Purpose/Significance To systematically analyze the text features of privacy policies for health information platforms, so as to enhance the transparency and user-friendliness of privacy policy design, and provide theoretical and practical basis for privacy policy optimization. Method/Process The privacy policy texts of 17 mainstream health information platforms are selected. Based on the technology acceptance model (TAM), an indicator system is constructed. From the dual perspectives of perceived ease of use and perceived usefulness, the latent Dirichlet allocation (LDA) model is adopted to identify the privacy policy topics. Result/Conclusion At present, privacy policies for health information platforms generally have problems such as poor understandability, heavy reading burden, and complex access paths, centering on the interests of the platforms and lacking consideration of users’ cognitive abilities. Privacy design concepts such as “default protection” should be introduced to enhance the identifiability and readability of privacy policies, to promote the transformation of privacy policies from compliance to a multi-directional orientation centered on user understanding and trust, and achieve the unification of usability, accessibility and user orientation.