autoRPA

autoRPA constructs cancer staging models using recursive partitioning analysis (RPA) to derive prognostic groups from survival data.

Key Features:

• Recursive Partitioning Analysis (RPA): Builds decision trees from survival data using RPA to define staging groups.
• Model Pruning: Provides pruning of decision trees to refine prognostic model structure.
• Covariate Evaluation: Evaluates the contribution of each covariate involved in the grouping process.
• Performance Validation Indicators: Computes hazard consistency, hazard discrimination, percentage of variation explained, and sample size balance to validate staging models.
• Model Comparison: Compares prognostic staging models using a standard bootstrap evaluation method.

Scientific Applications:

• Cancer staging model construction and validation: Develops and validates staging systems from clinical survival data.
• Prognostic stratification: Identifies patient groups with distinct survival outcomes for prognostic assessment.
• Covariate impact assessment: Assesses the influence of clinical covariates on stage grouping and prognosis.
• Model selection and benchmarking: Enables comparison and selection of staging models using bootstrap-based evaluation and quantitative performance indicators.

Methodology:

Implements recursive partitioning analysis (RPA) to build decision trees from survival data, supports tree pruning, evaluates covariate contributions, computes hazard consistency, hazard discrimination, percentage of variation explained and sample size balance, and performs model comparison via bootstrap evaluation.