Lung cancer prediction from microarray data by gene expression programming.

Area of Science:

• Oncology
• Bioinformatics
• Computational Biology

Background:

• Lung cancer remains a primary cause of cancer mortality globally.
• Early cancer diagnosis significantly improves treatment outcomes.
• Microarray technology offers potential for cancer diagnosis via gene expression profiling.

Purpose of the Study:

• To develop and evaluate a gene expression programming (GEP) model for predicting lung cancer from microarray data.
• To compare the performance of GEP models against established machine learning methods.

Main Methods:

• Utilized two gene selection techniques to identify significant lung cancer-related genes.
• Developed distinct GEP-based prediction models based on selected genes.
• Performed rigorous performance evaluations and cross-dataset validation on real-world lung cancer microarray datasets.
• Compared GEP models against support vector machine, multi-layer perceptron, and radial basis function neural networks.

Main Results:

• The GEP model employing fewer feature genes achieved superior performance across key metrics: accuracy, sensitivity, specificity, and area under the ROC curve.
• Cross-dataset validation confirmed the reliability and robustness of the GEP approach.
• GEP models outperformed traditional machine learning methods in lung cancer prediction.

Conclusions:

• Gene expression programming (GEP) presents a highly effective solution for lung cancer prediction using microarray data.
• The GEP model's efficiency is enhanced by utilizing a reduced set of informative feature genes.
• This study highlights GEP as a promising tool for improving early lung cancer diagnosis.