Semiparametric Partial Functional Regression Model for Estimating Optimal Individualized Treatment Regime.

Area of Science:

• Biostatistics
• Medical Informatics
• Personalized Medicine

Background:

• Personalized medicine aims to tailor treatments using patient data, including complex functional forms.
• Accurate estimation of individualized treatment regimes is crucial for effective patient care.

Purpose of the Study:

• To propose a novel semiparametric partial functional regression model for estimating optimal individualized treatment regimes.
• To incorporate both scalar and functional patient covariates for more precise treatment recommendations.
• To ensure model interpretability and reduce the risk of misspecification.

Main Methods:

• Developed a semiparametric partial functional regression model with nonparametric main effects and flexible treatment interactions.
• Employed B-spline estimation for model parameters.
• Established theoretical convergence rates for the proposed estimation method.

Main Results:

• The proposed model effectively estimates optimal individualized treatment regimes using diverse patient data.
• The single index interaction with a monotone link function ensures interpretability.
• Simulation studies and real data analysis demonstrate the method's strong performance.

Conclusions:

• The novel semiparametric partial functional regression model offers a robust approach for personalized medicine.
• This method enhances the estimation of optimal individualized treatment regimes by accommodating complex patient data.
• The model provides a valuable tool for improving treatment decisions in clinical practice.