A Novel Hybrid Neural Support Vector Algorithm for Lung Cancer Progression Prediction Using NN and SVM

Abstract

Lung cancer remains one of the most pressing global health issues, requiring innovative strategies This research article introduces the Hybrid Neural Support Vector Algorithm (HNSVA), a sophisticated predictive model that combines the strengths of Neural Networks and Support Vector Machines. HNSVA utilizes the feature extraction capabilities of Neural Networks and the classification prowess of SVM to address the limitations commonly found in traditional models. The lung cancer dataset, containing key attributes such as genetic markers, tumor properties, and patient demographics, underwent preprocessing that involved normalization and imputation techniques. The model was trained using cross-validation to ensure robust generalization across various data splits. Results demonstrated that HNSVA outperformed standalone Neural Networks and SVM models. Notably, the model significantly reduced false negatives while keeping false positives within an acceptable range. This improvement is credited to the fine-tuning of hyperparameters and the incorporation of an early stopping mechanism to prevent overfitting. Although the integration of both models incurs a slightly higher computational cost, the advantages offered by HNSVA render it suitable for real-time lung cancer detection applications. HNSVA presents a promising and highly accurate method for predicting lung cancer progression, contributing meaningfully to advancements in medical diagnostics.