[Weighted random forest for estimating individualized treatment rules]

With the rapid development of personalized medicine, recommending the optimal treatment regimes among multiple options for individual patients has become a key topic in the study of individualized treatment rules. Existing methods often face challenges such as limited accuracy and robustness when handling multi-category treatment problems. This study proposes a weighted random forest method that formulates the treatment decision problem as a weighted classification task. By incorporating the expected loss differences among treatment outcomes, the method enhances its learning process and improves recommendation performance with the non-parametric nature and flexibility of random forests. The weighted random forest method is further applied to real-world hypertension intervention data to generate personalized antihypertensive treatment recommendations based on the patient's baseline characteristics, demonstrating its potential value in clinical practice. This research aims to provide a new approach for individualized treatment rules in multi-treatment settings and to support the development of data-driven clinical decision-making systems.