临床预测模型常用统计模型及其SAS实现

Abstract

摘要： 临床预测模型在医学研究中的应用越来越广泛，欲达到良好的预测性能，选择正确的模型非常关键。对于预测模型类型的选择，预测结局的类型起着决定性作用。本文从数据类型的角度出发，将结局变量分为连续变量（正态分布、偏态分布）、分类变量（二分类、无序多分类、有序多分类）以及时间-事件变量（无竞争风险、有竞争风险），分别介绍不同类型结局变量的特点、对应的模型、建模案例以及SAS实现程序，以期为研究者构建预测模型提供参考。 Abstract: The application of clinical prediction models in medical research is becoming increasingly widespread. To achieve good predictive performance, selecting the correct model is crucial. Regarding the choice of prediction model, the type of prediction outcome plays a decisive role. This paper, from the perspective of data types, divided outcomes into continuous variables (normal distribution, skewed distribution), categorical variables (binary, nominal, ordinal), and time-to-event variables (with or without competing risks), introducing the characteristics of different types of outcomes, the model types, examples and SAS programs to serve as a reference for researchers to develop prediction models.

Keywords

• 临床预测模型
• 统计模型
• 逻辑回归
• clinical prediction model
• statistical model
• logistic regression