What is the Cox Proportional Hazards Model?
You'll learn what the Cox model is and why it's the most popular regression method in survival analysis.
The Core Idea
The Cox Proportional Hazards Model (or Cox regression) is a semi-parametric method that lets you predict how covariates (like age, treatment, or risk factors) affect survival time without assuming what the baseline hazard distribution looks like.
Think of it this way: imagine you're studying how different medications affect patient recovery time. Traditional parametric models would require you to assume the "natural" recovery pattern follows a specific shape (exponential, Weibull, etc.). The Cox model says: "I don't care what that baseline pattern is—I just want to know how each medication multiplies the risk compared to that baseline."
Why "Semi-Parametric"?
- Parametric part: The model assumes covariates affect hazard through a mathematical formula with parameters (coefficients) you estimate.
- Non-parametric part: The baseline hazard function (risk over time for someone with all covariates = 0) is not assumed to follow any distribution—it's left flexible.
This flexibility makes Cox regression extremely practical because you don't need to guess the underlying hazard shape.
The "Proportional Hazards" Assumption
The model assumes that the effect of covariates is multiplicative and constant over time. If treatment A doubles your hazard compared to treatment B at month 1, it doubles it at month 6 too—the proportion stays constant.
Key Takeaway: The Cox Proportional Hazards Model is a flexible, semi-parametric regression tool that relates covariates to survival time without requiring you to assume a specific baseline hazard distribution.