Developing Dynamic Prediction Methods for Survival Time Lost in Chronic Kidney Disease Progression under Competing...

Area of Science:

• Biostatistics
• Nephrology
• Epidemiology

Background:

• Chronic kidney disease (CKD) patients face competing risks of end-stage renal disease (ESRD) and mortality.
• Existing CKD prediction models often use hazard-based measures, which lack clinical interpretability and fail to quantify survival time lost.
• There is a need for prognostic models that account for competing risks and provide absolute measures of survival time loss.

Purpose of the Study:

• To develop a dynamic prediction model for CKD that incorporates competing risks using restricted mean time lost (RMTL).
• To provide clinically interpretable and individualized predictions of survival time loss.
• To capture the dynamic impact of time-varying covariates in longitudinal CKD data.

Main Methods:

• Developed a dynamic restricted mean time lost (RMTL) prediction model under competing risks.
• Incorporated the landmark approach to handle time-dependent covariate information in longitudinal data.
• Validated the model using Monte Carlo simulations and patient data from the African American Study of Kidney Disease (AASK) cohort.

Main Results:

• The proposed dynamic RMTL model provides accurate and robust estimates of survival time loss.
• The model effectively captures time-varying covariate effects, offering dynamic insights into CKD progression.
• The dynamic RMTL model demonstrated superior predictive performance compared to conventional static models.

Conclusions:

• The dynamic RMTL model offers a clinically interpretable and individualized approach to risk assessment in CKD.
• This model quantifies survival time loss under competing risks, aiding personalized risk assessment and intervention planning.
• The findings support the use of dynamic RMTL models for improved chronic disease management.