A Risk Stratification Model for Lung Cancer Based on Gene Coexpression Network and Deep Learning

Area of Science:

• Genomics
• Bioinformatics
• Cancer Research

Background:

• Lung cancer risk stratification using gene expression profiles is crucial.
• Previous models focused on individual genes, limiting system-level insights.
• A novel approach integrating gene coexpression networks is needed.

Purpose of the Study:

• To develop a system-level risk stratification model for lung adenocarcinoma.
• To utilize gene coexpression networks and deep learning for improved prognostic prediction.
• To validate the model's ability to predict patient survival independently of clinicopathological variables.

Main Methods:

• Gene coexpression network analysis was performed on microarray data to identify survival-related networks.
• A deep learning model was constructed using representative genes from these networks.
• The model was validated on independent training and test sets using survival analysis.

Main Results:

• Five gene coexpression networks were significantly associated with patient survival.
• The deep learning model demonstrated significant association with survival in all sets (p < 0.00001).
• Multivariate analyses confirmed the model's independence from clinicopathological features.

Conclusions:

• Gene coexpression networks offer a novel perspective for lung cancer risk stratification.
• Deep learning applied to genomic data provides a powerful tool for prognosis prediction.
• This integrated approach enhances the clinical application of genomic data science in oncology.