Semiparametric efficient estimation for additive hazards regression with case II interval-censored survival data

Area of Science:

• Biostatistics
• Survival Analysis
• Medical Statistics

Background:

• Interval-censored data are common in medical, biological, and demographic studies.
• Cox proportional hazards regression is standard for censored data, but additive hazards regression is less explored.
• Case II interval-censored data include right-, left-, and interval-censored observations.

Purpose of the Study:

• To propose a sieve maximum likelihood method for parameter estimation in additive hazards regression.
• To address the analysis of case II interval-censored data, encompassing various censoring types.
• To provide a robust statistical framework for a promising alternative to Cox regression.

Main Methods:

• Developed a sieve maximum likelihood estimation approach.
• Applied the method to additive hazards regression models.
• Utilized case II interval-censored data, including right-, left-, and interval-censored observations.

Main Results:

• Established the consistency and asymptotic normality of the proposed estimator.
• Demonstrated that the estimator achieves the semiparametric efficiency bound.
• Validated the method's performance through extensive simulation studies and a real-world clinical example.

Conclusions:

• The proposed sieve maximum likelihood method is effective for additive hazards regression with interval-censored data.
• The method offers statistically sound parameter estimation and achieves optimal efficiency.
• The approach is applicable to real clinical scenarios, as shown in a hemophilia patient study.