Optimizing lung cancer prediction: leveraging Kernel PCA with dendritic neural models.

Area of Science:

• Medical Informatics
• Computational Biology
• Machine Learning in Healthcare

Background:

• Delayed lung cancer diagnosis contributes to high mortality rates.
• Effective prediction models are crucial for early detection and treatment optimization.
• Traditional machine learning (ML) models require enhancement for improved lung cancer prediction.

Purpose of the Study:

• To enhance a dendritic neural model (DNM) for lung cancer prediction.
• To improve accuracy, precision, and sensitivity using collaborative feature selection and extraction.
• To compare the enhanced DNM against traditional ML models.

Main Methods:

• Utilized a dataset of 1000 lung cancer patients with 23 features.
• Employed collaborative feature selection and extraction techniques, including Principal Component Analysis (PCA), Kernel PCA (K-PCA), and Uniform Manifold Approximation and Projection (UMAP).
• Evaluated model performance using accuracy, precision, F1 score, sensitivity, specificity, and confusion matrix.

Main Results:

• The dendritic neural model (DNM) integrated with Kernel PCA (K-PCA) achieved 98.50% accuracy, 99.42% precision, and 98.84% sensitivity.
• Principal Component Analysis (PCA) resulted in 96.50% accuracy, 96.64% precision, and 97.45% sensitivity.
• Uniform Manifold Approximation and Projection (UMAP) yielded 98% accuracy, 98.82% precision, and 98.82% sensitivity.

Conclusions:

• The Kernel PCA-enhanced dendritic neural model (DNM) demonstrates superior performance for lung cancer prediction.
• The proposed method offers a significant improvement over traditional machine learning models.
• This enhanced DNM holds potential for advancing lung cancer diagnostics and improving patient outcomes in healthcare research.