The worldwide health challenge from smoking persists at a severe level since it creates major persistent diseases that extend to cardiovascular problems and both respiratory problems and cancer development. Standard methods of smoking behaviour analysis face challenges in detecting complex variable interactions that derive from demographic aspects alongside health-related and behavioral components. This research project depends on expert machine learning (ML) algorithms which include Logistic Regression along with Random Forest and XGBoost to extract practical data about smoking behaviors. The researchers analyzed the structured database which exceeded 3000 observations to establish smoking duration as the primary variable that causes disease development together with age and cigarette usage. The research produced a network to explain risk factors with smoking interval and smoking status at the core while creating a 3D disease risk model based on age and smoking duration and an anomaly detection method to detect abnormal smoking habits. The best model performance resulted from XGBoost because the algorithm predicted with accuracy and provided clear feature rankings that Random Forest also achieved. Research indicates that ML technology can reveal hidden patterns that help public health organizations create their strategic initiatives. The developed research insights will lead to effective strategies for helping high-risk smoking populations including aging smokers who use excessive cigarettes.

Feature Importance Disease Risk Prediction

Public Health Interventions

smoking behaviour

Smoking Cessation Programs

machine learning