CondiS

CondiS imputes censored survival times by sampling from their conditional distribution to enable the use of complete datasets in machine-learning-based survival analysis.

Key Features:

• Censoring Imputation: Imputes censored survival times by drawing from their conditional distribution given the observed portion of each subject's data.
• Incorporation of Covariates (CondiS-X): Extends imputation by conditioning on additional covariate information to refine imputed survival times.
• Prediction for New Patients: Produces predicted survival times for newly provided patient records using the same conditional imputation framework.

Scientific Applications:

• Survival Analysis with Censoring: Enables analysis of datasets containing right-censored survival times by providing complete-event time estimates.
• Machine-Learning Model Training: Facilitates training and evaluation of machine-learning prognostic models on datasets after imputation of censored values.
• Personalized Prognosis: Supports individualized survival time prediction by incorporating patient-specific covariates into the imputation process.

Methodology:

Imputation of censored survival times by sampling from their conditional distribution given observed data, with optional conditioning on covariates (CondiS-X) to generate predictions for new patients.