Survival causal rule ensemble method considering the main effect for estimating heterogeneous treatment effects

Area of Science:

• Medical research
• Machine learning
• Biostatistics

Background:

• Precision medicine necessitates understanding treatment effect heterogeneity based on patient characteristics.
• Existing machine learning methods for heterogeneous treatment effects often lack interpretability (black-box models).
• Current methods primarily focus on continuous or binary outcomes, neglecting important survival data.

Purpose of the Study:

• To develop an interpretable machine learning method for estimating heterogeneous treatment effects specifically for survival data.
• To address the limitations of black-box models in understanding patient characteristics influencing treatment outcomes.
• To provide a tool for analyzing survival outcomes in the context of personalized medicine.

Main Methods:

• Proposed a novel heterogeneous treatment effect estimation method for survival data.
• Utilized RuleFit, an interpretable machine learning algorithm, as the foundation for the model.
• Validated the method through numerical simulations and application to a real-world HIV clinical trial dataset (AIDS Clinical Trials Group Protocol 175).

Main Results:

• Numerical simulations demonstrated that the proposed method achieves prediction performance comparable to existing state-of-the-art methods.
• Application to the AIDS Clinical Trials Group Protocol 175 dataset illustrated the method's interpretability using real patient data.
• The survival causal rule ensemble method provides sufficient estimation accuracy and a clear interpretation of treatment effect variations.

Conclusions:

• The developed survival causal rule ensemble method effectively estimates heterogeneous treatment effects for survival data.
• The method enhances interpretability, crucial for understanding treatment-effect relationships in precision medicine.
• This approach offers a valuable, interpretable alternative for analyzing complex survival outcomes in clinical research.