Skip to Content
Interactive Textbook on Clinical Symptom Research Logo


Home Button

Statistical Models for Prognostication
Author Bio
Introduction
Predictions: Statistical Models
Insight: Statistical Models
Ingredients: Statistical Models
Theoretical Aspects
Central Concepts
Currently selected section: Regression Models
Problems: Regression
Practical Advice
Example 1
Example 2
Chapter 8: Statistical Models for Prognostication: Development of Regression Models
        
In addition to calibration and discrimination, overall performance measures have been proposed, which may be interpreted as the accuracy of predictions for individual patients. In addition to R2 -like measures, these include:
  • The model chi-square (difference in -2 log likelihood of the model and a null model that includes only an intercept), which can be scaled by division through the number of patients (known as performance measure D).
  • Brier score: Symbol(yi - pi)2 / n, where y denotes the observed outcome and p the prediction for subject i in the data set of n subjects.

Note that the (scaled) model chi-square is a logarithmic scoring rule, and the Brier score is quadratic.

QUESTION 7.6

Calibration of model predictions can be assessed with:

Return to Current Section