statistics / survival analysis · BEGINNER LESSON
Cox proportional-hazards model
The Cox proportional-hazards model is a semiparametric survival-analysis method for time-to-event data. It estimates how covariates change the hazard, meaning the instantaneous risk that an event happens at a given time, while leaving the baseline hazard unspecified.
What it is and what it measures
The Cox proportional-hazards model is used in survival analysis, which studies the time until an event such as death, recurrence, recovery, or loss to follow-up. It measures the hazard, the instantaneous risk that the event occurs at a particular time given the person’s covariates, or measured predictors. The model separates this into a baseline hazard and multiplicative effects of covariates, so the baseline risk does not need to be written as a fixed formula.
[1] Machine learning prediction of tuberculosis mortality - PMC [3] Applied comparison of stepwise Cox and Cox-LASSO regression in ...
How it works
In the model, each covariate has a coefficient that is interpreted through the log hazard ratio. A hazard ratio compares the hazard for one person or group with another. If the model is proportional hazards, that ratio is assumed to stay constant over time. This makes the method useful for comparing several factors at once while keeping the results interpretable. Right censoring is allowed, which means some people do not have an observed event time before the study ends or they leave follow-up.
[1] Machine learning prediction of tuberculosis mortality - PMC [3] Applied comparison of stepwise Cox and Cox-LASSO regression in ...
Common uses and related ideas
The model is widely used for medical and other time-to-event data because it can handle censoring and produce hazard ratios for multiple predictors. It is a standard reference method in survival analysis and is often compared with more flexible approaches such as random survival forest or extensions like penalized-spline Cox models. Those alternatives may fit non-linear or time-varying patterns better, but the Cox model is often valued for simplicity and interpretability.
[1] Machine learning prediction of tuberculosis mortality - PMC [2] Comparison of Methods for Analyzing Environmental Mixtures ... [3] Applied comparison of stepwise Cox and Cox-LASSO regression in ...
Limitations
The main limitation is the proportional-hazards assumption: the hazard ratio between two sets of covariates should remain constant over time. If hazards cross or effects change with time, the model may fit poorly or give misleading summaries. It can also be unstable with many correlated predictors, and because it is usually linear on the log-hazard scale, it may miss complex non-linear relationships unless extended.
[1] Machine learning prediction of tuberculosis mortality - PMC [2] Comparison of Methods for Analyzing Environmental Mixtures ... [3] Applied comparison of stepwise Cox and Cox-LASSO regression in ...
SEE THE IDEA
How the Cox proportional-hazards model organizes survival data
Measure how long until an event happens or the observation is censored.
Starting hazard pattern when covariates are set to the reference value.
Predictors such as age or weight adjust the hazard through estimated effects.
Hazard ratios are assumed to stay constant across time.
Outcome time is modeled through a baseline hazard that is multiplied by covariate effects, under a constant hazard-ratio assumption.
The model predicts relative hazard over time rather than directly specifying the full baseline risk curve.
TRY IT YOURSELF
Test your understanding
Choose an answer, then check the explanation.
PUT IT TOGETHER
Match each concept to its meaning
Drag a label onto its description, or select a label and then select a description. This also works with a keyboard.
4 pairs to match
Sources and further reading
This general lesson was generated from public source material. It explains a concept; it does not interpret a private paper or provide medical advice.