Update docs/examples/example_survival.ipynb

Co-authored-by: AronPapQC <[email protected]>

docs/examples/example_survival.ipynb

@@ -65,7 +65,7 @@
     "\n",
     "To address these issues, we need to construct a model to act as the nuisance model. Here's how we can address the associated problems:\n",
     "\n",
-    "* **Time to event prediction:** The principle to remember is that this nuisance model, when invoking the ``predict`` method, must return a time to event prediction for each observation. In this case, we will employ [``xgboost`` with Accelerated Failure Time](https://xgboost.readthedocs.io/en/stable/tutorials/aft_survival_analysis.html), since its output is inherently the time to event. However, other models may necessitate integration over the survival function or necessitate other transformations.\n",
+    "* **Time to event prediction:** The principle to remember is that this nuisance model, when invoking the ``predict`` method, must return a time to event prediction for each observation. In this case, we will employ [``xgboost`` with Accelerated Failure Time](https://xgboost.readthedocs.io/en/stable/tutorials/aft_survival_analysis.html), since its output is inherently the time to event. However, other models may necessitate integration over the survival function or require other transformations to get expected survival time predictions.\n",
     "\n",
     "* **Censoring Support:** For handling this issue, we keep ``y`` as is. If the event was observed, the value represents when it occurred, and if not observed, it is the censoring time. We then augment the ``X`` dataset with an extra column (which won't be passed to the model), labeled ``\"censored\"``, indicating whether the observation was censored or not. This column aids in constructing the upper and lower bounds for the output label from the model.\n",
     "\n",