Path2Surv: Pathway/gene set-based survival analysis using multiple kernel learning.

Area of Science:

• Bioinformatics
• Computational Biology
• Cancer Genomics

Background:

• Survival analysis is crucial for understanding patient outcomes.
• Integrating molecular pathways can reveal mechanisms of patient survival.
• Current methods often involve separate pathway selection and model training steps.

Purpose of the Study:

• To develop a novel machine learning algorithm, Path2Surv, for integrated pathway and survival analysis.
• To conjointly identify predictive pathways and train a survival model.

Main Methods:

• Developed Path2Surv, a machine learning algorithm using multiple kernel learning.
• Path2Surv integrates pathway/gene set information directly into the survival model training.
• Algorithm tested on gene expression data from 7655 patients across 20 cancer types.

Main Results:

• Path2Surv significantly outperformed survival random forest (RF) on 12 out of 20 cancer datasets.
• Achieved comparable predictive performance to survival support vector machine (SVM).
• Utilized less than 10% of the gene expression features compared to survival RF and SVM.

Conclusions:

• Path2Surv offers an efficient and effective approach for molecular mechanism discovery in cancer survival.
• The algorithm enhances predictive accuracy while reducing feature dimensionality.
• Provides a valuable tool for cancer research and personalized medicine.