Evaluation of CatBoost for Diabetes Prevention in Comparison to XGBoost: To avert mortality by developing an AI model capable of predicting the onset of diabetes

Diabetes is a metabolic condition defined by consistently elevated blood glucose levels. Hyperglycemia is characterized by heightened hunger, thirst, and urination. Numerous consequences may arise from untreated diabetes. Diabetic ketoacidosis, hyperosmolarity, hyperglycemia, and even mortality may occur as acute consequences. Heart disease, stroke, chronic renal failure, foot ulcers, and eye impairment are some of the worst long-term consequences. A total of 463 million individuals were found to have diabetes in 2019, according to the worldwide diabetes organisation. Assuming predictions come true, that number will soar to 700 million by 2045, up 578 million by 2030. It has modern medical applications, allowing for the rapid and precise diagnosis of illness. Thus, we may use machine learning to avert mortality by developing an AI model capable of predicting the onset of diabetes; to determine which algorithm is most suited for this task, we will compare the CatBoost and XGBoost methods. In the end, the research compared two diabetes prediction models, XGBoost and CatBoost, using a variety of health indicators from the dataset. Using a real-world dataset obtained from Kaggle, our suggested system is trained and built using Python. Performance assessment criteria such as precision, recall, F1score and accuracy measurements are used to thoroughly analyse our methods. In comparison to XGBoost's 88.33% accuracy rate, 85.25 percent precision, 85.63 percent recall, and an F1 Score of 86.50. CatBoost's 91.08% percent accuracy, 85.25 percent precision, 85.63 percent recall, and 87.38 F1 Score. CatBoost marks it as the most successful ensemble approach.