====== Survival regression ======


Formula of the survival regression is: {{::survival_regression.png?200|}}
Where γ is the scaling parameter. And //Xβ// is the model linear predictor, estimated by the summation of all the β values multiplied by that variable. E.g. //β<sub>var1</sub> ⋅ var1 + β<sub>var2</sub> ⋅ var2 + β<sub>var3</sub> ⋅ var3..........    β<sub>varN</sub> ⋅ varN//\\




===== Practical example =====
To demonstrate how to add a survival regression model on Evidencio, the addition of the [[https://www.mskcc.org/|Memorial Sloan Kettering Cancer Center (MSKCC)]] will be shown step by step below.\\
First of all, we need all the model coefficients. They can be found in the table below:\\
{{::survival_regression1.png?100|}}\\

  * Go to the //"Model"// tab in the model editor (more information on model editting [[creating_a_model|here]]). Select **Survival regression, Beta coefficients, Single value** and enter the **Intercept**(6.14381457) and **C-statistic**(0.82)
  * Now add the variables found in the list above. Here is an {{::survival_regression2.png?linkonly|example}} on how to add the variables to Evidencio. Note that there were two splines for the preoperative PSA levels. These splines were addes using a full range variable conversion. The equation to calculate spine variables were also found on the website of the MSKCC. 
  * Next go to the sub-tab "//Survival options//" Where you can specify time parameters. Selecting **No** will show no slider on the model page, selecting **Point in time** will only show the point in time slider on the model page, and selecting **All** will show both the point in time and years no event slider on the model page. 
  * See {{ ::survival_regression3.png?linkonly |this}} example on how the survival options can be filled in. The point in time on the left is set at 10 years with 0 years without an event, so when using the model, the calculated survival concerns the 10 year probability of survival directly after surgery (baseline is 0 years). With the "point in time" slider, the user is able to adjust the 10 year probability to a survival probability between 1 and 15 years. In this case, no slider for "years no event" was set, so this will always be 0. This 0 is reflected as t<sup>*</sup> in the formula. The scaling parameter (1.0814495) is reflected in the formula as γ.
  * Our model is now complete and ready for use. Make sure to set the **Pre-result text** at the **result** tab in the model editor, otherwise the users of the model don't know what they're predicting. If everything was filled in correctly the result might look like {{ ::survival_regression4.png?linkonly |this}}