[Statistical analysis methods for identifying multimorbidity patterns].

Abstract

Multimorbidity has become a widely recognized public health problem worldwide. Identifying multimorbidity patterns can improve not only the efficiency of healthcare resource utilization but also patients' prognosis. This article summarizes three common approaches for the identification of multimorbidity patterns: association analysis methods (including association rule mining and network analysis), classification methods (including cluster analysis, latent class analysis, and latent transition analysis), and dimensionality reduction and feature extraction methods (including principal component analysis, factor analysis, and multiple correspondence analysis), introduces the application of these methods using data from the UK Biobank to identify multimorbidity patterns and discusses and compares the results of case analysis to provide reference for the selection of appropriate methods for multimorbidity pattern research.