Identifying treatment responders using counterfactual modeling and potential outcomes

Area of Science:

• Biostatistics
• Translational Medicine
• Clinical Trial Methodology

Background:

• Personalized medicine requires individualizing treatment based on patient characteristics.
• Statistical methods are crucial for predicting individual treatment effects.

Purpose of the Study:

• To develop a statistical model for estimating the probability of treatment response for individual patients.
• To apply this model to censored data in both randomized controlled trials (RCTs) and observational studies.

Main Methods:

• Utilized potential outcomes and principal stratification frameworks.
• Developed a multinomial model for left and right-censored data.
• Assumed monotonicity: no patients respond to control but not experimental treatment.

Main Results:

• Simulation studies demonstrated good calibration of predicted responder probabilities.
• Observed some variability and small bias when estimating numerous parameters.
• Successfully applied the method to a cohort study for radiotherapy selection in soft-tissue sarcoma.

Conclusions:

• The proposed statistical method effectively estimates individual treatment response probabilities.
• The method is applicable to various study designs with censored data.
• This approach aids in optimizing treatment selection for personalized medicine.