In this article, some prediction strategies are introduced for event times, where multivariate data with competing or semi-competing risks are simultaneously collected. Without loss of generality, the proposed methods can be used to analyze multivariate longitudinal data with competing or semi competing risks, often encountered in social sciences and sports activities. Regarding the situations mentioned earlier, we can provide the prediction values of: I. Time of occurrences of any cause for specific individuals II. Time of subsequent events for some cause in other individuals III. The covariate values on predicted time of I and II. Accordingly, doctor assistants or nurses can schedule good visiting times based on I and II. Item III can provide the missing values of all covariates that are utilized for better modeling. The corresponding statistical background is extensively discussed. Finally, an actual data set has been analyzed, the prediction values are provided, and their performances are assessed.

Keywords