Machine learning has emerged as a powerful tool in healthcare, enabling prediction of medical conditions based on patient profiles. However, imbalanced datasets and limited representation of minority classes often hinder model performance in medical predictions. This study addresses these challenges by applying data augmentation techniques, specifically the Synthetic Minority Over-sampling Technique (SMOTE), to generate synthetic samples for underrepresented classes, achieving a more balanced dataset. The methodology includes a RandomForest classifier, enhanced with feature engineering and hyperparameter tuning, to improve the accuracy and robustness of predictions. Additionally, K-means clustering was employed to identify patterns within patient health profiles, providing insights into characteristic groupings, while feature importance analysis highlights the most influential indicators, such as BMI and glucose levels. The optimized model achieved an accuracy of 87%, with balanced precision, recall, and F1-scores across all classes. These findings underscore the importance of data augmentation and feature selection in improving model performance for medical condition prediction. The study’s implications suggest that targeted feature engineering and clustering can enhance interpretability, aiding clinicians in focusing on critical health indicators for early diagnosis and intervention.