Sickle cell disease (SCD) poses significant diagnostic challenges, necessitating precise and reliable classification methods for effective patient management. This study employs various machine learning models, including CatBoost, K-Nearest Neighbors (KNN), Extra Trees, Random Forest, Gaussian Naive Bayes, Logistic Regression, Decision Tree, XGBoost, SVM, MLP, and AdaBoost and Gradient Boosting, to enhance diagnostic accuracy for SCD. The comprehensive dataset, extracted from the western part of Odisha and mapped with Alpha-2_20032, was crucial for training and validation. We optimized model performance using Genetic Algorithm (GA) feature selection and optuna driven hyperparameter tuning. CatBoost achieved the highest accuracy of 98.97%, benefiting from its effective handling of categorical variables and robust boosting framework. KNN and Extra Trees also performed well, each with an accuracy of 97.95%, due to optimized decision boundaries and feature importance assessments. In contrast, Gaussian Naive Bayes had a lower accuracy of 77.55%, likely due to its assumption of feature independence, which may not suit the interdependent dataset variables. Hyperparameter tuning, focusing on parameters such as ‘max_depth’, ‘n_estimators’, ‘C’, and ‘gamma’, was critical for enhancing predictive performance across all models. This study underscores the importance of model selection and hyperparameter optimization in improving SCD diagnostic accuracy, with significant implications for patient care. Future research will incorporate additional clinical features and explore deep learning techniques to further enhance diagnostic capabilities.