Among 299 heart failure patients (96 deaths), the strongest independent mortality predictor in the Cox model is Hypertension. Median survival could not be determined from the observed follow-up. The Cox model achieves a C-statistic of 0.741, indicating good discriminative ability. The proportional hazards assumption holds globally (Schoenfeld test p > 0.05).

The overall Kaplan-Meier curve shows the probability of surviving beyond each follow-up time point for the full cohort of 299 patients. Survival at 30, 90, and 180 days is 88.2%, 76.3%, and 65.4% respectively, with a median survival of not reached. The shaded band shows the 95% confidence interval; wider bands reflect fewer patients at risk at later time points.

Survival curves are stratified by patient sex to compare mortality trajectories between male and female heart failure patients. The sex difference in survival is not statistically significant (log-rank p = 0.9498). Each step in the curve represents one or more observed deaths; flat segments between steps indicate periods with only censored observations.

Survival curves for four binary clinical risk factors (anaemia, diabetes, hypertension, and smoking) compare patients with and without each condition. A wide vertical separation between 'Present' and 'Absent' curves indicates a strong univariate association with mortality. The largest average survival gap between present/absent groups is observed for Hypertension. Log-rank significance for each factor is shown in the adjacent bar chart.

Each bar shows -log10(p) from the log-rank test for that binary covariate; bars beyond the dashed line at 1.3 indicate p < 0.05. Hypertension showed a statistically significant survival difference (log-rank p < 0.05). The factor with the strongest univariate signal is Hypertension. Note that log-rank significance does not imply independent effect — use the Cox model for multivariable inference.

Forest plot of hazard ratios from the multivariable Cox model with 11 covariates; bars show 95% confidence intervals. An HR > 1 means increased hazard (shorter survival); HR < 1 means protective. 5 of 11 covariates have CIs that exclude HR = 1.0. The strongest predictor is Hypertension with HR = 1.609 (95% CI: 1.053–2.458).

Complete Cox regression output for all 11 covariates. 6 predictors are significant at the 0.05 level. The overall model is statistically significant (likelihood ratio test p = 0). C-statistic = 0.741, confirming the model's discriminative capacity.

Scaled Schoenfeld residuals are plotted against event time for each of the 11 covariates in the Cox model. A random scatter around zero (horizontal) indicates the hazard ratio is constant over follow-up, consistent with the proportional hazards assumption. A clear trend (rising or falling) signals a time-varying hazard ratio. Global Schoenfeld test p > 0.05 — the proportional hazards assumption is not rejected.

Time-dependent AUC estimates the Cox model's ability to separate patients who die by each landmark from those who survive beyond it. An AUC of 0.5 means no discrimination; AUC of 1.0 is perfect. The model achieves good discrimination (AUC ≥ 0.70) across the 3 evaluated time points, with best AUC of 0.77 at 180 days. Average AUC across landmarks is 0.757.