Regression modelling of interval censored data based on the adaptive ridge procedure

Area of Science:

• Biostatistics
• Dental Research
• Survival Analysis

Background:

• Ankylosis is a significant complication following tooth replantation.
• Accurate analysis of time-to-event data, especially with censored observations, is crucial for understanding ankylosis progression.
• Existing statistical models may not adequately handle the complexities of censored data in dental trauma.

Purpose of the Study:

• To propose a new statistical method for analyzing time to ankylosis in replanted teeth.
• To develop a robust approach for handling left-censored, interval-censored, and right-censored data.
• To identify specific time intervals associated with increased ankylosis risk.

Main Methods:

• A Cox model with a piecewise constant baseline hazard function was employed.
• The Expectation-Maximization (EM) algorithm was used for parameter estimation, treating true event times as unobserved.
• A penalized likelihood method was implemented to automatically determine the number and location of baseline hazard cuts.

Main Results:

• The EM algorithm produced a block diagonal Hessian matrix, facilitating penalized likelihood estimation.
• The penalized likelihood method effectively determined baseline hazard structure and identified high-risk periods for ankylosis.
• Simulation studies demonstrated good baseline hazard fit and precise regression parameter estimation.

Conclusions:

• The proposed penalized Cox model offers an effective method for analyzing time to ankylosis in replanted teeth with censored data.
• This approach enhances the identification of critical time windows for ankylosis development.
• The methodology is extendable to include exact observations and cure fractions, offering broader applicability in survival analysis.