Investigating network structures in recurrent event data with discrete observation times

The article investigates the application of stochastic block models to recurrent event data with discrete observation times, a common scenario in longitudinal studies. It begins by discussing the prevalence of network data across various fields, such as social sciences, biology, and computer science, and the importance of understanding the structures and interactions within these networks. The text then delves into the stochastic block model, a popular statistical model for investigating network structures, and its extensions to handle more complex networks. A key focus is on recurrent event processes with discrete observation times, which are often encountered in real-world scenarios due to cost considerations. The article presents a variational EM algorithm for parameter estimation, which offers faster computing speed compared to traditional methods. It also introduces a perturbation loop to escape local optima and achieve global convergence. The effectiveness of the proposed method is demonstrated through simulations and an analysis of the French schoolchildren dataset, showcasing its ability to uncover the underlying structure in longitudinal networks. The article concludes with a discussion on the potential for further research and the development of more complex models to meet real-world needs.