survival-analysis-tools

Kaplan-Meier survival analysis summarizes time-to-event data by estimating survival probabilities across time points while properly handling censored observations. This skill generates survival probability tables and plots directly from datasets using pandas and matplotlib.

Cox proportional hazards models estimate hazard ratios by fitting regression models to identify factors influencing survival time. This skill fits these models using statsmodels, allowing you to account for covariates like age and disease stage in research datasets.

Comparing survival distributions between different groups is typically done using log-rank tests. This skill performs log-rank tests to determine if survival curves differ significantly between groups, producing reproducible outputs for clinical trial analysis.

You do not need scikit-survival for standard workflows; numpy, pandas, statsmodels, and matplotlib provide the core functionality for Kaplan-Meier summaries, log-rank tests, and Cox models. The sksurv package is only an optional dependency for advanced machine-learning models.

Survival analysis handles censored data by accounting for subjects whose event has not occurred or is lost to follow-up before the study ends. This skill processes censored time-to-event observations to generate accurate survival probability estimates without biasing the results.