A Hybrid Ensemble Model for Accurate Lung Cancer Detection Using Harris Hawks Optimization and Advanced Data Preprocessing

Medical diagnostics faces an essential challenge with detecting lung cancer because diagnostic models need precise accuracy and dependable outcomes together with clear interpretation methods. The proposed method introduces an enhanced predictive framework for lung cancer that includes improved data cleaning steps combined with ensemble classification techniques and feature selection procedures to boost performance measures. The data preprocessing starts with three sequential stages which encompass MICE to handle missing values alongside ADASYN for class imbalance correction combined with IQR filtering to find out and discard outliers. By implementing the HHO algorithm the model becomes more efficient and interpreter-friendly through the process of selecting important clinical and biological variables. The Extra Random Trees Classifier Hybrid (ERTC) represents the core classification model because it unites Random Forest (RF) and Extra Trees Classifier (ETC) strengths for increased model diversity and robustness. The model receives thorough testing through measures of accuracy and precision and its ability to recall patients and F1-score along with Area Under the Curve (AUC). The ERTC model delivers outstanding results through its 98.87 % accuracy and 98.92 % precision and 98.85 % recall which lead to an outstanding F1-score of 98.88% and an AUC of 99.14%. The model demonstrates readiness for direct implementation in real-time clinical decision support systems. The conducted research demonstrates that the proposed model demonstrates exceptional potential for lung cancer early detection in clinical settings as an effective analytical tool.