LogSum + L₂ penalized logistic regression model for biomarker selection and cancer classification

Area of Science:

• Genomics
• Bioinformatics
• Machine Learning

Background:

• Genomic data analysis is crucial for cancer classification and biomarker discovery.
• Accurate diagnosis and improved machine learning model performance rely on identifying key gene biomarkers and biological pathways.

Purpose of the Study:

• To propose a novel penalized logistic regression model for biomarker selection and cancer classification.
• To evaluate the model's effectiveness using simulations and real-world experimental data.

Main Methods:

• Development of a LogSum + L2 penalized logistic regression model.
• Application of a coordinate descent algorithm for model optimization.
• Comparative analysis against state-of-the-art methods.

Main Results:

• The proposed model shows high competitiveness compared to existing methods.
• Excellent performance achieved in group feature selection tasks.
• Effective classification of different cancer types demonstrated.

Conclusions:

• The LogSum + L2 penalized logistic regression model is a powerful tool for genomic data analysis in cancer research.
• The method offers significant advantages in both biomarker identification and cancer classification.
• This approach enhances knowledge discovery from genomic datasets.