In many biomedical applications, outcome is measured as a “time-to-event” (e.g., disease progression or death). To assess the connection between features of a patient and this outcome, it is common to assume a proportional hazards model and fit a proportional hazards regression (or Cox regression). To fit this model, a log-concave objective function known as the “partial likelihood” is maximized. For moderate-sized data sets, an efficient Newton–Raphson algorithm that leverages the structure of the objective function can be employed. However, in large data sets this approach has two issues: (i) The computational tricks that leverage structure can also lead to computational instability; (ii) The objective function does not naturally decouple: Thus, if the data set does not fit in memory, the model can be computationally expensive to fit. This additionally means that the objective is not directly amenable to stochastic gradient-based optimization methods. To overcome these issues, we propose a simple, new framing of proportional hazards regression: This results in an objective function that is amenable to stochastic gradient descent. We show that this simple modification allows us to efficiently fit survival models with very large data sets. This also facilitates training complex, for example, neural-network-based, models with survival data.

This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (