Utilizing a variety of machine learning techniques, including Random Forest, Decision Tree, XGBoost, K-nearest neighbors, and Logistic Regression, this research investigates the issue of Smoking Status Prediction. Using critical metrics including F1-score, precision, accuracy, and recall, the research carefully assesses each model's performance. A 38,984 instance, 23 feature dataset obtained from Kaggle is used for predictive modeling. With 81% accuracy, Random Forest stands out as an outstanding performer among them, showcasing its great predictive capacity. Beyond quantitative metrics, the study employs a customized Confusion Matrix for Random Forest to provide additional in-depth insights. The findings not only advance the science of predictive health analytics but also have real-world implications for customized healthcare interventions. The study's findings underscore how critical it is to apply machine learning to challenging public health issues. This study offers the foundation for further research by emphasizing the potential for interdisciplinary collaborations, advanced computational techniques, and the incorporation of diverse dataset to further develop predictive models