survcomp

survcomp provides functions to assess and compare the performance of risk prediction models for survival analysis, enabling quantitative evaluation of prognostic models using gene expression data.

Key Features:

• Model Comparison: Quantitative assessment and a structured framework for comparing the accuracy of predictive survival models.
• Univariate and Interpretability: Evaluation of simpler univariate models based on single genes, including proliferation markers, and direct comparison to more complex methods.
• High-dimensional Data Handling: Consideration of issues arising from a high number of variables, limited sample sizes, and elevated noise levels in datasets.
• Microarray Gene Expression: Support for analysis of gene expression profiles derived from microarray experiments for prognostic assessment.

Scientific Applications:

• Breast Cancer Prognostication: Assessment and comparison of prognostic models in breast cancer (BC), including prediction of survival outcomes independently of treatment.
• Gene Expression Analysis: Exploration of BC biology and evaluation of prognostic markers using microarray gene expression profiles.

Methodology:

Computational methods explicitly include quantitative assessment and comparison of predictive models using cross-validation, evaluation of univariate (single-gene, e.g., proliferation markers) versus more complex methods, and analysis approaches that address high-dimensional microarray data with limited sample sizes and noise.