In this study, logistic regression analysis, orthogonal experimental design, user experience evaluation, and A/B testing are used to explore the key design factors of children’s app icons and make suggestions for optimization. Firstly, user information is collected through a questionnaire survey, and the core elements affecting users’ adoption willingness are analyzed by logistic regression. Second, orthogonal experimental design is used to evaluate the influence of function combinations on users’ preferences and screen the optimal function configuration. Once again, the user experience method is used to collect feedback on actual use, evaluate usability, ease of use, and satisfaction, and propose improvement measures. Finally, the A/B test verifies the effectiveness of the optimization scheme. Taking children’s science apps as the research object, the results show that the methodological framework proposed in this study significantly improves user experience and provides designers with a way to accurately understand children’s preferences and needs and develop high-quality science apps.