The graphical isomorphism network algorithm for pcod for disease prediction and risk stratification

The Graphical Isomorphism Network (GIN) algorithm is employed for disease prediction and risk stratification of Polycystic Ovary Disease (PCOD/PCOS), which is an intricate condition impacting numerous women worldwide. The research leverages the power of graph-based algorithms, specifically GIN, to better understand complex relationships between various health parameters and predict the likelihood of PCOD. The dataset consists of vital health features such as age, height, weight, menstrual cycle data, and more, which are analyzed using state-of-the-art machine learning techniques.

In the project, multiple machine learning models are evaluated, including Random Forest (RF), XGBoost, and Convolutional Neural Networks (CNN), to ascertain their efficacy in PCOD diagnosis and stratification. Among them, Random Forest has shown remarkable performance, achieving an accuracy of 93.43%, making it the most reliable model for the task at hand. XGBoost also performs well with an accuracy of 93.50%. Meanwhile, CNN showed promising results with an accuracy of 84.90%.

The overall framework integrates Flask, a lightweight Python web framework, with a MySQL database for managing user registrations and model predictions, offering a seamless user experience. The system allows users to upload datasets, select different models, and receive health predictions with actionable recommendations based on the output. This approach enables personalized health management, ensuring women are more informed about their risk factors for PCOD and can make better decisions regarding their health.

Keywords:

Graphical Isomorphism Network (GIN), PCOD, Disease Prediction, Risk Stratification, Random Forest (RF), Health State Prediction, Machine Learning, Feature Engineering, Medical Data Analysis, Personalized Health Recommendations, Flask Web Application, Model Performance, Health Technology