Evaluation of Machine Learning Algorithm Utilization for Lung Cancer Classification Based on Gene Expression Levels.

Area of Science:

• Bioinformatics
• Computational Biology
• Oncology

Background:

• Lung cancer is a leading cause of cancer deaths globally, with high case fatality.
• Accurate classification of lung cancer type and malignancy is crucial for effective treatment strategies.
• Tumor morphology significantly influences anticancer treatment efficacy.

Purpose of the Study:

• To evaluate the effectiveness of various machine learning algorithms for lung cancer classification.
• To assess algorithm performance on gene expression data across multiple public datasets.
• To determine the optimal machine learning approach for lung cancer morphology determination.

Main Methods:

• Utilized four publicly available lung cancer gene expression datasets (Dana-Farber, Michigan, Toronto, Brigham and Women's).
• Applied machine learning algorithms including k-nearest neighbor, naive Bayes, support vector machine, and C4.5 decision tree.
• Evaluated algorithm performance using the Matthews correlation coefficient.

Main Results:

• Support vector machines demonstrated superior performance on Dana-Farber and Brigham and Women's Hospital datasets.
• Most algorithms achieved high effectiveness on the University of Michigan dataset, excluding the C4.5 decision tree.
• The C4.5 decision tree yielded the best results for the University of Toronto dataset.

Conclusions:

• Machine learning algorithms are viable tools for lung cancer morphology classification.
• Gene expression level evaluation combined with machine learning aids in diagnostic tasks.
• These computational approaches can support improved lung cancer diagnosis and treatment planning.