Model-Based Clustering of Nested Lifetime Data: Profiling COVID-19 Heart Failure Patients

This study presents a novel multilevel cluster weighted model\r\nfor lifetime data with random covariates and frailties. In the framework\r\nof model-based clustering, we extend cluster-weighted model to handle\r\ntime-to-event responses and data with hierarchical structures. The proposed\r\nmethodology allows to identify latent clusters of observations characterised\r\nby random covariates with different distributions, whose time-to-event dynamic and heterogeneity at the grouping level also differ. The objective function that defines the model is maximized using a stochastic\r\nEM algorithm tailored to right-censored lifetime data. The development\r\nof this method is motivated by a study on the survival of COVID-19 heart\r\nfailure patients, admitted to multiple hospitals in Lombardy region. By\r\napplying the proposed method, we identify latent clusters of patients\r\nthat differ in terms of clinical characteristics and, for each cluster, we\r\ninvestigate the survival pattern, the association of particular respiratory\r\ndiseases with their death hazard and the supplementary effect of the\r\nfacility.