Article type
Year
Abstract
Objectives:
Participants will be introduced to individual participant data (IPD) meta-analysis in risk prediction research. We will discuss the opportunities and challenges of combining IPD from multiple studies when developing and/or validating a risk prediction model. Appropriate statistical methodology and software packages will be introduced and IPD meta-analysis using case studies and example papers will be illustrated.
Description:
Risk prediction models are developed and validated to predict future occurrence of a particular outcome (prognostic) but also to predict the presence of a certain disease (diagnostic). Prediction models aim to provide absolute probabilities of a certain outcome or disease in an individual. Risk prediction models offer opportunities in clinical practice and public health, although their anticipated performance is often overoptimistic and their generalizability tends to be insufficient. Testing a model’s predictive abilities in other datasets is thus essential. IPD meta-analysis is increasingly used to address issues in risk prediction modelling, however little guidance exists. We will discuss why IPD meta-analysis offers unique opportunities to risk prediction research, describe how to combine multiple datasets appropriately and demonstrate how to accommodate for heterogeneity across populations. We will show how to interpret the achieved meta-analytical model performance and provide strategies for improving generalizability.
Participants will be introduced to individual participant data (IPD) meta-analysis in risk prediction research. We will discuss the opportunities and challenges of combining IPD from multiple studies when developing and/or validating a risk prediction model. Appropriate statistical methodology and software packages will be introduced and IPD meta-analysis using case studies and example papers will be illustrated.
Description:
Risk prediction models are developed and validated to predict future occurrence of a particular outcome (prognostic) but also to predict the presence of a certain disease (diagnostic). Prediction models aim to provide absolute probabilities of a certain outcome or disease in an individual. Risk prediction models offer opportunities in clinical practice and public health, although their anticipated performance is often overoptimistic and their generalizability tends to be insufficient. Testing a model’s predictive abilities in other datasets is thus essential. IPD meta-analysis is increasingly used to address issues in risk prediction modelling, however little guidance exists. We will discuss why IPD meta-analysis offers unique opportunities to risk prediction research, describe how to combine multiple datasets appropriately and demonstrate how to accommodate for heterogeneity across populations. We will show how to interpret the achieved meta-analytical model performance and provide strategies for improving generalizability.