HYBRID APPROACHES IN MACHINE LEARNING: INTEGRATING ANT COLONY OPTIMIZATION FOR IMPROVED LUNG CANCER DETECTION

Abstract

Early detection of lung cancer is crucial for improving survival rates, making it one of the deadliest diseases worldwide. Machine learning (ML) techniques have been increasingly adopted to improve diagnostic processes. However, individual algorithms face trade-offs between accuracy, precision, recall, and F1 scores. This study aimed to evaluate the performance of four ML algorithms—SVM, Random Forest, KNN, and CNN—for predicting lung cancer. Each model revealed strengths and weaknesses, particularly low recall values, suggesting a need for more comprehensive solutions. The research introduced a stacking ensemble model to enhance prediction performance. SVM showed a high precision of 0.983 but a low recall of 0.5, while Random Forest balanced accuracy (0.967) and recall (0.741). KNN and CNN also performed well, though they struggled with precision. To overcome these limitations, the stacking ensemble model combined SVM, Random Forest, and KNN, achieving an accuracy of 0.977, precision of 0.701, recall of 0.711, and an F1 score of 0.711. This approach maximized the strengths of each model, offering better generalization and robustness. Overall, the ensemble model proved more effective in healthcare applications, where accurate and early diagnosis is essential for successful treatment.