CKD Detection in Patients with Reproductive Health Challenges Utilizing Ensemble Models in Machine Learning

Infertility and Chronic Kidney Disease (CKD) have garnered global recognition as significant health challenges, and their interconnected nature has been acknowledged over time. Infertility has affected numerous couples worldwide, while CKD is notorious for its harmful effects on kidney function and overall health. The relationship between these conditions is strengthened by shared factors such as chronic illnesses, lifestyle choices, and CKD’s impact on reproductive health. Our research relied on extensive clinical data collected from hospital databases, which provided a wealth of information encompassing both CKD and infertility. The primary goal of this study was to identify CKD among individuals who were already dealing with infertility and to conduct factor-wise analysis. To achieve this objective, we employed a diverse set of machine learning models, including Random Forest, AdaBoost, Gradient Boosting, XGBoost, LightGBM, and CatBoost, for diagnosing CKD. Essential data preparation steps, such as encoding and scaling features, played a pivotal role in the research process. After thorough training and evaluation, our findings indicated that CatBoost emerged as the most effective model, demonstrating superior accuracy and ROC curve performance. These results suggested that CatBoost had the potential to facilitate early CKD detection among individuals facing infertility-related challenges.