Joint clustering multiple longitudinal features: A comparison of methods and software packages with practical...

Area of Science:

• Biostatistics
• Medical Informatics
• Data Science

Background:

• Clustering longitudinal features is crucial for identifying disease developmental trajectories in medical studies.
• Integrating multiple longitudinal features enhances clustering by incorporating more information, potentially revealing co-existing patterns and deeper biological insights.
• Limited practical guidance exists for implementing and evaluating cluster analysis for multiple longitudinal features in medical datasets.

Purpose of the Study:

• To provide an overview of commonly used approaches for clustering multiple longitudinal features.
• To offer practical guidance on implementation using R software.
• To compare the performance of different clustering approaches in medical contexts.

Main Methods:

• Overview of model-based (frequentist and Bayesian) and algorithm-based approaches for clustering multiple longitudinal features.
• Emphasis on application and implementation using R software.
• Comparative performance evaluation using real-life and simulated medical datasets.

Main Results:

• Comparison of various clustering methods for multiple longitudinal features.
• Evaluation of performance across different dataset types (real-life and simulated).
• Identification of practical guidance for applied researchers.

Conclusions:

• The study offers practical guidance for researchers applying clustering methods to multiple longitudinal features.
• Recommendations are provided for applied researchers and future research directions in this field.
• Comparative analysis aids in selecting appropriate methods for medical data analysis.