Zhengyin Li ¹ Zhen Xu ^1* Qingyang Ma ^2*

• ¹ Institute of Agricultural Economics and Development, Chinese Academy of Agricultural Sciences, Beijing, China
• ² University of Chinese Academy of Social Sciences, School of Law, Beijing, China

Public health policy evaluation serves as a crucial foundation for enhancing health outcomes, optimizing healthcare resource allocation, and promoting fairness and transparency in decision-making. This article summarizes the principles, scope, and specific model configurations of traditional public health policy evaluation methods, analyzes the challenges these methods face in the era of big data. It also explores the application of machine learning in evaluating the effects of public health policies, outlining the specific steps involved and offering practical examples.Additionally, it identifies several key challenges associated with the application of machine learning in public health policy evaluation, including issues with model interpretability, data bias, the perpetuation of historical health inequities, and concerns related to data privacy and ethics. To address these challenges, the article proposes several critical development directions: integrating data-driven and theory-driven approaches to improve model interpretability; developing a multi-level data strategy to reduce data bias and address the continuation of historical health inequalities; utilizing a comprehensive combination of technical safeguards, legal frameworks, and social oversight to ensure data privacy and ethical handling in public health data; and employing validation and benchmarking strategies to ensure robustness and reproducibility.