Analysis of competing risks in the CoxPH model for progressive censorship with binomial removal

Background: In medical research and survival analysis, it is common for an individual or item’s failure to be attributable to multiple causes, also known as competing risks. This article focuses on examining the competing risks model as the data increasingly becomes type II censored and randomly removed. The model assumes that the causes of failure are independent and that the lifetimes of individuals are described by the Cox model. At each failure time, the number of items or people removed follows a binomial distribution. The article derives estimators for the indefinite parameters in the model. The study presents a set of detailed data and includes a simulation study that also illustrates the results. Methods: Different reasons, frequently known as competing risks, are frequently embroiled in an individual’s or an item’s failure in medical research survival analysis. The competing risks shown under sort II dynamic censoring with random removals are the subject of this research. We get the maximum likelihood and inexact most extreme probability estimators of the obscure parameters. The asymptotic distribution of the maximum probability estimators is utilized to decide the CIs. Then, Monte Carlo simulations were applied to demonstrate the approach. The analyses were performed utilizing R 4.0.4 software. Results: For stroke, systolic blood pressure (SBP) and hypertension status are the only significant variables. In contrast, gender, body mass index (BMI), smoking status, the logarithm of urinary albumin and creatinine ratio, and diabetes status are significant variables for coronary heart disease (CHD) and other cardiovascular diseases (CVDs). The results suggest that significant risk factors differ for different types of CVD events. Conclusion: The outcomes of the simulation study indicate that progressively right-censored type II sampling designs outperformed the usual censored type II sampling designs. Therefore, the estimated parameters on the defined pattern setting are recommended. They can be used in many practical situations when competing risks occur, and progressive censoring could be considered.