Competing risk between in-hospital mortality and recovery: An application of DeepHit on COVID-19 clinical data

Abstract BackgroundUnexplained pneumonia appeared in Wuhan was soon determined to be a novel coronavirus disease, referred to as COVID-19. On March 11, 2020, WHO characterized COVID-19 as a pandemic. A plethora of studies on this pandemic is being carried out using various statistical and mathematical models. Though most of them are focusing on building predictive models, concentrating on the length of hospital stay can improve decision making and treatment plans. While modeling the length of stay, possible outcomes observed are either discharge or mortality.ObjectiveThe study aimed to analyse the survival data of COVID-19 patients with the competing risk methodology.MethodologyThe performance DeepHit, a deep learning-based competing risk model is compared with Fine-Gray, a traditional statistical model using a time-dependent concordance index. ResultsThe deep learning-based competing risk model outperformed the statistical model in terms of discriminative power.ConclusionModeling the duration of recovery and death provides valuable information for health officials to design proper strategies during the outbreak. These outcomes should be considered as competing events to model the data adequately.