk of developing GDM at an early stage, thereby supporting timely intervention and improved maternal and fetal health management. The input data includes clinical features such as number of pregnancies, glucose levels, blood pressure, insulin levels, BMI, DiabetesPedigreeFunction representing family history, age, and outcome. The system performs data preprocessing through missing-value handling, feature scaling using StandardScaler, optional outlier detection, exploratory data analysis, and dataset splitting, with SMOTE included in the system architecture for addressing class imbalance. Multiple models are considered in the Model Training Module, including Random Forest, Support Vector Machine (SVM), Logistic Regression, Gradient Boosting, K-Nearest Neighbors (KNN), and Neural Networks. The major modules include Data Collection, Data Preprocessing, Model Training, Prediction, and Model Evaluation. The trained models process new patient data and generate a prediction indicating whether GDM is likely, along with a confidence score or probability. The system is intended for use in prenatal care, clinics, and hospitals to assist healthcare professionals with early risk identification and timely management.

Complete Technical Specifications