SEMIPARAMETRIC REGRESSION MODEL FOR RECURRENT BACTERIAL INFECTIONS AFTER HEMATOPOIETIC STEM CELL TRANSPLANTATION

Area of Science:

• Statistics
• Infectious Disease Epidemiology
• Hematology

Background:

• Hematopoietic stem cell transplantation (HSCT) patients are highly susceptible to recurrent bacterial infections post-procedure.
• Standard statistical models for recurrent events are inadequate for analyzing infection patterns after HSCT due to distinct event timings.

Purpose of the Study:

• To develop a novel semiparametric regression model to simultaneously analyze risk factors influencing the time to the first infection and subsequent infection intervals post-HSCT.
• To address statistical challenges including dependent censoring and length bias inherent in recurrent event data.

Main Methods:

• Proposed a semiparametric estimation procedure to model covariate effects on both time-to-first-event and recurrent event gap times.
• The method accounts for within-subject correlation and length-biased sampling in censored gap times.
• Validated the model's performance using simulation studies and applied it to real-world post-HSCT infection data.

Main Results:

• The developed statistical model effectively estimates the impact of patient- and transplant-related factors on infection occurrence.
• Demonstrated the model's ability to handle complex dependencies and biases in recurrent event data.
• Successfully applied the methodology to identify key risk factors for bacterial infections in HSCT recipients.

Conclusions:

• The proposed semiparametric model provides a robust framework for analyzing recurrent infection patterns in HSCT patients.
• This approach enhances the understanding of infection dynamics and aids in identifying high-risk individuals.
• The findings have implications for optimizing infection prevention and management strategies in HSCT care.