clusterMLD: An Efficient Clustering Method for Multivariate Longitudinal Data
- 2024-06-12 (Wed.), 16:40 PM
- Auditorium, B1F, Institute of Statistical Science;The tea reception will be held at 16:30.
- Lecture in English. Online live streaming through Cisco Webex will be available.
- Prof. Ying Zhang
- University of Nebraska Medical Center College of Public Health
Abstract
Longitudinal data clustering is a challenging task, especially with sparse and irregular observations. It lacks reliable methods in the literature that deal with clustering complicated longitudinal data, particularly with multiple longitudinal outcomes. In this manuscript, a new agglomerative hierarchical clustering method is developed in conjunction with B-spline curve fitting and the construction of a unique dissimilarity measure for differentiating longitudinal observations. In an extensive simulation study, the proposed method demonstrates its superior performance in clustering accuracy and numerical efficiency compared to the existing methods. Moreover, the method can be easily extended to multiple-outcome longitudinal data without too much cost in computation and shows its robust results against the complexity of the underlying mixture of longitudinal data. Finally, the method is applied to a date set from the SPRINT Study for validating the intervention efficacy in a Systolic Blood Pressure Intervention Trial and to a 12-year multi-site observational study (PREDICT-HD) for identifying the disease progression patterns of Huntington’s disease (HD).
Please click here for participating the talk online.
Download
Update:2024-06-06 15:20