Statistical Models for Prediction

Abstract

In this chapter, we consider statistical models for different types of outcomes: binary, unordered categorical, ordered categorical, continuous, and survival data. We discuss common statistical models in medical research such as the linear, logistic, and Cox regression model. We consider simpler approaches and more flexible extensions, including regression trees and neural networks. We also discuss competing risks and the concept of dynamic prediction modeling. We focus on the most relevant aspects of these models in a prediction context. All models are illustrated with case studies. In Chap. 6, we will discuss aspects of choosing between alternative statistical models for the same type of outcome.