Mixture proportional hazards cure model with latent variables.

Area of Science:

• Biostatistics
• Epidemiology
• Medical Statistics

Background:

• Chronic kidney disease (CKD) is a significant complication of type 2 diabetes (T2D).
• Understanding risk factors for CKD progression and cure rates in T2D patients is crucial for effective management.
• Existing models may not fully capture the complex interplay of observed and unobserved risk factors.

Purpose of the Study:

• To propose a mixture proportional hazards cure model incorporating latent variables.
• To assess the impact of observed and latent risk factors on CKD hazard and cure rates in T2D patients.
• To provide a flexible statistical framework for analyzing survival data with cure fractions and unobserved heterogeneity.

Main Methods:

• Development of a mixture proportional hazards cure model with latent variables.
• Utilizing factor analysis to measure latent variables via correlated multiple indicators.
• Employing maximum likelihood estimation with a Gaussian quadratic approximation for integration over latent variables.
• Using a piecewise constant function for the unspecified baseline hazard.
• Implementation via SAS Proc NLMIXED.

Main Results:

• The proposed model effectively estimates the effects of both observed and latent risk factors.
• Simulation studies demonstrate the satisfactory performance of the developed approach.
• The model successfully identifies risk factors associated with CKD in a cohort of T2D patients.

Conclusions:

• The mixture proportional hazards cure model with latent variables offers a robust method for analyzing complex survival data in T2D patients with CKD.
• This approach enhances understanding of disease progression and cure dynamics by accounting for unobserved factors.
• The model's practical implementation facilitates its application in clinical and epidemiological research.